Elections in the 21st Century: from paper ballot to e-voting · partnership with an online voting...

Elections inthe 21stCentury:from paperballot to e-voting

The Independent Commission on Alternative Voting Methods

Elections inthe 21stCentury:from paperballot to e-voting

The Independent Commission on Alternative Voting Methods

Dr Stephen Coleman (Chair), Director of theHansard Society’s e-democracy programme, andlecturer in Media and Citizenship at the LondonSchool of Economics & Political Science

Peter Facey, Director of the New PoliticsNetwork

Paul Gribble CBE, Independent consultant andadvisor on Election Law to the ConservativeParty; Editor of Schofield’s Election Law

Steven Lake, Director (Policy & External Affairs)of the Association of Electoral Administrators, andElectoral Registration Officer for SouthOxfordshire District Council

Gerald Shamash, Solicitor. Electoral Lawyer to theLabour Party

The first Chair of the Commission, KeithHathaway, resigned in June 2000 when he retiredas Chair of the Association of ElectoralAdministrators and became involved in electionmonitoring and supervision missions overseas.Sadly, we lost to ill health our esteemedcolleague, George Smith, Chair of the Associationof Electoral Administrators. A third formerCommissioner, Joe Wadsworth, of ElectoralReform Services, resigned to avoid any conflict ofinterests when that company entered intopartnership with an online voting vendor. PeterFacey joined the Commission to replace KeithHathaway and Steve Lake replaced George Smith.

The Commission has been extremely well servedby its clerk, Rebecca Williams, who has beensupported by research assistance from SimonCollingwood and David Pepper.

The Commission was established by the ElectoralReform Society, and is grateful to them forhosting our meetings.

The Electoral Reform Society is very grateful tothe members of the Independent Commission forthe time which they have so freely given and fortheir commitment to the development of goodelectoral practices in the UK.The Society, whichhas since 1884 been campaigning for thestrengthening of democracy (although principallythrough reform of the voting system), believesthat the Commission’s report is a majorcontribution to the debate on the modernisationof the way we vote and that the report shouldguide the development of policy and practice inthis area in the coming years.

Although the Independent Commission wasestablished by the Electoral Reform Society andhas been serviced and supported by it, theSociety has not sought to influence theCommission’s work.The views expressed in theCommission’s report therefore do notnecessarily represent the views of the Societyand, similarly, neither the Commission as a bodynor its members individually necessarily supportall of the policies of the Society.

Elections in the 21st Century: from paper ballot to e-voting

2

Membership of the Independent Commission on Alternative Voting Methods

Membership of the Independent Commission on Alternative Voting Methods

Terms of Reference

Preface

Executive Summary

Overall Recommendations

Introduction

Postal Voting

Electronic Counting

Electronic Machine Voting

Telephone Voting

Online Voting

Submissions and Other Information

Bibliography

02

04

05

09

13

15

19

43

63

73

83

99

101


3

Contents

The Commission shall look into new methodsof voting and related matters and draw uprecommendations that will ensure that boththe security and secrecy of the ballot aremaintained and to ensure continued publicconfidence in the electoral process.

1. By electronic voting we mean all forms ofelectronic or mechanical equipment forassisting in the counting of votes, e.g.

a) the mechanical and/or electronic counting,or sorting and counting, of ballot papers;

b) computer programs to assist returningofficers with calculations and in theproduction of results sheets;

c) equipment by which voters enter their voteson a console rather than using a ballotpaper, and computer disks, or the equivalent,take the place of ballot boxes;

d) systems in which registers are online and whichallow voters to vote at any polling station;

e) telephone voting;

f) internet voting;

g) universal postal voting.

2. The Commission should consider each formof electronic voting, looking at:

a) the state of the art technology;

b) its existing use, in the UK or internationally;

c) the ease of voting and risks of votes beingwrongly cast;

d) the safeguards required against voterimpersonation;

e) transparency and the facility with whichcandidates and their agents can ensure thatthere has been no malpractice;

f) whether there is any place/need forrecounts with the different systems;

g) the ease and speed of counting;

h) risks and consequences of mechanical failure,power cuts, etc.;

i) danger of tampering, e.g. by technical staff or hackers;

j) possible biases which might be introduced bydifferential access to the voting system, e.g. bythose with telephones or internet access;

k) any implications for turnouts;

l) advantages which might be offered for morecomplex forms of vote counting (e.g. STV);

m)approximate costs of implementation (andsavings over manual counting).

n) data protection problems.

3 The Commission should produce:

a) guidelines for the safeguards required topreserve the integrity of the election witheach form of electronic voting;

b) recommendations on how far and how fastthe Government should move in introducingelectronic voting.


4

Terms of reference

Democracy and Participation

A democracy in which the public does notparticipate is in trouble. Falling turnout atelections is a worry for all of us, because weknow that voting is the most basic act ofdemocratic participation; people who do notvote tend not to participate in other civicactivities. It is not the job of the Commission totell people that they should vote.We respectthose who choose to abstain. But we areconcerned to promote public participation indemocratic life. It is beyond the remit of thisreport or this Commission to consider mattersother than alternative methods of casting andcounting votes, but we are strongly in favour ofattempts to reinvigorate public participation,whether through the teaching of democraticcitizenship in schools, the work of the ElectoralCommission or the Government’s developinginterest in e-democracy.

Our work began before the 2001 election, in theshadow of the 71.3% voter turnout in 1997 –the worst since 1935. In 2001 voter turnout fellby a staggering 12 percent, leaving us in no doubtthat the democratic process in the UK is indeedin trouble.Three out of four 18-24 year-olds didnot cast a vote in 2001. Historically, voting isassociated with higher levels of affluence andeducation; so, it is particularly disturbing toobserve that, as the population as a whole hasbecome progressively more affluent andeducated in recent years, voting levels havedeclined. It would be simplistic and naive toimagine that new methods of voting couldredress this drift, unless they were part of amuch broader revitalisation of democratic life. So,any recommendations made here must beconsidered in the context of an agenda formaking democracy more accessible and

meaningful to citizens.This report is thereforeoffered as a contribution to a wider commitmentto nourish and energise democratic life in Britain.

Do Methods Matter?

Superficially, it would not seem to be the casethat the public cares very much about how theycast their votes. If voting in elections wereabolished or limited tomorrow there would be ahuge public outcry, but we doubt very muchwhether many people are much bothered byhow votes are cast or counted. On the otherhand, the public, in a number of recent opinionpolls and surveys, have stated that current votingmethods are inconvenient. People are becomingused to conducting their transactions in flexibleways – by post, phone, in the street orsupermarket, and increasingly at home via emailand the web – and the burden of walking to apolling station to cast a vote could seemanachronistic.We cannot be sure that all thosewho cite inconvenience as their reason for non-voting are telling the whole truth; maybe it iseasier to blame voting procedures than to admitto inertia or apathy. Critics of the convenienceargument say that there is a civic obligation tovote and that those who cannot even bebothered to walk to a nearby polling station aresimply irresponsible. Such a view fails to recognisethe logistical complications of modern lifestyles.

Whatever the arguments for and against making iteasier for people to vote, we are convinced thatculture is more important than convenience andthat politics is a greater motive for voting thanprocedures. In short, people vote when they feelthat there is something worth voting about andthat their votes count. It is up to the parties andcandidates, in their campaigning, to provide suchincentives; where they do, people will vote, even if


5

Preface

it is not very convenient; where they fail to dothis, people will not vote, even if the easiestvoting technologies are available to them.Voting isa political act, not merely a procedural one.Thatdoes not mean that voting methods need not beconsidered; as long as a single person who ismotivated to participate in an election isprevented from doing so because the method onoffer is inconvenient there is a need to addressthe questions raised in this report.

Assessing Risk

We live in a world of risks. Just as personal health,air and road transport or commercial transactionsare prone to risks and newly-perceived dangers, sois democracy. But, unlike most other risks that faceus, democracy is a truly collective good: we all needit to work well.The public can only be expected tohave confidence in the electoral system if it is basedupon transparently fair and robust procedures andworking methods. One could argue that publicconfidence in our current voting methods is notwell founded; certainly, the current system is moreopen to fraud than many believe.Voters in the2000 US presidential election discovered muchabout their electoral procedures which underminedtheir previous confidence. One thing is certain:public confidence in democratic elections takesdecades to develop and far less time to destroy.

Dealing with risk is not a matter of eliminating alluncertainties, but of setting clear limits upon thescope for accidents, attacks and errors. In thinkingabout voting methods, we need to decide howmuch risk is acceptable. For example, let usimagine that there is a method of voting that islikely to result in higher voter turnout thancurrent methods, but at greater risk to theprobity, accuracy or security of the electoralprocess. Should it be accepted or rejected? Theanswer does not lie in an absolutist rejection ofrisk, but a clear policy about where on thespectrum of risks one decides to draw a line.TheCommission has spent much time over the pasteighteen months trying to draw such lines.

The Commission: Purpose and Principles

The Independent Commission on AlternativeVoting Methods was established to examine,

analyse and offer recommendations about a rangeof possible methods for casting and countingvotes in UK elections.We have been motivatedthroughout by a keen awareness of the need fordemocratic practices to evolve and improve, butalso a commitment to resisting changes whichwould fail to win public confidence or meet thehighest democratic standards.The probity,accuracy and security of electoral arrangementsare integral to the vitality and credibility ofdemocracy. Everything in the following pages isintended to reflect that principle.

As its name suggests, the Commission is a totallyindependent body, with no ties to externalinstitutions or organisations. In the course of ourdeliberations, we have had discussions withGovernment, and have been invited to submit ourfindings to them; however rather than seeking toset out a policy for Government, we haveproduced a set of principles andrecommendations upon which we would wish tosee such policy based.We have been pleased totake evidence from vendors of electionmachinery and software; however we havetreated this critically and have no ties to suchcorporate interests. In all of our investigationsover the past year and a half, and in the writing ofthis report, the Commission has acted in anindependent and principled way and should bejudged in these terms.

I have been privileged to chair this Commissionand learn an enormous amount from my fellowCommissioners, as well as those who gaveevidence to us.

In examining alternative voting methods, we haveconstantly kept in mind that any increase, orperceived opportunity for increase, in electoralfraud or malpractice would damage the integrityof the electoral process, the trust of the publicand the legitimacy of our elected representatives.We set out ten key criteria which we applied toall methods of voting that we considered:

Security – protection against voter impersonationor tampering

Secrecy – protection against undue influence ortraceability of votes other than by court order

Elections in the 21st Century: from paper ballot to e-voting Preface

6

Ease of voting and risk of mistakes – votingshould be convenient, but also foolproof

Transparency – opportunities for scrutinythroughout the process

Speed and efficiency of counting – having beencast, the counting of votes should not be delayedor made less trustworthy

Accuracy of results – voters deserve an accurate count

Effects on turnout – the maximum number ofeligible voters should be encouraged to cast theirvotes, but not at the expense of differential access

Cost – the process must be affordable andadequately funded

The Commission welcomes the fact that theGovernment has set in motion the process ofexamining and piloting alternative voting methods.It is important that these are independentlyevaluated and we would suggest that the criteriaoutlined in this report would be appropriate forsuch studies. Using evidence from places wherealternative methods have already been used, thisreport makes recommendations about the speedand degree to which the Commission believesthe Government should be moving forward withvarious new forms of voting.Theserecommendations are accompanied by guidelinesfor the safeguards that the Commission believesneed to be in place to preserve the democraticintegrity of the electoral process.

Dr Stephen Coleman

Elections in the 21st Century: from paper ballot to e-voting Preface

7


8

The Independent Commission on Alternative VotingMethods welcomes the current climate ofinvestigation into renewing the electoral andbroader democratic processes of the UnitedKingdom. One part of this agenda is the pilotingand possible further extension of new votingmethods, and as part of an attempt to addressdisengagement from the political process we viewthese as positive developments. However, anychange can bring with it suspicion and uncertainty,and it is vital that changes to our electoral processdo not render that process more vulnerable, anddo not undermine voters’ confidence in the system.

It is not only the mechanics of the systemsthemselves, but also the way in which they areimplemented which affects the confidence ofvoters, and the efficiency and effectiveness of thewhole process. Recent elections in the UnitedKingdom have been characterised by theintroduction of last minute legislation, insufficientfunding and insufficient time for preparation andthe testing and validation of equipment. Nothingwill cause a greater loss of confidence than forsystems to be introduced without this work beingproperly carried out.

Elections in the 21st Century: from ballot paperto e-voting examines five alternative votingmethods.These are listed below along with asummary of our conclusions and recommendations.The full conclusions and recommendations on eachmethod are at the end of each chapter.

(Universal) Postal Voting

The increased use of postal voting, whether ondemand or universal, offers increasedconvenience to the voter and therefore has thepotential to increase turnout at a manageablecost. Access is equal to all.

Problems arise however, with regard to theidentity of the individual casting the vote andthe secrecy in which the vote is cast. Postalvoting does not make voting any more user-friendly or vote-counting any more efficient oraccurate.

The Commission recommends:

p that further pilots take place in local elections,European Parliamentary elections andParliamentary by-elections, across a wholeauthority, electoral region or constituencyrespectively, and also in parish and communitycouncil elections

p that voters be required to provide their datesof birth (both at registration and in completingthe postal vote) in order to avoid personation;otherwise, that Declarations of Identity beretained

p the introduction of a marked register for postalvotes, so as to maintain a record of postalvoters whose votes were returned and allowchecks to be made on whether a vote hasbeen cast in the name of, for example, arecently deceased elector

p that postal voters who have not received theirpapers and do not have time to be issued withreplacements, be allowed to attend a pollingstation and cast a tendered ballot, so as not tobe totally disenfranchised

p that postal voters be made fully aware of thetimetable for issuing postal ballots

p that consideration be given to thearrangements for the delivery and return ofpostal votes


9

Executive Summary

Electronic Counting

The main area in which electronic counting canbe beneficial is in increasing the speed andaccuracy of the count. However, it does so atconsiderable cost.

Since this kind of technology affects the countingof votes, rather than the voting itself, it will nothave any effect on turnout or increase the user-friendliness of the voting process. Opportunitiesfor scrutiny are somewhat reduced, though apaper audit trail is retained.


p that if such technology is introduced, it shouldbe at polling station level, rather than at acentral count; this would allow theinadvertent spoiling of ballot papers to beavoided

p that recounts must be available and that thenature of the recount (electronic or manual)must be governed by fixed, publicised rules

p that rigorous and realistic testing of softwareand hardware be carried out in situ by theReturning Officer, with candidates and agentsgiven the opportunity to be present


As a polling-station based technology, electronicmachine voting would not offer any benefits interms of voter turnout, but equally would notraise any problems of differential access.

Electronic machine voting could make the votingexperience more user-friendly by warning peoplebefore they cast a spoilt ballot, and by providingvoting formats that allow voters with visual andhearing impairments to cast their votesunassisted. Increased speed and accuracy ofcounting would also be a benefit.

The introduction of this kind of technologywould be extremely costly. Opportunities forscrutiny would be substantially reduced, andthere would not be a paper audit trail in thetraditional sense.


p that further pilots take place at any level ofpublic election (including Parliamentary by-elections), except at a General Election

p that the equipment used be specificallyelection-dedicated voting equipment

p that rigorous and realistic testing of softwareand hardware be carried out in situ by theReturning Officer, with candidates and agentsgiven the opportunity to be present; testmodes should not be allowed and the testvotes should be entered by hand, not as pre-prepared data

p that voting data be recorded and stored induplicate in case of damage to data

p that voting machines be programmed to allowvoters to cast a blank ballot, but that voters bewarned before doing so

p that voting machines log all events, by votersand administrators

p that the security of the machines andcartridges be as high as ballot paper security

p that percentage turnout by polling station bemade available on request to candidates andtheir agents during election day

p that a detailed breakdown of voting by pollingstation be made available to candidates andtheir agents as soon as possible after thedeclaration of the result

p that the election data be made available to acourt, in the case of a challenge to the result

Telephone voting

Telephone voting offers increased convenience tothe voter and therefore has the potential toincrease turnout at a manageable cost.Voterscould be prevented from inadvertently spoilingtheir vote, and vote-counting could be muchquicker and more accurate than at present.

Elections in the 21st Century: from paper ballot to e-voting Executive Summary

10

Problems arise however, with regard to theidentity of the individual casting the vote and thesecrecy in which the vote is cast.

Opportunities for scrutiny would be substantiallyreduced, and there would not be a paper audittrail in the traditional sense.


p that further pilots take place, but that theseinitially avoid combined elections and electionswith more than three vacancies

p the introduction of Elector Cards inconjunction with PINs to avoid personation;otherwise to require voters to use their dateof birth along with a PIN

p that the voter’s identity be stripped from thevote and stored separately, and that noindividual or individual agency should havethe capacity to match the two sets ofrecords; the two sets of records should onlybe matched if a court order requires suchaction to be taken

p that all telephone calls from within the UnitedKingdom be free of charge

p that PINs be randomly generated

p that buying or selling PINs be made an offenceand that the penalties be widely publicised

p that the interactive voice system be as user-friendly as possible and offer the possibility ofhaving the instructions and options repeated atany stage; voters must be given theopportunity to review their choices beforeconfirming their vote

p that there be sufficient telephone lines that atno point do electors have difficulty gettingthrough to the voting system

p that the system should log all aspects of the call

p that voters be allowed to cast a blank ballot,but that they be warned before doing so

p that voters who have not received theirsecurity information and do not have time tobe issued with replacements, be allowed toattend a polling station and cast a tenderedballot, so as not to be totally disenfranchised

p that a detailed breakdown of voting by thesmallest appropriate polling division (equivalentto a polling station) be made available tocandidates and their agents as soon as possibleafter the declaration of the result


Online voting

Online voting offers increased convenience to thevoter and therefore has the potential to increaseturnout.Voters could be prevented frominadvertently spoiling their vote, and vote-counting could be much quicker and moreaccurate than at present.

Problems arise however, with regard to the identityof the individual casting the vote and the secrecy inwhich the vote is cast. Differential access to onlinetechnology would be a serious issue.

Opportunities for scrutiny would be substantiallyreduced, and opportunities of external attackwould be significantly increased, particularly inview of the vulnerability of personal computers.There would not be a paper audit trail in thetraditional sense.


p that a Technology Taskforce be established priorto any pilots in order to evaluate and challengethe system

p that electors who already have a digital certificateand the necessary software be allowed to usethem as a form of voter authentication

p that electors have the option of being issuedwith an Elector Card by their local authority tobe used in conjunction with a PIN; otherwisethat voters be required to use their date ofbirth in conjunction with a PIN


11

p that the voter’s identity be stripped from thevote and stored separately, and that noindividual or individual agency should have thecapacity to match the two sets of records; thetwo sets of records should only be matched ifa court order requires such action to be taken

p that PINs be randomly generated

p that buying or selling PINs be made an offenceand that the penalties be widely publicised

p that there be sufficient servers with adequatecapacity that at no point do electors havedifficulty getting through to the voting system

p that voters be allowed to cast a blank ballot,but that they be warned before doing so

p that voters who have not received theirsecurity information and do not have time tobe issued with replacements, be allowed toattend a polling station and cast a tenderedballot, so as not to be totally disenfranchised

p that a detailed breakdown of voting by thesmallest appropriate polling division (equivalentto a polling station) be made available tocandidates and their agents as soon as possibleafter the declaration of the result



12

The Independent Commission on AlternativeVoting Methods makes the followingrecommendations in addition to those referringto specific voting methods contained in theindividual chapters that follow.Theserecommendations apply to the introduction ofany new voting methods.

1. We recommend pilots in the use of ElectorCards in conjunction with postal, telephone oronline voting.

2. We recommend pilots in electoral registrationwhich provide for more secure methods ofvoting. Subject to approval from the DataProtection Commissioner, these would gatherelectors’ dates of birth which could be used toenhance the security of all forms of remotevoting.

3. Any pilots using new voting methods must beformally assessed by an independent bodyrather than by those involved in the conduct ofthe election. Specific criteria must be laid downfor the assessment of such pilots.

4. In order to build confidence amongst voters, aninformation campaign must be a central part ofthe introduction of new voting methods.

5. With the exception of all-postal elections, werecommend that a multiplicity of votingmethods be maintained for the foreseeablefuture.

6. The Electoral Commission should haveresponsibility for :

a) the validation to the highest internationalstandards of any hardware and softwareused in the electoral process

b) putting standards in place for the testing andcertification of software and hardware at allstages of the electoral process; returningofficers should be party to testing at certainstages, and candidates and their agentsshould also have the option to be present

c) oversight of the use and validation of anybarcodes used in the electoral process

d) the standardisation of contracts withcompanies providing equipment

7. Sufficient time must be allowed for anynecessary legislative processes to fully taketheir course before implementation of the newvoting methods begins.

8. It is not possible to run effective electionswithout adequate resources. Since costs willinevitably be incurred in the introduction ofnew voting methods, we urge the Governmentto dedicate sufficient funds to implementingany changes.

9. Sufficient time must be available for detailedplanning, to include areas such as training and adetailed rehearsal for all staff; a back-up plan incase of power-failure; briefings for all keyparticipants (e.g. candidates and their agents,media etc).


13

OverallRecommendations


14

Our electoral arrangements are being challenged.The procedures that govern the way in which wecast our votes have remained largely unchanged andlargely unquestioned in this country for over acentury. Now they are having to face up toincreasing competition both from new methods thathave emerged as a result of technological advances,such as electronic and online voting, but also fromthe possibility of extending existing methods whichuntil now have only been used by a small minority ofvoters, such as postal voting. Consequently, ourtraditional polling station, paper and pencil method ofvoting is coming under unprecedented scrutiny.Thecombination of rapid technological development,concerns about public engagement in the electoralprocess and the fact that ‘alternative’ methods arenow routine in many other countries, has served toopen up the debate in the United Kingdom.

Though technological advances and the desire forprogress exert pressures of their own, the mostseductive reason for change in the current climateis probably the possibility of increasing turnout atelections, even more so in view of the 59.4%turnout in the General Election of June 2001.Other factors also play their part in the debate,such as the potential for reducing the number ofinvalid votes, improving the speed and accuracy ofelection counts, reducing the cost and generallystreamlining the process.These factors, which mayonce have seemed peripheral, have taken on fargreater importance in view of the U.S.Presidential election in November 2000 where aclose result shone a bright light on these areas ofelectoral procedure, and in many cases revealedthat they did not bear up to such scrutiny.

Current Electoral Arrangements

The electoral law which governs electionprocedures in the United Kingdom today

originated in the nineteenth century, a periodwhen corruption was rife and the votes of thesmall electorates could be freely bought and sold.The introduction of the Ballot Act in 1872provided for the first time a secret ballot, greatlyundermining the ability of candidates and theiragents to buy votes. Several other Acts followedwhich affected the legal and administrativeprovisions for elections both at national and locallevel, including the 1883 Corrupt and IllegalPractices Act which laid down basic rules for theprocedures to be followed at elections.

The procedures adopted at the end of thenineteenth century have survived, and they havebeen recognised for providing free andtransparently fair elections. For General Elections,each parliamentary constituency has a returningofficer who is responsible for ensuring that all theprocedures and election rules are properlyapplied. For Local Elections, there is a returningofficer for each local authority.

Registration: In order to vote, individuals musthave their name on the Electoral Register. Eachhousehold gets a regular opportunity to fill in aform and list all those residing in that propertywho are of voting age or who are nearing thatage.These forms when returned to the electoralregistration officers within local authorities providethe basic information used to compile the localarea register of electors.With the recentintroduction of ‘rolling registration’, people are ableto change their details, for example if they movehouse, on a monthly basis at any point during theyear. In the period leading up to the election,registered voters are sent a polling card givingdetails of the date and nature of the election andthe location of the appropriate polling station.Provisions exist under the present system for anelector either to vote in person at a polling


15

Introduction

station, or (by prior arrangement) by post or byappointing a proxy to vote on their behalf.

Voting at a polling station: On arrival at theappropriate polling station, the voter states his/hername and address.The voter is not required toshow any form of identification.When the entryon the Electoral Register has been found, thevoter’s name is crossed off.The voter’s electoralnumber (from the Register) is then marked onthe counterfoil of the ballot paper which is aboutto be issued.The ballot paper and the counterfoilare both already marked with an identical printednumber.The ballot paper is marked with theofficial stamp.The voter is given the ballot paperwhich s/he takes to the polling booth in order tocast the vote in secret, by placing a mark, asrequired, on the ballot paper. Ballot papers, oncemarked, are deposited in ballot boxes which arein full public view and sealed against tamperingprior to the commencement of the poll.

The Count: At the close of poll the ballot boxesare transported to the count location, which forlocal elections may be in the polling station.Thenthe boxes are opened and the votes counted byhand in full view of any candidates and agentswho choose to be present. Rules are applied toensure that the Returning Officer, in consultationwith the candidates and their election agents, candeal with any ballot papers which arequestionable, challenged or void. If the result isclose, a recount may be held, at the discretion ofthe Returning Officer.

Challenges: If there is a formal challenge to theprocess through the courts, then it is possible totrace individual ballots (by finding the ballot paperwhich matches the counterfoil number with theparticular voter’s electoral number written on it),and verify whether or not fraud has taken place,and whether it would have made any differenceto the outcome of the election.

Towards Reform

Reasons for CautionTransparency, security and secrecy. The currentelectoral arrangements have many strengths, notleast their transparency. Candidates and theiragents are able to observe each stage of the

process.The fact that votes are required by lawto be traceable means that there is a deterrent tofraud and a way of identifying it where it hasoccurred1. In this way, both security and secrecyare relatively easy to maintain.

Equal access. The current system provides equalaccess to everybody on the electoral register,particularly now that polling stations are requiredto be more accessible to voters with physicaldisabilities and to provide facilities that will allowpartially sighted voters to cast their vote unaided.

Recounts. Recounts, though time-consuming, arenot problematic.

Reasons for ChangeTurn-off and turnout. Despite the strengths ofthe current electoral arrangements, recent yearshave witnessed increasing civic disengagementfrom the electoral process.Turnout in UKelections at every level is falling.The past decadehas witnessed some of the lowest voter turnoutsin local elections since records began.The DETR’sGreen Paper, Modernising Local Government: LocalDemocracy and Community Leadership (1998)points out that ‘turnout in local governmentelections in Great Britain is at the bottom of theEuropean Union league table below almost allindustrialised nations.’ In the 2001 GeneralElection turnout fell to 59.4%, the lowest levelsince 1918 (a year in which a significantpercentage of the electorate was overseas onmilitary service, and prior to any arrangementsfor absent voting). Only 23% of eligible votersturned out for the 1998 European election, alower percentage of the UK population thanthose who voted in the phone ballot for the finalepisode of Channel 4’s Big Brother game show. AsRallings and Thrasher have observed,

“Since 1992 the slump in participation inelections of all kinds has been marked.Theturnout at the local elections, which followedjust four weeks after that year’s generalelection was the lowest since 1945.Turnoutat parliamentary by-elections over the 1992-97 parliament was on average some 24percentage points below general electionlevels. Labour’s … successful defence of LeedsCentral at a 1999 by-election saw the lowest

Elections in the 21st Century: from paper ballot to e-voting Introduction

1. Many other countries have amore preventative approach tofraud and require that votersshow proof of identity beforecasting a vote.They specify thatno link should be maintainedbetween a voter and his/her vote.For a more detailed discussion ofthis issue see Ballot Secrecy:Electoral Reform Society/LibertyWorking Party Report, ElectoralReform Society, 1997

16

peacetime turnout in an individual seat(19.6%). Coupled with plummeting figures forthe 1999 European Parliament elections(24%) and the first recorded instance in 1998of fewer than one in three electors acrossEngland voting at local elections (28.8%),concern has been expressed about a crisis ofdemocracy in Britain.”

There appears to be a prevalent mood ofdisenchantment with the political process,combined with widespread ignorance about howit works or whether it matters.The Governmentset up an Advisory Group on Education forCitizenship and the Teaching of Democracy, underthe chairmanship of Bernard Crick, which hasreflected upon the need for encouraging politicalliteracy within the school curriculum (1998). But itwill take more than lessons in school to revitalisethe democratic process. Redressing the drifttowards civic disengagement must involveconsideration of a number of issues which arebeyond the scope of this Commission’s work.These include the failure of the parties tocampaign on issues in ways that convey realpolicy differences to electors; the move awayfrom organised party politics towards single-issuecampaigning; and the rise of e-politics and newforms of political communication. In short, we arenot suggesting that by improving voting methods,as one factor within the electoral landscape,overall turnout will necessarily increase.

There are a number of variables that affectturnout at elections and these have a differentimpact in different constituencies and at differenttypes of elections.These include: the marginalityof the contest, the size and composition of theelectorate, the number of parties contesting theelection, partisanship of ward, the strength andnature of the party campaigns2, the identificationof an issue, mobility of the population, the mediaand strength of local identity. All these and morehave an effect upon the number of people whoturn out to vote at an election.

If it can be shown that a new method of voting –whether by post, online or via digital TV – wouldsubstantially increase the number of citizens choosingto cast a vote, that in itself would be a strongargument in favour of providing such an option.

A voter attitude survey conducted by MORI onbehalf of the Electoral Commission3 found that21% of non-voters in the 2001 General Electionsaid that they “couldn’t get to the polling stationbecause it was too inconvenient” with a further16% saying that they “were away on electionday”, suggesting that voting arrangements couldhave a significant impact on voter turnout.

Invalid ballot papers. Though casting a vote iscurrently not an especially complicated matter,invalid ballot papers are far from uncommon(0.26% in the 2001 General Election4 and 0.3%in 19975).This means that the voter’s intentionwas unclear, generally because they did notmake a mark, they put a mark in the wrongplace or in too many places, or the markrevealed the voter’s identity. Methods ofreducing the number of unintentionally spoiltballot papers would be welcomed.

Speed of the count. Under the currentarrangements, a number of factors contribute tothe speed of the count.The level of voter turnouthas an effect on the number of individual votes tobe counted, and the number of people employedto do the counting is clearly also an importantfactor. In General Elections, the quickest countstake just under two hours and, barring recounts,most voters can expect to know the result intheir constituency by the time they wake up thefollowing morning. A hastening of theannouncement of results could be advantageous.

Accuracy of the results. The accuracy of theresults is open to a certain margin of human error,but the process is scrutinised by candidates andagents. It would be unusual for the margin ofhuman error to be greater than the differencebetween the two leading candidates. If thedifference is especially close, then a recount can becalled.This was the case in sixteen constituencies inthe 1997 General Election. Clearly, any way ofreducing inaccuracy would be a positive step.

Other factorsCost. It is estimated that the 1997 General Electioncost £52 million of public money6, a figure whichcovers Returning Officers’ Expenses, but does notinclude electoral registration.This represents £1.19per elector.Although the introduction of new


2 For further information on thispoint, please refer to Attitudes toVoting and the Political Process, asurvey conducted by MORI onbehalf of the Electoral Commission,4th July 2001

3 Attitudes to Voting and thePolitical Process, 4 July 2001

4 Calculation based on figuresprovided by the University ofPlymouth for the ElectoralCommission, September 2001

5 Calculation based on figures fromElection Expenses,The StationeryOffice, February 1999.

6 The British General Election of1997, David Butler and DennisKavanagh, Macmillan Press Ltd.,1997, p.223

17

technology would be likely to increase costs initially,it is likely that savings would be made in certainareas of expenditure. Over time, new methodscould potentially become more cost-effective, andthis is certainly a factor worth examining.

Alternative Electoral OptionsThe Representation of the People Act 2000made provision for limited experimentation withour electoral procedures.This resulted in pilotprojects being carried out by 32 local authoritiesin the local elections of 4th May 2000 and in allLondon boroughs in the elections for LondonMayor and Assembly on the same day.Theseincluded postal votes on demand, universal postalballots, electronic machine voting, electroniccounting, early voting, extended polling hours,weekend voting and mobile polling stations7.

Elections within private organisations, not boundby legislation in the same way as public elections(other than the organisation’s own constitution),have used alternative methods of voting andcounting for many years, with electroniccounting, telephone voting and increasinglyonline voting being quite usual.This is also thecase for many Trade Union elections, apart fromthe top level of elections which are usuallyrequired to be postal ballots8.

Other countries have been experimenting withtheir electoral procedures for public elections formany years.The state of Tasmania in Australia hasheld council elections by universal postal votesince 1994; in Brazil, voters have been able to casttheir votes electronically since 1996, and as isnow well-known, a plethora of different electoralarrangements are in place across the UnitedStates, including the use of mechanical levervoting machines since 1892.

The options that the Commission has chosen toconsider are those which are likely to have thegreatest implications for the integrity of theballot, rather than those which merely extend oralter the polling period.They are listed and alsodefined here so that the terms used by theCommission are clear from the outset and aredistinguished from other kinds of electoralarrangements which are sometimes, confusingly,known by the same names.

Postal voting: this can either mean postal votingon demand as has recently been introduced inGreat Britain; or it can mean universal postalvoting, where there are no polling stations andeveryone on the electoral register gets a postalvote delivered to them as a matter of course.Thevote is to be returned to the appropriate placeby a certain deadline on polling day.Votes arethen counted in the traditional way.

Electronic counting: this refers solely to thecounting process.The voter goes to the pollingstation as usual and makes a mark in theappropriate manner on the ballot paper.Thecompleted ballot paper is placed in the ballotbox.The ballot papers are then counted by amachine, rather than by hand.

Electronic Machine Voting: this is also polling-station based, but combines the voting andcounting processes. It is sometimes known asData Recording Electronic (DRE). It usuallyfunctions as a closed circuit within the polling-station, not linked to any external network(though this need not necessarily be the case);the voter votes by, for example, pressing buttonsor touching a screen, and the vote is stored on anelectronic memory device within the system.There are not usually any paper ballots.Thecounting is then a relatively speedy tallying ofthese electronic records.

Telephone Voting: this is remote voting from land-line touch-tone telephones as found in homes andoffices, or from mobile phones.There are no paperballots and there is no need for polling stations.

Online Voting: this can be in polling-stations orfrom remote locations such as computer terminalsat home or at work, lottery machines, digitaltelevisions, mobile phones or other devices thatare connected to an existing external network.Again, there are no paper ballots.The countingprocess would also be incorporated, and wouldbe an electronic tallying of electronic records.

The report contains a chapter on each of theseoptions. Each chapter examines any existing useof the technology in public elections in the UnitedKingdom along with any more long-standing orlarge-scale experience overseas.


7 Piloting New ElectoralArrangements, a Local GovernmentAssociation briefing for May 2000elections, LGA Publications,April 2000

8 Here ‘top level’ refers to electionsfor the Principal ExecutiveCommittee, the General Secretaryand the President (if elected).

18

Traditionally in the United Kingdom, the use of thepostal vote has been restricted to those who, inspecific circumstances, are unable to cast their vote atthe polling station. Recent changes to postal votingarrangements in Great Britain have, for the first time,allowed for postal votes on demand to any elector.However, postal voting in the United Kingdom, otherthan in specific pilot schemes, remains a complementto polling-station-based voting.The General Election ofJune 2001 will provide the material for our examinationof postal-votes on demand, with the pilot schemes ofMay 2000 and overseas examples providing the basisof our discussion of universal postal-voting.

Background

In order to vote by post, electors must apply to havetheir name put on the Absent Voters List, along withthose wishing to cast a vote by proxy, for a singleelection, a limited period or indefinitely.The firstprovisions for postal votes in the United Kingdom weremade to enable those carrying out national serviceduties to vote in the General Election of 1945.TheRepresentation of the People Act 1948 allowed for thecreation of an Absent Voters List containing the namesof those entitled to vote by proxy or by post.Theseincluded the names of service voters, and others whocould not be expected to vote in person due to:

p the ‘general nature’ of their occupation (e.g.long-distance lorry-driver)

p duties in relation to the election

p blindness or other physical incapacity

p having to make a journey by sea or air to reachtheir polling station

p having changed address, beyond the boundariesof the borough, urban district or parish1

More recently, holiday-makers and those withshort-term illnesses were also added to the list.The Representation of the People Act 2000made postal votes available on demand for anyregistered elector in England, Scotland or Wales,even if they are currently resident overseas.Restrictions on eligibility remain forParliamentary elections in Northern Ireland (seeAppendix 1, p. 37).

With restrictions on eligibility in place, postalvotes included in the count tended to account for2 – 3% of all valid votes across the UnitedKingdom (2.4% in 1997), though the figure isusually higher in Northern Ireland (3.3% in 1997)2.In the 2001 General Election, when postal votingwas made available to all electors the figure wascloser to 5.3%3.

Postal votes at UK General Elections

General Election Postal ballots included in the count as a % age of valid votes

1974 (Feb)* 2.0%

1974 (Oct)* 2.9%

1979* 2.2%

1983* 2.0%

1987* 2.4%

1992* 2.1%

1997* 2.4%

2001** +/- 5.3%


1 The British General Election of1950, H.G.Nicholas, Macmillan &Co. Ltd, 1951

2 Election Expenses,The StationeryOffice, February 1999

3 Calculation based on figuresprovided by the ElectoralCommission in July 2001

4 For example, Strathclyde WaterReferendum, March 1994; CouncilTax Referendums in Croydon andBristol, February 2001

* Statistics for 1974 – 1997 from:Election Expenses,The StationeryOffice, February 1999

** Statistics for 2001 – apreliminary calculation based onfigures from the ElectoralCommission in July 2001

19

Postal Voting

Postal voting is more commonly used in theUnited Kingdom for elections in the privatesector, such as trade union or professionalassociation ballots where voters are scatteredacross the country, but not in sufficient numbersto merit the establishment of polling stations. Ithas also been used for a number of referendums4.

In other countries, attitudes to postal voting aremixed. In some countries, such as Italy andBelgium, there are no provisions for postal votingfor voters residing within the country. Elsewhere,such as the Republic of Ireland and France it isrestricted to voters in specifically definedcircumstances whereas in countries such asGermany, Canada and New Zealand, postal votingis relatively unrestricted.

Current Postal Voting Arrangements5

(excluding Parliamentary Elections in NorthernIreland – see Appendix 1, p. 37)

Eligibility: All registered electors in England,Scotland and Wales.The register effectively closesat the beginning of the month during which theelection timetable begins.

Application procedure & timing: By filling in andsigning a form available from various sourcesincluding the local authority Electoral RegistrationOffice, government websites or circulated withother election material. Applications must bereceived by the Electoral Registration Office by5pm on the sixth working day before theelection.

Timing of mailing: The timing and arrangementsfor the mailing of ballot materials is at thediscretion of the Returning Officer who isrequired to send out voting materials ‘as soon aspracticable’ after the close of nominations. Postalballots can be dispatched continuously from thatmoment onwards until shortly after the sixth daybefore polling day.These are sent by first-classpost by the Post Office or a commercial deliveryfirm, or are hand-delivered by clerks.Theorganisation responsible for deliveryacknowledges receipt of the number of envelopesand the date on which they were received foronward delivery.

Issuing ballot papers: Only the Returning Officerand his staff are permitted to be present whenpostal ballots are issued6. Every ballot paper mustbe stamped with the official mark, which isdifferent from the official mark on non-postalballot papers.The name and number of the voter,as they appear on the electoral register, is calledout as each ballot paper is issued.The number isnoted on the counterfoil of the ballot paper, anda mark is made against the name on the AbsentVoter’s List, to signify that a ballot paper has beenissued.The requisite number of ballot boxes forreceipt of postal voting material are provided andsealed in view of all present at the first issue ofpostal voting materials. An ‘A’ is written next tothe voter’s name on the polling station register, sothat the voter is not able to cast a vote at thepolling station on Election Day.

Other voting materials: As well as a ballot paper,the voter is sent a Declaration of Identity.Thisrecords the ballot paper number sent to thevoter. It requires the signature of the voter andthe signature and address of a witness known tothe voter to testify that s/he is the rightfulrecipient of the ballot paper.The completed ballotpaper is inserted into a small, inner envelopewhich also bears the ballot paper number.Thesmall envelope and the Declaration of Identity arethen inserted into a larger, covering envelope (seeAppendix 3, p. 38-40).

Voting period: This depends on when the ballotmaterials were sent, but is usually between oneand two weeks before Election Day.Voters areprovided with a pre-paid envelope in which toreturn their vote (as long as it is in the UK)though votes may also be handed in at a pollingstation in the relevant constituency on ElectionDay, or to the Returning Officer.Votes must bereceived by close of poll on Election Day.

Provision for spoilt/undelivered ballots: If a voterspoils his/her ballot paper, s/he is entitled torequest a replacement up until 5pm on the daybefore Election Day.The Returning Officer willissue a new ballot paper upon receipt from thevoter of all relevant voting material (includingdeclaration of identity and envelopes) from theinitial issue. If, four days before Election Day(three days in Scotland), a voter has not received

Elections in the 21st Century: from paper ballot to e-voting Postal Voting

5 Many of these provisions wereintroduced in the Representation ofthe People Act 2000 and theRepresentation of the PeopleRegulations 2001. Prior to thislegislation, the rules regarding postalvotes in Great Britain more closelyresembled those still currently inforce in Northern Ireland (seeAppendix 1, p. 37)

6 Prior to the legislation mentionedin the previous footnote, candidatesand their agents were entitled to bepresent at the issue of postal ballotpapers.This causes disquiet to somemembers of the Commission.

20

ballot materials, s/he may contact the localElectoral Registration Office to requestreplacement materials. If the request is receivedbefore 5pm on the day before Election Day, andthe Returning Officer is satisfied as to the identityof the voter and has no reason to doubt that thevoter did not receive the original materials,provisions will be made to replace the votingmaterials. A list is kept of all electors who haverequested replacement voting materials.

Verification of votes: Ballot boxes containingpostal voting materials may be opened before theclose of poll, as long as one ballot box remainsavailable for receipt of incoming materials.Candidates and their agents are entitled to bepresent.The ballot box is opened and thereturned envelopes are checked. Any envelopeswhich contain no Declaration of Identity, or anincomplete Declaration, are “provisionallyrejected” at this stage.

Counting of votes: When the inner envelopesare opened, the number on the envelope ischecked against the number on the ballot paper. Ifany of these do not match, they are rejected.

Information from Parliamentary andLocal Elections – 7th June, 2001

Parliamentary elections were held on the 7thJune, 2001; these were combined with CountyCouncil elections in county areas and with unitaryelections in a number of Unitary Authorities.

Elections were also held on the same day inmany constituencies to fill casual vacancies inindividual wards.

It is estimated that combined elections were heldin about 35% of the 659 Parliamentaryconstituencies.

These were the first major elections held afterthe introduction of rolling registration, postalvoting on demand, the shortening of thedeadlines for postal and proxy vote applicationsand a number of other changes to the postalvoting process.These changes were introducedby the Representation of the PeopleRegulations, 2001.

The County and Unitary Elections were originallyscheduled to be held on the 3rd May but werepostponed to the 7th June because of the Footand Mouth disease epidemic.

Issues around postal voting received significantpublicity, particularly the possibility of fraud andalso the extreme differences in the take-up ofpostal voting.

The effect of the Registration of Political Parties Act,1998 meant that there were no nominationproblems of substance. Candidates seeking toconfuse the electorate with descriptions close tothose of political parties or with similar names werefortunately absent so there was little for the mediato focus on in that area.With the result of theParliamentary election not looking in great doubt itmay well be that the publicity about the possibilityof postal voting fraud was one of the few electionstories left for the media to follow up, and arguablythe publicity was disproportionate to the problem.

The key issues which arose in the postal votingarea were:-

1. Following the pilots (referred to in pages 22-24) a number of Electoral Registration Officersundertook campaigns to give much greaterpublicity to postal voting. A number of EROswrote at least once to every elector invitingapplications for postal voting.Well before theelections were announced several authoritiesalready had their highest ever number of postalvoters – Stevenage, Cardiff, Newcastle andNorwich all undertook major campaigns and inStevenage 30% of the electorate registered tovote by post.The national publicity in the runup to elections on postal voting was alsosignificant and effective and many ERO/ROstook advantage of the later deadlines toinclude information about the availability ofpostal voting on the poll cards.The overalleffect of these measures was a significantincrease in the number of postal voters.

There were however very wide variances fromconstituency to constituency in the percentageof the electorate registered to vote by post,believed to range from less than 1% in someareas to over 30% in others7.


7 Figures provided by the ElectoralCommission, July 2001

21

2. The later deadline (moved from 5pm 11working days before polling day to 5pm 6working days before polling day) also resultedin increased numbers of postal voters.The 11day deadline remained for alterations toexisting postal / proxy voters.

3. Returning Officers were able to start issuingpostal votes as soon as practicable after the 11day deadline. Practices varied but most ROsissued votes on at least 3 occasions with inmany cases the final issue not taking place untilFriday 1st June.

4. The 2001 Regulations enabled ReturningOfficers to deliver postal votes through RoyalMail, by hand or using a commercial deliveryservice. In practice the vast majority of ROscontinued to use Royal Mail. Whilst theperiod between the first issue and polling daygave sufficient time for postal votes to bedelivered and returned, in the cases wherepostal votes were issued on Thursday 31stMay or Friday 1st June the period for RoyalMail to deal with delivery and return wasquite short. It is very difficult to gatheraccurate information as to how many ballotpapers were “lost in the post” either on theirway to the elector or on the return.Anecdotal information gives rise to concernabout the potential numbers.The aftermathof an unofficial strike by Royal Mail workersmay well have exacerbated the problems.

5. Each Electoral Registration Officer is requiredto prepare a list of postal voters.This is thenused by the Returning Officer (usually the sameperson) to issue the postal votes. It is thereforeknown which electors are to receive a postalballot paper.There is no provision for recordingon the list of postal voters whether a ballotpaper has been returned, meaning that there isno postal vote equivalent to the marked copyof the register.Whilst this has always been thecase, the tremendous increase in postal votenumbers has focussed attention on the issue. Amarked register serves the purpose of helpingto detect fraud, and it seems anomalous that anequivalent should not exist for postal votes. Italso means that there is no way of checkingwhether a postal vote was actually received by

the RO, and makes it very difficult to conductan audit of postal vote return.

6. Another amendment to the Regulations wasthe introduction of an arrangement for thepostal voter to request a replacement ballotpaper if the postal ballot had not beendelivered.The replacement could be requestedbetween 4 and 1 working days before pollingday.The voter or the person applying on behalfof the voter has to complete a declaration.Thisarrangement undoubtedly benefited thosevoters whose ballot papers had genuinely notarrived. However because of the tighterdeadlines and delays in deliveries the numbersof replacement ballot papers issued weresignificant in some cases. Final statistics remainto be evaluated but one constituency issuedover 250 replacements.

7. Further amendments for these elections were:-

p the ability of the voter to deliver the postalballot paper to a polling station in theconstituency. It has always been possible forthe voter or a person on behalf of the voterto hand deliver the ballot paper to the ROsOffice;

p the RO could send a postal ballot outsidethe UK (previously this was not allowed) butthe later deadlines made the time scale forthis very tight;

8. At the opening of postal ballots anotheramendment was the arrangement for theRO to “connect” a ballot paper to aDeclaration of Identity if these weredelivered separately. Prior to thisamendment, it was common where, forexample, two elderly people lived at thesame address and had postal votes, for thedeclarations and ballot papers to be mixed.Under the new regulations, this would nolonger invalidate the ballot papers.

Information from Pilot Projects

As part of the local election pilot schemes, sevenlocal authorities experimented with universalpostal ballots on 4th May 2000.These were:


22

p Bolton Metropolitan Borough Council – 3wards

p Doncaster Metropolitan Borough Council – 1ward

p Gateshead Metropolitan Borough Council – 2wards

p Norwich City Council – 2 wards

p Stevenage Borough Council – 2 wards

p Swindon Borough Council – 4 wards

p Wigan Metropolitan Borough Council – 3wards

Four authorities (Amber Valley Borough Council,Eastleigh Borough Council, Gloucester CityCouncil and Milton Keynes Council)experimented with postal votes on demand. Sincewe have already examined the General and LocalElections of June 2001 as a more recent andlarger-scale example of postal votes on demand,we shall confine ourselves here to the universalpostal voting pilots.

The pilot schemes were unusual in severalrespects.The Representation of the People Act2000, which gave rise to the pilot schemes (seeAppendix 2, p. 37), allowed the schemes to differfrom normal provisions made underRepresentation of the People Acts in relation towhen, where and how voting takes place.Thisallowed local authorities not only to holdamongst the first universal postal elections in thehistory of UK public elections8, but also toexperiment with postal voting procedures.Thismeant that in several significant respects therewere variations in the approach taken by differentauthorities.The following description of theschemes will concentrate on those areas whichdepart from normal postal-voting procedures.

Also noteworthy was the fact that, despite theRepresentation of the People Act 2000 makingprovision for schemes to take place across awhole authority, the universal postal schemes onlycovered a handful of wards in each of theauthorities.This meant that authorities were

running two different sets of voting arrangementssimultaneously.They were also doing so underconsiderable pressure of time due to therelatively late passage of the legislation.

Eligibility: In the wards concerned in the sevenauthorities, all voters on the electoral registerwere automatically sent postal voting materials.Doncaster and Gateshead in particular madeconcerted efforts to contact eligible voters whohad changed address since the compilation of theElectoral Register so that ballot papers could besent to them at their current address. Gatesheadplaced an advertisement in the Newcastle Journalto this end. Doncaster sent letters to allhouseholds asking them to inform the Council ofany changes.

Public information: All of the local authoritiesinvolved went to considerable lengths to informthe electorate of the new voting arrangementsand to explain how they would work. For mostauthorities, this involved sending informationdirectly to each household, either once or twiceduring March and April, often instead of a pollingcard.Wigan concentrated its pre-election effortson the local press, and only sent informationdirectly to households with the ballot materialsthemselves. Some authorities, including Bolton andGateshead, established a telephone helpline forthe relevant wards.

Timing of mailing: Wigan began to forward votingmaterials for hand delivery from the weekbeginning 10th April (24 days before ElectionDay). Doncaster and Gateshead were next todispatch their voting materials, by Royal Mail, on20th April (14 days before Election Day).Theothers dispatched their materials over thefollowing days up until Norwich’s final dispatch, byhand, on 27th April (7 days before Election Day).

Issuing ballot papers: This appears in all cases tohave been done in accordance with currentpractice, with the one difference being the vastincrease in numbers. All authorities had toemploy extra staff for extra hours in order tocomplete the issue.

Other voting materials: Six of the sevenauthorities included all voting materials required


8 With the exception of someCommunity Councils in Scotland,elected by universal postal ballotsince 1997 (Community Councilswithin Stirling Council) and 1998(Community Councils within Argyll& Bute Council).

23

under the current electoral provisions. Bolton, inits application to conduct a pilot schemerequested permission to dispense withDeclarations of Identity.The request was rejectedby the Home Office. However, the same requestwas granted to Gateshead which did indeed, forthe first time in a recent public election in the UK,dispense with the Declaration and the doubleenvelope system.

Voting period: As noted above, apart from somevoters in Wigan, voters in the various authoritieshad between just under two weeks and justunder one week to cast their postal ballot. Mostauthorities, with the exceptions of Bolton andDoncaster, also provided specific locations wherevotes could be delivered in person on ElectionDay.These locations were usually Council Officesand the facility was available either until 5pm oruntil close of poll at 9pm. Gateshead, where 11%of votes were returned by hand, also arranged forthis facility to be available until 9pm at WhickhamPublic Library.

Provision for spoilt/undelivered ballot papers: Itseems that this was not a major issue in any ofthe authorities. In Norwich, where six electorscontacted the Elections Office because they hadnot received their voting materials, the ReturningOfficer arranged for hand-delivery of a set ofpapers. Each elector was required to sign adeclaration to the effect that they had not alreadyreceived voting materials, that they understoodthat to vote more than once in the same electionwould be an offence, and undertaking to returnany voting materials subsequently received to theElections Office. Although it did not arise,Gateshead had decided that it would onlyprovide replacement papers to voters whosevoting materials had been returned to theReturning Officer by the Post Office as beingundeliverable.

Verification and counting of votes: Wigan wasthe first authority to start opening its votingmaterials, eleven days before Election Day. Otherauthorities began opening the materials from upto four days before Election Day. Candidates andagents either were present or had theopportunity to be. Counting appears to haveproceeded according to normal procedures.

Media attention: Above and beyond their ownpress releases, the seven local authorities appearto have had good media coverage of the pilotschemes.

Information from Overseas

In the United Kingdom, experience of universalpostal ballots in public elections has beenconfined to one-off pilots in a handful of wardswithin individual local authority areas.There areseveral instances elsewhere of such electionstaking place across an entire state/territory oreven an entire country either as a one-offreferendum or election, or on a more regularbasis.The following examples are far fromexhaustive, but have been chosen because oftheir significance in terms of scale or regularity.They have also taken different approaches tosignificant aspects of the postal-voting process.Those aspects will provide the main focus ofwhat follows.

The 1997 Australian Election for theConstitutional Convention

The issue of constitutional change in Australia hadbeen gaining prominence for some time whenthe Liberal-National coalition government waselected in 1996 and pledged to hold aConstitutional Convention to discuss the issuesrelating to the possibility of Australia becoming aRepublic.Those attending the ConstitutionalConvention, in February 1998, were to be 152delegates of whom 76 would be elected and 76appointed.

It was decided that the 76 elected delegateswould be chosen by non-compulsory publicelection conducted by the Australian ElectoralCommission (AEC) and that the number ofdelegates would broadly reflect the balance ofrepresentation of the States and Territories in theFederal Parliament. After considering severaloptions, it was decided that the election would bea nationwide universal postal ballot. Election Daywas on 9th December 1997.

It is worth noting that registration to vote iscompulsory in Australia, and that it is a ‘rolling’process.When the register closed on 7th


24

October 1997, it covered an electorate of 11.9million.Voting in public elections in Australia isalso compulsory and this particular election wasan exception in that respect. Although the AEChad previously conducted local elections byuniversal postal ballot, this was to be the largestconducted in Australia.

Australia is made up of six States and twoTerritories.The AEC has its Central Office inAustralian Capital Territory, a Head Office in theremaining Territory and each of the six States. Italso has Divisional Offices in or near each of the148 electoral divisions.

Eligibility: Everybody on the electoral register on7th October 1997, two months before election day.

Public information: This was substantial andincluded advertising, public relations, a nationaltelephone enquiry service and the internet.Advertisements were placed (multi-lingually) inthe press and on the radio and television. Local,regional and national media were regularly briefedand featured interviews with AEC staff.Thenational telephone enquiry service operated onevery weekend throughout the election periodand was available in several languages. Particulargroups were specifically targeted; these weremostly non-English-speaking or Aboriginalcommunities, or visually impaired electors.

Timing of mailing: The first consignments ofvoting materials were delivered to Australia Postoffices on 28th October (42 days before close ofpoll) and the final ones were delivered to postoffices on 11th November (28 days before closeof poll).This process was largely automated, withthe help of mailing houses for producingpersonally addressed envelopes.These weredelivered by Australia Post except for areas (180remote communities) without a regular postalservice, where voting materials (19,000 votingpackages) were delivered by hand. Delivery ofthese packages began earlier than the othersbecause of the extra time needed both fordelivery and return.

Issuing ballot papers: Candidates were notpermitted to be present at the issue, or any otherpart of the conduct of the election, however, they

were entitled to appoint scrutineers. Inaccordance with Australian electoral law, theballot papers did not bear any mark that could beassociated with an individual voter.

Other voting materials: Along with the ballotpaper, voters received a blue ballot paperenvelope and a reply-paid envelope.The ballotpaper envelope had a tear-off flap attached to it,which could be removed without opening theenvelope.The flap was pre-printed with thevoter’s name, address and unique barcode whichrepresented the voter’s electoral registrationnumber. In addition, on this flap, was a declarationwhere the voters were required to enter theirdate of birth and to sign a statement to the effectthat they were the person named on theenvelope, that they were entitled to vote, andthat they had not already voted in this election.The voting package also included a State andTerritory-specific information booklet entitled‘Your Guide to the Constitutional ConventionElection’ which included information on how tovote and a list of candidates, their identificationnumbers and statements.

Voting period: Voters had between almost sixweeks and almost four weeks to return theirvoting papers, by 6pm on 9th December. Ballotpapers were returned by post to the DivisionalOffices, though voters were also entitled to delivertheir ballot papers by hand to any AEC office.

Provision for spoilt/undelivered ballot papers:Voters were able to request replacement votingpackages if they had lost or spoilt their ballot paper,or if the package had not been received.Therequest had to come directly from the elector.These were sent by post until 6pm five days beforeclose of poll, and were available to be collected inperson from any AEC office until close of poll. Over49,000 replacement packages were sent, mostly inresponse to requests via the national telephonehelpline from voters who had not updated theiraddress before the close of the electoral register.

Verification of votes: This was done at theDivisional Offices and the process began as soon asthe envelopes started to arrive.The outerenvelopes were opened and the blue ballot paperenvelopes extracted. Each Divisional Office had a


25

CD-ROM with a list of its electors, their addressesand dates of birth, created at the close of theelectoral register.These offices were also providedwith a barcode reader and software which, uponscanning the barcode, could bring up the elector’sdetails on the screen from the CD-ROM.Thebarcodes were scanned to ensure that the electordetails on the envelope and the CD-ROMmatched, to make sure that they were entitled tovote and also to ensure that the elector had notalready voted. If there were any discrepanciesbetween the voter’s details (including the date ofbirth), or if more than one vote was received in thename of a particular voter, the Returning Officer –checking the date of birth and comparing thesignature to that on the most recent application forelectoral registration – decided which, if either/any,of the votes would be counted.Any envelopeswhich did not meet the verification requirements(for example, those without a signature) wererejected at this stage.The ballot paper envelopeswere then stored, unopened until the count.

Counting: Counting was conducted at DivisionalOffices and also at State and Territory Head Offices.Candidates’ scrutineers were entitled to be presentat the count, though the candidates themselveswere not.The detachable flaps on the ballot paperenvelopes were removed at this stage (withoutopening the envelopes) to maintain the secrecy ofthe ballot.The envelopes were then opened andthe ballot papers extracted and a count done tomake sure that the number of ballot papersmatched the number of envelopes.The countingbegan once all of the ballot papers had beenclassified as being valid or invalid. Because the votingsystem being used was a transferable one, thecounting process was necessarily somewhat lengthy,however some of the results were data-inputtedand counted by a computerised system.Though thedeadline for the result was 6th January 1998, theresult was actually finalised within two weeks andofficial notification of the results, in a formalannouncement by the Electoral Commissioner, wason 24th December 1997.

Universal Postal Voting in Oregon, USA

The State of Oregon has been experimentingwith universal postal voting since 1981 when theState Legislature approved its use for local

elections. By 1987, it was the method used bymost of Oregon’s 36 counties for their local andspecial elections and 1993 saw the first State-wide universal postal election.

1998 was an important year in the developmentof universal postal voting in Oregon. 41% ofOregon’s electors were now registered aspermanent absentee electors, meaning that theyhad opted for a postal vote for every electionuntil further notice. At a polling-station-basedPrimary election in May, almost two-thirds of thevotes cast were postal votes. In November, votersin Oregon passed a measure directing all publicelections to be held by universal postal ballot.

Ten State-wide elections have taken place usinguniversal postal voting.The most recent of thesewas the Presidential Election in November 2000with an electorate of 1.9 million and the electoralarrangements used for this particular election willform the basis of the following discussion. It isimportant to note that decisions about manyaspects of electoral arrangements are made notby the State, but at county level.

Eligibility: Applications for registration had to bereceived by (or postmarked no later than) 17thOctober, 21 days before election day. However,electors already registered in Oregon werepermitted to update their details (including theiraddress) after this date up until and includingelection day itself.

Public information: Information was sent to allelectors both in advance of the ballot and insome counties with the voting materialsthemselves.This consisted of a voters’ pamphletfrom the Secretary of State which includedinformation about the nature of the electionand the location of ‘drop-sites’, about thecandidates and issues to be voted on andstatements from the political parties.Thisinformation was also available online on theSecretary of State’s website. Some countiesproduced county-specific voters’ pamphlets inaddition.Telephone numbers for countyelections offices were also provided.

Timing of mailing: Most voter packs weredispatched, via the United States Postal Service,


26

18 – 14 days prior to election day9.Voter packsfor overseas electors were sent from 45 daysbefore the election, and for electors in other USStates from 29 days before the election.

Issuing ballot papers: Members of the publicwere permitted to observe all stages of theelectoral process.There were no counterfoils forthe ballot papers, and there is not permitted tobe any method of matching an individual ballotpaper with a voter.

Other voting materials: In addition to the ballotpaper, electors received a secrecy envelope, anouter envelope in which to return the ballotpaper and in some counties a localised electioninformation booklet.The completed ballot paperhad to be inserted into the secrecy envelopewhich in turn was to be inserted into the outer‘return’ envelope.The outer envelope, apartfrom the return address, included the name,address and voter identification of the voter andrequired the voter’s signature. In addition, thevoters needed to affix a 33cent stamp for thereturn postage.

Voting Period: Voters had roughly two weeksduring which to return their ballot papers, whichhad to be received by the deadline of 8pm onthe 7th November.

Ballot papers could be returned either by post, orby hand to any county election office (one percounty) or to any designated ‘drop-sites’ withinthe State.There were at least two drop-sites percounty and they were required to be open onElection Day for a minimum of eight hours upuntil 8pm. Some drop-sites were available duringthe week preceding the election, and some weresited in ‘drive-by’ locations.

Voters who, for whatever reason, did not wish tocast their vote at home, had the right to cast theirvote in a privacy booth.These were availablethroughout the period that ballot papers werebeing issued, at county election offices and atsome of the drop-sites.These booths were alsomade available for electors who had changedtheir registration details too late to enable a ballotto be sent to them by post.

Voters were permitted to call their countyelection office to confirm that their completedballot paper had been received.

Provision for spoilt/undelivered ballot papers: Inthe case of lost, damaged, spoilt or undeliveredballot papers, voters were able to request areplacement voter pack up until 5 days beforeclose of poll from their county elections office.After that, replacement voter packs were availablefor collection from the county elections office. Arecord was kept of all voters to whomreplacement voter packs were sent, in case ofreceipt of duplicate/multiple ballot papers.

Verification of votes: This process, which beganwithin a week of the initial dispatch of voterpacks, took place at the county election offices.Members of the public were permitted to bepresent.The process involved verifying thesignature on every outer envelope by comparingit with either an electronic signature held oncomputer, or the signature on the most recentregistration application. If there was no signatureon the envelope, attempts were made to contactthe voter and arrange for them to come to thecounty elections office to sign the package, or fora replacement voter pack to be sent, or for thepackage to be returned to them for signing. If theenvelope was not signed before close of poll, thevote was not counted. In cases where thesignature could not be verified an attempt wasmade to contact the voter and resolve theproblem.

A record was kept of each voter who hadreturned a ballot paper, enabling administrators tocheck that they were registered to vote, and thatthey had not already voted in the election.Theenvelopes were then stored, unopened until thenext stage.

Five days before close of poll, the outerenvelopes were permitted to be opened. Onceagain, members of the public were entitled to bepresent, and election personnel were required tobe from different political parties.The totalnumber of returned envelopes was recordedbefore the outer envelopes were opened and thesecrecy envelopes removed.The outer envelopeswere retained in order to serve the purpose of a


9 In previous universal postal ballotsin Oregon, voter packs were sent21 – 18 days before election day.This was reduced in order to allownew voters to register up until 21days before election day.

27

‘poll book’. From this point onwards, the ballotswere totally anonymous. If there was no secrecyenvelope (only a ballot paper) or if the secrecyenvelope was unsealed, but the outer returnenvelope was sealed, then the ballot paper wasaccepted. If the outer return envelope wasunsealed and the secrecy envelope was alsounsealed or absent, the ballot paper was rejected.The secrecy envelopes were then opened andthe ballot papers extracted. Ballot papers werethen inspected for irregularities and prepared forthe counting process.

Counting: The actual process of counting beganon Election Day itself, allowing a significantproportion of votes to be counted, or otherwisedealt with, before close of poll at 8pm, though noresults were released until then. Different countiesused different vote-counting methods and thesefell into the categories of punch-card/data votesystems and optical scan systems.

Evaluation of Universal Postal Voting

Our discussion of the implications of universalpostal voting will fall into the following categories:

p Integrity (secrecy and security)

p Participation (turnout and the quality of theprocess for the voter)

p Efficiency (cost and time)

p Complementary processes (public informationand party campaigning)

Integrity

a) SecrecyThe right to a secret vote is one of the keyfactors in the conduct of a free and fair election. Ifthis right can be guaranteed, then bribery andcorruption in the form of vote-buying or undueinfluence cannot exist.The circumstances in whicha voter casts his/her vote is crucial to themaintenance of this right. Evidently, in a polling-station-based election, it is relatively easy toguarantee that voters have total privacy in whichto cast their vote. As soon as voters arepermitted to cast a vote in a location other than

a polling booth within a polling station, thiselement of the process becomes impossible tomonitor and extremely difficult to control.

Under United Kingdom law, undue influence ispunishable by up to five years in prison and anunlimited fine.

Other than publicising the potential penalties, it isextremely difficult to deal with the problem ofsecrecy. Although there were no allegations ofundue influence during the UK pilot schemes, theAustralian universal postal election or in any ofthe Oregon elections, it is a permanent risk, and onthat basis, cannot be ignored. It is worth mentioningthat in Oregon, ‘privacy booths’ were providedthroughout the polling period at county electionoffices (where the ballot papers are issued) and inother accessible locations, for those who wished tomake use of them for whatever reason.

b) SecurityThere are a number of issues regarding thesecurity of the electoral process that are raisedby postal voting.These can take place on a small,domestic level or on a large organised scale.

Since a detailed examination of the process ofelectoral registration is beyond the remit of thisCommission, and since improvements to thatprocess, in terms of rolling registration, haverecently come into force, for the purposes of thisdiscussion we shall merely state that an accurateand up-to-date register is an essential pre-requisite for a secure election, particularly whenpostal votes are involved.

Assuming that voting materials are delivered tothe correct addresses, there still remains theproblem of multiple votes being delivered to asingle residence. On a small-scale this could meanindividuals within a house having access to theballot papers of several other people. On a largerscale, there is the situation of a large multipleresidence such as an old people’s home or astudent hall of residence which may have open‘pigeon-holes’ for post, accessible not only to allresidents but also, potentially, to any visitors. Inthese situations, it is not difficult to imagine asingle person (whether or not they are aregistered elector) obtaining multiple ballot


28

papers and voting several times.This is the crimeof ‘personation’ under UK law.

This problem is not limited to postal-votingsince no formal personal identification isrequired in order to cast a vote in a polling-station beyond simply stating one’s name andaddress. However, large-scale fraud would bedifficult in this context due to the danger ofbeing recognised as a multiple-voter by polling-station staff, or to needing to organise a largenumber of different people willing to votefraudulently.

“…in reviewing the scheme, both politicians andourselves the officers feel that we would need tolook again at the security aspects of all-postalvoting.While no system is immune fromabuse…having to vote in person is a checkagainst personation.The existence of markedregisters…means that there is a check on who’svoted – it’s a matter of public record…andunder postal voting I would acknowledge thatyou don’t have these checks.”

Jonathan Rew, Gateshead MetropolitanBorough Council

The Declaration of Identity has been themethod used to deter such actions in theUnited Kingdom, both in ordinary elections andin the pilot schemes (see Appendix 3, p. 38-40).The Declaration requires the signature of avoter and a witness to the effect that the voteris indeed the person they claim to be.Theexception to this rule was the pilot scheme inGateshead which, with the permission of theHome Office, dispensed with Declarations ofIdentity and did not replace them with anyother security measures.There were noallegations of fraud relating to any of theuniversal postal pilot schemes.

Declarations of Identity are problematic in anumber of respects. Firstly, they work as a purelytheoretical deterrent since they are not, andcannot be, checked in any way; ElectoralRegistration Offices do not have a record ofelectors’ signatures. It would not be difficult for amotivated individual to find a single willingaccomplice to act as a false witness, or even for

an individual to fill in the form in two apparentlydifferent hands; in either case it is unlikely thatsuspicion would be aroused.

Secondly, there is evidence to suggest that theDeclarations are not especially user-friendly, andcan cause electors, inadvertently, to spoil theirvote. Most of the authorities conducting universalpostal pilot schemes had significant percentages ofspoilt ballots, ranging from Bolton’s 2.5% toDoncaster’s 8.4%, the majority of which were dueto incomplete Declarations of Identity. Gateshead,without Declarations, had 0.25% spoilt ballots. Atthe 2001 General Election, the rate of spoiltballots amongst postal voters was 2.25%, whilstthe average rate was 0.26%10.

“The biggest public reaction was against theDeclaration of Identity. People resented whatthey saw as the third party interference in thevoting process. And also that by returning thevoting paper, albeit in a separate envelope, witha Declaration of Identity, that there was a loss ofsecrecy in the ballot.There was a strong reactionto that.”

Des Grogan, Bolton Metropolitan Borough Council

As shown in the case-studies from Australia andOregon, there are alternatives to the kind ofDeclaration of Identity currently used in theUnited Kingdom.These involve a check on thevoter’s identity by comparing information heldabout the voter (collected at the time ofregistration) with information provided by thevoter at the moment of casting a ballot.Thesealternatives mean that it is necessary to collectand store extra information relating to electors,such as a signature and/or a date of birth, duringthe process of electoral registration. Automatedprocesses, whether by means of barcodes, oridentification/serial numbers, for the storage andretrieval of this information can evidentlystreamline the process.This is a possibility that isalready being considered in the United Kingdomin relation to elections in Northern Ireland11.Thiskind of solution would require the collection ofadditional information when the annual canvasstakes place or on completion of the registrationapplication and could significantly increase the


10 Calculation based on figuresprovided by the ElectoralCommission, September 2001

11 Electoral Fraud (NorthernIreland) Bill, introduced in theHouse of Commons on 28th June2001

29

cost of preparing the register.There is a possibilityalso that it could discourage registration.

Though voters would be required to providesupplementary information about themselves,they would no longer have to involve a thirdparty, as a witness, in the process. It would also beinformation easily memorable for the voter, butunlikely to be known to many other people andwould have the benefit of being checkable. Large-scale fraud could be virtually eliminated, andthough small-scale fraud could conceivably takeplace (e.g. within a family who are familiar witheach others’ dates of birth and, possibly,signatures), the deterrent would be significantlygreater than at present due to the facility tocheck the information.

These kinds of mechanisms would also have thebenefit of allowing the creation of a record ofthose who have voted (an equivalent of themarked register produced for votes cast atpolling-stations), and checks to be made onwhether a vote has already been cast in the nameof a particular voter.

These identity checks could take the form of aDeclaration of Identity, involving a signatureand/or a date of birth, to be completed by thevoter (without a witness), and then enclosed,along with the ballot paper envelope, in the outerreturn envelope.This solution, however, wouldretain the high number of separate pieces ofstationery sent to electors. It would also do littleto reduce the complexity of the current processand the consequent potential for error leading tothe invalidation of ballot papers.

Alternatively, as in Australia, the identity checkdetails could be printed on a detachable flap onthe ballot paper envelope.This would reduce thepieces of stationery and would remove some ofthe complexity.The rate of rejected ballots dueto incomplete details on the detachable flap inthe Australian Constitutional Convention electionwas 1.49%.

In Oregon the identity checking information wason the outer return envelope, reducing to aneven greater extent the amount of stationeryand the complexity of the process. Questions of

security could arise however over outerenvelopes bearing the identity of the voter.

Clearly, whichever method is used, for the sake ofsecrecy, the voter details must be separated fromthe ballot papers by the time of opening the ballotpaper envelopes and extracting the ballot papers.

Participation

a) TurnoutAs discussed in the Introduction to this report,electoral turnout is a complicated issue involvingmany different and largely immeasurable factors.The basis on which postal voting can claim tohave a bearing on turnout, is convenience.Votersare not compelled to vote during specific hours,or even on a specific day.They can vote at amoment which is convenient to them, which doesnot clash with, for example, professional, domesticor religious commitments, and with any relevantmaterials (instructions about how to vote, orparty campaigning material) to hand.Votes can bereturned from the most convenient postbox.

Several of the local authorities conducted post-pilot surveys and these overwhelmingly showedpositive responses to universal postal voting interms of convenience (92% of respondents inSwindon found universal postal voting moreconvenient; 82.6% of respondents in Stevenagepreferred to vote by post and 100% ofrespondents in Wigan found voting by post easy).A survey conducted in Oregon in 199612 foundthat 76.5% of voters preferred to vote by post,mainly due to increased convenience.

With the wards involved in the pilot schemes it ispossible to make a direct comparison with theturnout in the concerned wards in the previousyear’s polling-station-based elections, as well aswith the average turnout across the authority,though it is important to bear in mind that levelsof public information and media attention arelikely to have been higher than normal.

There was an increase in the percentage of votescast in all 17 universal postal voting wards.Theincreases ranged from 6% in Bedwell (Stevenage) to32% in Whickham North (Gateshead).The turnoutfigures for all wards appear in Appendix 4, p. 41.


12 Final Report: Survey of Vote-by-Mail Senate Election, PriscillaL.Southwell, Department of PoliticalScience, University of Oregon,April 1996

30

“…7,300 people over the two wards voted,which was roughly 4,000 more than in 1999. So,it actually brought 4,000 people into thedemocratic process.”

Jonathan Rew, Gateshead MetropolitanBorough Council

The Australian Constitutional Convention electionhad a turnout of 47%. It is not easy to make anymeaningful comparisons as this was a one-offelection on a very specific issue. It was also avoluntary ballot rather than a compulsory ballot asis the case for most public elections in Australia.

Oregon provides more scope for makingcomparisons.The turnout in the PresidentialElection in November 2000 was 79.8% whereasthe national average was 50.3%, though it shouldbe noted that Oregon has a tradition of higherthan average turnout in Presidential elections.Theprevious Presidential elections in 1996 werepolling-station-based (though almost 39% of theelectorate were registered as absentee voters);the turnout was 71.31%.

Since 1996, the average turnout of polling-station-based elections in Oregon has been 47.44%whereas the average for universal postal ballotshas been 56.12% (see Appendix 5, p. 42).

It is interesting to note that a nationwidereferendum in New Zealand in 199713, the first inthe country to be conducted as a universal postalelection, achieved a turnout of 80.3%.Thisrepresented the highest turnout ever recorded inthe country for a stand-alone referendum.

b) Assistance to electorsWhen votes are cast in a location other than apolling station, voters are to some extentdeprived of assistance from an impartial official.

On the other hand, material can be included withthe voting materials which gives step by stepinstructions about how to cast a vote. It is alsopossible to make assistance available on atelephone helpline.

c) Community considerationsMany people value the experience of

attending a polling station on election daybecause of the feeling it gives that they areparticipating in the exercise of democracy andare seen to be doing it. Many, particularly inrural areas, look forward to the social elementthat is part of the present election day activity.This emerged in several of the pilot schemewards (3 calls to the helpline in Boltoncomplained about not being able to vote in apolling station; 7% of respondents to the post-election survey in Swindon and 16% ofrespondents in Stevenage would havepreferred to have voted in person).

The only measures which address this issue toany extent, are the provision of dropping-offpoints for ballot papers in public places, such aspublic libraries or town halls, and the provision ofprivacy booths in suitable locations.

Efficiency

a) Cost The evidence relating to the cost of runninguniversal postal elections is mixed. Both Oregonand Australia show significant savings whilst thelocal authority pilot schemes were significantlymore expensive than polling-station-basedelections.

In Australia, the Electoral Commission reportedsubstantial savings.The Constitutional Conventionelection cost AUS$ 4.28 per voter (or AUS$ 2.01per elector) whereas the cost of the FederalElection in the previous year, 1996, withcompulsory voting, was AUS$ 5.10 pervoter/elector.

In Oregon, the polling-station-based 1998Primary Election cost $5.10 per voter ($1.78 perelector) in contrast with the universal postal2000 Primary Election which cost $3.03 pervoter ($1.56 per elector).

All of the local authorities that piloted universalpostal voting experienced net increases in cost,with per ward increases ranging from £3,400+ inSwindon and Wigan to £10,686 in Doncaster.Doncaster estimated that universal postal votinghad meant an increase in cost per elector from£0.75 to £1.12.


13 Compulsory Retirement SavingsScheme Referendum, 5th – 26thSeptember 1997

31

Authority Overall Net increasenet increase per ward

Bolton £12,674 £4,225

Doncaster £10,686 £10,686

Gateshead £11,000 £5,500

Norwich £11,720 £5,860

Stevenage £7,200 £3,600

Swindon £13,728 £3,432

Wigan £10,463 £3,488

Universal postal elections provide the opportunityfor savings to be made because the followingitems of expenditure can be dispensed with:

p hiring buildings for polling stations

p employing polling station staff (PresidingOfficers, Poll Clerks and Polling StationInspectors)

p delivering and collecting polling booths andballot boxes

p printing and delivering poll cards

However, there are also areas of greaterexpenditure:

p additional stationery printing

p staff time for issuing and opening ballot materials

p postage (outward and return)

The extent of additional printing can be limitedby replacing a Declaration of Identity with detailsprinted on one of the envelopes.There could alsobe economies of scale if universal postal votingwere extended across areas greater than two orthree wards.

The issuing and opening of ballot materials wasdone by hand in the local authority pilot schemes.

In both Australia and Oregon a significant degreeof mechanisation was introduced into theprocess, to make the process quicker and lessdemanding of staff time.

“If we’re going to have to do it in the future,we’d have to look at mechanical insertionequipment”

Des Grogan, Bolton Metropolitan BoroughCouncil

Although the outward postage is fixed andpredictable in relation to the size of theelectorate, the return postage will increase thehigher the turnout. In Oregon significant savingswere made by putting the burden of the returnpostage on the voter.

b) Time considerationsThere are three key areas, in terms of timing, inwhich postal votes can differ considerably fromvotes cast at a polling-station.

p issuing voting materials

p length of polling period

p verification and counting

Issuing voting materials for an entire electoralarea is a potentially time-consuming process.Mechanisation of the process of inserting materialinto envelopes would speed the process upconsiderably, and would make staff available forother tasks.There are machines which can fillenvelopes with several items at a maximum speedof 30,000 envelopes per hour, and machines whichopen envelopes at a rate of about 10 – 12,000per hour. After the pilot schemes, both Norwichand Stevenage pointed out that the requirementfor each ballot paper to be stamped with theofficial mark at the time of issue also contributedto inefficiency in the process of issuing votingmaterials. Stevenage recommended the use ofballot papers with watermarks instead. Norwichalso suggested that the requirement for eachballot paper to have a counterfoil be dropped.

The length of the polling period will be affectedby a number of factors, including compatibility


32

with other elements of the electoral timetableand the length of time that is considered to bepreferable for voters to cast their vote.

The most important part of the electoraltimetable in relation to the timing of postal votesis the close of nominations.This is on the eleventhday before the Election and the statement ofpersons nominated is published as soon aspossible after 5pm on that day. Postal votescannot be sent prior to close of nominations;however, the process can begin as soon aspracticable afterwards.

In the local authority pilot schemes, most votershad a period of 1 – 2 weeks in which to casttheir votes.The Australian Constitutional Electionallowed voters 4 – 6 weeks whilst in Oregon thepolling period is about 2 weeks.The 1997 universalpostal referendum in New Zealand14, mentionedabove, allowed a 3-week period for voting.

Voting patterns tend to show that the shorter thevoting period, the sooner people cast their votes.In the New Zealand referendum, 45.7% of ballotpapers had been returned by the end of the firstweek of voting. Similarly, in Gateshead, which hada two-week voting period, 55% of papers werereturned within the first week. Australia with itssignificantly longer polling period, received only11% of all ballots during the first week, thoughthis had risen to a total of 48% by the end of thesecond week.Thereafter, the weekly rate droppedto 27%, 14% and 7% with a slight increase in thelast few days15.

A lengthy voting period could lead to a numberof problems. For example, Doncaster, whosevoting period, at two weeks, was not especiallylengthy, received a comment from a candidate inone of the pilot wards who felt that the length ofthe polling period meant that some electors hadbegun to lose interest.Wigan’s post-pilot surveyof electors found that 31% of non-voters claimednot to have voted simply because they hadforgotten.The authority felt that posting votingmaterials closer to Election Day would help toreduce such forgetfulness. Evidently, anyextension in the voting period could exacerbatethese problems and potentially have a negativeeffect on turnout.

A longer voting period could be problematic interms of political party campaigning (see page34). In addition, any move away from the snap-shot that is provided by voting on a singleelection day, allows the possibility of significantdevelopments occurring during the voting period,meaning that later voters could be influenced bydifferent factors from the earlier voters.Thelonger the voting period, the greater the potentialfor this kind of discrepancy.

On the other hand, a voting period should not beso short as to reduce the possibility for increasedconvenience that is offered to the elector.Therealso needs to be adequate time to provide anyassistance that voters might need and to replacespoilt or undelivered ballot papers.

Verification of votes can be more-or-lesscompleted before election day, and this was donein almost all cases examined above. However, theissue of when to open the inner envelope is amore difficult one. On the one hand, time couldbe gained by beginning to open the envelopesearly on during Election Day, and preparing themfor the count, so that the count could proceedrelatively quickly on close of poll. However, in thecase of a universal postal election, with candidatesand their agents entitled to be present, it wouldbe possible for them to gain a fairly accuratepicture of the final result before the close of poll,simply by viewing the opening of the envelopes.This was an issue raised by a candidate in one ofthe pilot wards in Gateshead which beganopening envelopes at 1pm on Election Day. Onthe other hand, waiting until close of poll beforeopening the inner envelopes would considerablylengthen the counting process.

Complementary processes

a) Public InformationPublic information is an important part of anyelection. It can inform electors that an election isabout to take place, of the nature of the electionand the candidates who are standing, of the votingarrangements that are in place and how to cast aballot, as well as who to contact for assistance.With postal voting, when voters do not necessarilyhave access to information or assistance at thepoint of voting, it is even more important.


14 Compulsory Retirement SavingsScheme Referendum, 5th – 26thSeptember 1997

15 Voting by Mail, InternationalInstitute for Democracy andElectoral Assistance, 1999

33

Public information can be through the media, orcan be delivered directly to each elector eitherbefore the polling period or with the votingmaterials themselves. In Australia, efforts wereconcentrated on a high-profile national, regionaland local media campaign leading up to theelection. Detailed information was delivered toelectors with the voting materials. In Oregon,information is delivered to electors prior to thepolling period, including candidates’ statements.

The local authorities differed slightly in theirapproaches.Wigan focussed primarily on themedia and did not deliver information directly toelectors except with the voting materialsthemselves. Its post-election survey revealed thatonly 49% of respondents could rememberreading anything about the pilot scheme.Gateshead, on the other hand, deliveredinformation directly to the electors on twooccasions prior to the delivery of votingmaterials and found that 96% of respondents toits survey had been aware of the universal postalvoting pilot scheme. It is also interesting to notethat the Elections Office in Gateshead receivedonly eight enquiries or complaints about thevoting process.

Where telephone helplines were in place theyappear to have been heavily used. In Australia,over 227,000 calls were received. Bolton receivedabout 75 calls. In both cases, a large proportion ofthese enquiries related to electoral registration,and in Australia, requests for replacement votingmaterials. Over 10% of Bolton’s telephoneenquiries were from electors concerned that theinclusion of a Declaration of Identity would meanthat their vote would not be secret.

b) Political CampaigningThe extension of the voting period from onespecific “polling day” to a longer period inevitablyhas an effect on the way in which political partiesconduct their campaigns.Whilst campaigning cancontinue throughout the polling period, partieswill not know who has and who has not cast avote or even whether there are areas of high orlow turnout.This means that the most influentialperiod in terms of reaching electors who havenot already cast a vote, will be prior to thedelivery of voting materials, effectively curtailing

the campaigning period. For political parties, thismeans focusing campaigning efforts on the periodbetween close of nominations and delivery ofvoting materials.

The other important factor for political parties toemerge from the pilot schemes was that, due toHome Office restrictions, no marked register wasproduced.Whereas, in a polling-station-basedelection, each elector who casts a vote has amark placed next to their name on the electoralregister, with the universal postal voting schemes,a mark was placed next to every name on theregister because a ballot paper was issued toevery elector. No record was made of those whohad voted as votes were returned.This is a usefultool for political parties in planning future electioncampaigns, and potentially also for localauthorities for the purpose of analysis. Becausethe marked register is open for public inspection,it can also help to identify fraud.

Conclusions and Recommendations

The cumulative effect of the alterations to theRepresentation of the People Acts andRegulations, the increase in the numbers of postalvoters, and piloting of universal postal voting hasbeen to focus attention on the postal votingprocess.This has happened amongst theelectorate, the media, those concerned about thesecurity of the process and those attempting toundermine it and use it to their own ends. Clearlyvoting other than in the polling station increasesthe potential for fraud, and the more postal votesin question, the greater the opportunities.Whilsta number of cases are being investigated, theissue of fraud issue needs to be kept underreview and in perspective.

Of equal importance is the efficiency of theprocess of delivering and returning ballot papers.It is now impossible to know how many voters inthe General Election of 2001 weredisenfranchised due to their votes being held upin the post, but it is certainly an area in whichimprovements need to be made.

Evidence from overseas suggests that there areways in which these and other aspects of postalvoting could be improved and made more secure.


34

The Commission believes that the potentialbenefits in terms of increased convenience andparticipation coupled with the desire to findcreative ways of limiting the potential riskscertainly merit further structuredexperimentation.

Our recommendations, below, set out the basison which we believe further experimentationshould proceed.

Recommendations

1. Further piloting of universal postal voting isnecessary, and the Commission recommendsthat universal postal voting be piloted in localelections, European Parliamentary elections,and Parliamentary by-elections across awhole authority, electoral region orconstituency respectively, and also in Parishand Community Council elections. Evidencediscussed in this chapter suggests that theseelections would lose least and potentially gainmost from using universal postal voting,particularly as regards potential increases inturnout. For the time being, turnout atGeneral Elections remains high in comparisonwith public elections at any other level, andso we believe that postal-voting on demandremains appropriate and would not suggest amove to universal postal voting.

2. The current methods of adding security tothe postal voting process are far fromperfect. Declarations of Identity are at best aweak deterrent, they create considerabledifficulties and expense for electoraladministrators, and confuse electors to theextent that votes are unintentionally spoilt.

Should the Government be minded to makechanges to the way in which electoralregistration functions (as has been proposedin the Electoral Fraud (Northern Ireland) Bill),then procedures involving dates of birth andpossibly signatures could be used tostreamline the security process from the pointof view of the electors and theadministrators, whilst actually increasing thesecurity of the process. Instead of finding awitness, voters would simply provide their

date of birth and/or signature, ideally on atear-off strip attached to the ballot paperenvelope.The strip would be removed afterverification in order to maintain secrecy.

If such changes are not envisaged, then theCommission sees no option but to retain theDeclaration of Identity, providing as it does alimited deterrent.Thought should be given,however, to the simplification of the languageand layout of the form, and increasing theprominence of warnings about the penalties ofcommitting fraud.

3. The lack of a postal voting equivalent to themarked copy of the register is a major failing –the introduction of such a system should belooked at as a matter of urgency.This could bedone in several ways:

a) by printing the voter’s (name and)electoral number on the reverse of theballot paper envelope, and simply tickingoff the name on the postal voters list.However, this option would make it moredifficult to ensure the secrecy of the ballotsrequiring, as it would, those present tokeep the envelope face upwards whilst theballot paper was removed

b) encoding the voter’s name and/orelectoral number in a barcode to beprinted on the ballot paper envelope andthen swiped during the verificationprocess, thereby creating a marked registerand ensuring secrecy

c) printing the voter’s name and number on atear-off strip attached to the ballot paperenvelope; the voter could be ticked off thepostal voters’ list and the removal of thestrip after this process would preservesecrecy

Such arrangements would give a list of postalvoters whose votes were returned andprevent double voting where replacementshad been issued. It would also allow checksto be made on whether a vote had been castin the name of, for example, a recentlydeceased elector.


35

4. In the case of postal voting on demand, whenan elector has applied for a postal vote, butthe postal vote has not been delivered andthere is not enough time to apply for areplacement, the voter should be entitled toattend a polling station in the relevantconstituency and be issued with a tenderedballot paper after answering the statutoryquestions (suitably amended). In the case of auniversal postal election, similar proceduresshould be in place at a location such as theTown Hall, or wherever the Returning Officeris based. At present a postal voter in thesecircumstances is totally disenfranchised.

5. The postal vote application form (in the caseof postal voting on demand) should includedetailed information as to the timetable forissuing postal votes (this is the same for allelections).This would help to ensure thatelectors going away during the election periodwere aware of the likely date of delivery of thepostal vote in relation to their departure, and givethem the opportunity to consider the option of aproxy vote if that were more appropriate. In auniversal postal election, such information shouldbe sent to electors at the earliest opportunity,possibly in lieu of a polling card.

6. The Commission is aware of concernsexpressed by electoral administrators inrelation to the performance of the Post Officein delivering postal votes.The Commissionshares those concerns and believes that greaterconsideration should be paid to thearrangements for the delivery and return ofpostal votes and that procedures should be asformalised as possible.


36

Appendix 1 – Eligibility for postal votesfor elections in Northern Ireland

a) Registered service voters

b) Those who cannot reasonably be expected

i. to go in person to the polling station allotted

ii. to vote unaided there

by reason of blindness or other physicalincapacity

c) Those who cannot reasonably be expected togo in person to the polling station allotted byreason of the general nature of theiroccupation, service or employment or that oftheir spouse

d) Those who cannot go in person from theirqualifying address to the polling station allottedwithout making a journey by air or sea

A person applying to vote by post must providean address in the United Kingdom as theaddress to which his ballot paper is to be sent.

Source: Representation of the People Act 1985

Appendix 2 – Background to pilotschemes

After the 1997 General Election, the Secretary ofState for Home Affairs, Jack Straw, directed that areview of electoral procedures should beconducted, and that it should take into account thedecreasing participation in public elections.

In order to conduct this review, the Working Partyon Electoral Procedures was established, under theChairmanship of George Howarth MP, and met forthe first time in January 1998.The Final Report waspublished in October 1999 and included arecommendation that the Home Secretary bepermitted to amend electoral legislation toapprove pilot schemes to test alternative votingarrangements at a local authority level.

Many of the Working Party’s recommendations,including the recommendation on pilot schemes,were subsequently incorporated into theRepresentation of the People Bill. Predicting thesafe passage of the Bill and in accordance withits provisions, the Home Secretary invited localauthorities to make applications to conductpilots schemes.

These applications were submitted by the 11thJanuary 2000 and by 14th February theReturning Officers from each authority receiveda letter indicating whether or not the applicationhad been successful, pending the Bill’s passagethrough Parliament.

The Bill gained Royal Assent to become theRepresentation of the People Act in April 2000and the local authorities whose applicationshad been successful were able to proceed withtheir arrangements for the local election on 4th

May 2000.


37

Appendix 3 – Declaration of Identity + inner envelope

Front of form


38

DECLARATION OF IDENTITY REPRESENTATION OF THE PEOPLE ACTS

Ballot Paper No:

I hereby declare that I am the person to whom the ballot paper numbered as above was sent.

Voter’s signature:

The voter, who is personally known to me, has signed this declaration in my presence.

Witness’s signature

Name of witness (WRITE CLEARLY)

Address of witness (WRITE CLEARLY)

SEE INSTRUCTIONS ON THE BACK OF THIS FORM

Appendix 3 (continued)

Back of form

INSTRUCTIONS TO THE VOTER

1. You must sign this declaration of identity in the presence of a person known to you.That personshould then sign this declaration as a witness, adding his or her name and address.Without this thedeclaration will be invalid.

2. Vote for one candidate only. Put no other mark on the ballot paper or your vote may not becounted.

3. Mark a cross (X) in the box on the right hand side of the ballot paper opposite the name ofthe candidate you are voting for. Do this secretly. If you cannot without assistance, the personassisting you must not disclose how you have voted.

4. Put the ballot paper in the small envelope marked “A” and seal it.Then put the envelopemarked “A”, together with the declaration of identity, in the larger envelope marked “B”. Return itwithout delay.The ballot paper must be received by the returning officer not later than the closeof the poll. Alternatively, it may be delivered to a polling station in this constituency on polling day.

5. If you receive more than one ballot paper, remember that it is illegal to vote more than once(otherwise than as proxy) at the same election.

6. At this election you cannot vote in person at a polling station, even if you receive an officialpoll card.

7. If you inadvertently spoil your ballot paper, you can apply to the returning officer for anotherone.With your application you must return, in your own envelope, the spoilt ballot paper, thedeclaration of identity and the envelopes marked “A” and “B”. Remember that there is little timeavailable if a fresh postal ballot paper is to be issued and counted.

Regulation 66(b)

Source: Draft Statutory Instrument, Representation of the People (England and Wales) Regulations 2001


39

Appendix 3 (continued)

Inner Envelope


40

General Election

Ballot Paper Envelope

No(s): _______________

BALLOT PAPER(S) ONLY TO BE ENCLOSED IN THIS ENVELOPE

This envelope, together with the Declaration of Identity, must beplaced in the envelope addressed to the Returning Officer.

_______________________ Ward/Electoral Division

A

Appendix 4 – Election Turnout Figures of Wards in 1999 and 2000


41

Authority Ward 1999 2000 % change Average turnout across authority (2000)

Bolton Bromley Cross 31.6% 45.8% +14.2 26.8%

Farnworth 15.1% 26.7% +11.6

Smithills 29.1% 45.5% +16.4

Doncaster Conisbrough 24.5% 40.1% +15.6 26.2%

Gateshead Bensham 19% 46.3% +27.3 29.6%

Whickham North 30% 62.2% +32.2

Norwich Bowthorpe 18.3% 3 0.7% +12.4 29.3%

Catton Grove 21.1% 31.4% +10.3

Stevenage Bedwell 32% 38% +6 28.1%

Old Town 33% 49.2% +16.2

Swindon Abbey Meads N/A 33.4% N/A 28.6%

Moredon 19.6% 31.5% +11.9

Parks 17.7% 27.3% +9.6

Penhill 19.5% 30.9% +11.4

Wigan Bedford Astley 18.4% 26.8% +8.4 19.5%

Whelley 19.3% 25.9% +6.4

Winstanley 15.1% 25.7% +10.6

Sources:

Local Authority reports

Local Elections Handbook 2000, Colin Rallings and Michael Thrasher, Local Government ChronicleElections Centre, 2000

Appendix 5 – Elections in Oregon since 1996

Source: Oregon Secretary of State, Elections Division


42

Date Election Voting method Turnout

30th January, 1996 Special US Senate General Universal postal 66.32%

12th March 1996 Presidential Preference Primary Universal postal 57.62%

2nd April 1996 Special Congressional Primary Polling-station 34.2%

21st May 1996 Biennial Primary Polling-station 37.75%

5th November 1996 General Polling-station 71.31%

20th May 1997 Special Universal postal 42.11%

4th November 1997 Special Universal postal 57.62%

19th May 1998 Primary Polling-station 34.9%

3rd November 1998 General Polling-station 59.02%

2nd November 1999 Special Universal postal 38%

16th May 2000 Primary Universal postal 51.29%

7th November 2000 General Universal postal 79.8%

Unlike postal voting, there has been no long-standing use of electronic counting in publicelections in the United Kingdom, though it hasbegun to be piloted. As is now well-known,mechanised counting of one form or another hasbeen used for many years in the United States ofAmerica, and some parts of Norway have alsoused electronic counting for about 10 years.Theexperience of the voters in casting their votes isnot generally greatly affected by electroniccounting beyond relatively minor changes to theformat of the ballot paper. Rather than having aneffect on levels of participation therefore,electronic counting is intended to improve thespeed and accuracy with which the result of theelection is delivered.

Current vote counting arrangements inthe United Kingdom(rules for elections to the European Parliamentdiffer from the following in several respects)

In General Elections, counting usually takes placein a suitable centralised location within eachconstituency. Local Election counts also tend totake place in centralised locations, though thesecan sometimes be split into two or three sites. Inboth cases, votes are counted as soon aspracticable after close of poll, which usually meanson the evening of polling day. Ballot boxes aretaken from polling stations to the central countunder the direction of the Returning Officer(often with the assistance of the police), orotherwise are put into secure storage until thebeginning of the count.

Those entitled to attend the count include theReturning Officer, his/her staff, the candidates andtheir spouses, the election agents and specificallyappointed counting agents.The Returning Officeris able to admit others to the count as long as

s/he is satisfied that their presence will not in anyway impede the progress of the count orcompromise the secrecy of proceedings. Allpresent at the count are made aware of therequirements of secrecy at the count and thepenalties for infringing those requirements.

The first stage of the count is to open the ballotboxes and count and record the number ofballot papers in each.The next step is theverification of the ballot paper account assubmitted by the presiding officer of each pollingstation; the Returning Officer draws up astatement as to the result.This allows theReturning Officer to check that all votes casthave reached the central count, and that nounused ballot papers have been illegally added tothe votes cast.

The ballot box(es) containing the alreadyprocessed valid postal ballot papers are alsoadded to the count.

Before the ballot papers are counted in termsof the votes cast, ballot papers must be mixedwith those from at least one other ballot boxso that it is not possible to deduce the votingpreferences of any particular polling district.Postal votes must be mixed with ballot papersfrom at least one polling-station ballot box, andone ballot box must be kept back in order tomix with the final ballot box to reach thecentral count.

Ballot papers are then sorted into piles of votesfor each candidate, and are bundled into batchesof between 10 and 100. Doubtful ballot papersare set aside for adjudication by the ReturningOfficer in the presence of the candidates andtheir agents. Counting agents sit and watch thecounting process.

Elections in the 21st Century: from paper ballot to e-voting Electronic Counting

43

Electronic Counting

Procedures vary somewhat if there is more thanone vacancy to be filled.Votes cast in a block forcandidates of the same party can be sorted andcounted together relatively easily. For votes castacross party lines, as is often the case in parishcouncil elections with large numbers ofcandidates, more complicated procedures arerequired.This can involve counting clerks sittingin pairs, with one calling out the names of thecandidates for whom votes have been cast, andthe other making a stroke on a counting sheet inthe appropriate place. Another solution involvessticking 15 – 20 ballot papers alongside eachother on a large piece of card, overlapping sothat only the voted part of the paper shows; acolumn on the far right hand side of the cardenables tallying of votes cast for each candidateon all of the ballot papers.

During the counting process, ballot papers arekept face up, so as not to expose the serialnumber printed on the reverse which could giverise to an opportunity to identify the voter.

Ballot papers which are unquestionably invalid arethose:

p which do not bear the official mark

p which bear a mark by which the voter couldbe identified (other than the serial number)

p on which the voter has not placed any mark

p on which the voter has cast votes for morecandidates than s/he was entitled to do so

For other doubtful ballot papers, the criteria bywhich they are judged to be valid or invalid iswhether the voter’s intention is clear. Any ballotpapers deemed to be invalid are marked‘rejected’ by the Returning Officer. If any of thecounting agents disagree with the ReturningOfficer’s decision, the Returning Officer mustalso add ‘rejection objected to’ to the ballotpaper. An account is drawn up of the invalidballot papers and the reasons for which theywere rejected.

When the counting is complete, the ReturningOfficer informs the candidates and their agents of

the results. Candidates and their agents mayrequest a recount, particularly if the result isclose, and the Returning Officer will agree if s/heconsiders the request to be reasonable.

The final result is declared forthwith by theReturning Officer.

All ballot papers, counterfoils and other electionmaterials are sealed up in packets and kept insecure conditions for a year and then destroyed.

Electronic Counting Systems

Punchcard systemsPunchcard systems have been used in the UnitedStates since the 1960s.The ballot paper is a pieceof card, and instead of making a mark with a penor pencil on the ballot paper, the voter isrequired to make a hole in the ballot paper toindicate a preference.

There are two different kinds of punchcard, bothof which can have counterfoils: one is the‘votomatic’ and the other is the ‘datavote’ card.The votomatic card is most widely used in theUnited States, because it allows voters to expresspreferences on many different issues at the sametime, all on the same ballot card. A series ofnumbered boxes, arranged in columns, areprinted on the card (without any words).Thenames of candidates or the ‘referendum’ issuesare printed in a special, hinged booklet. Both theballot card and the hinged booklet are insertedinto a special frame. As the voter turns each pageof the booklet, only the printed boxes in thecolumn relevant to the election or referendumissue on that page are exposed.The places whereholes may be punched are pre-scored, and votersuse a stylus to punch holes in the relevant boxesto mark their preferences.

Datavote cards differ from the votomatic cardsin that they have the candidates’ names andreferendum issues printed onto them, with acorresponding box in which to punch a hole toshow a preference.The cards are not pre-scored, and special mechanical hole-punchingdevices are used, which cleanly remove the‘chad’ (the little piece of card to be removed)from each hole.


44

The cards are then fed through a computerisedcard reader which detects the location of theholes punched and records the totals.The readercan be programmed to reject ballots papers withtoo many votes, without a vote, or those whichare unreadable.

Counting can take place within the polling station,in which case the reader functions as the ballotbox and can warn a voter if they have cast aninvalid vote, giving the opportunity to recast avalid vote.The results would then becommunicated to the central tallying location.Otherwise, the ballot papers can be transportedto a centralised location for counting.

Optical scanning systemsThere are a number of different types of opticalscanning systems, for example Optical MarkReading (OMR), also sometimes known asMarksense, Optical Character Recognition (OCR)and Intelligent Character Recognition (ICR). Fromthe point of view of the voter, they alter thevoting process less than punchcard systems.Candidate names or referendum issues areprinted on the ballot paper and voting involvesmaking a mark next to the appropriate name(s)or option(s) with a pen or pencil.The mark mustbe in a very specific area on the ballot paper, andmay be a cross, or an area (for example, an ovalor a rectangle) to be filled in to mark apreference, or an arrow to be completed, or withpreferential voting, numbers in order ofpreference.

As with punchcards, the ballot papers are thenfed through a computerised scanner whichdetects the location of the darkest marks in thespecified areas on the paper by measuring thereflected light (OMR) or the location of differentnumbers on the paper (ICR) and records thetotals. Once again, the scanners can beprogrammed to reject ballot papers which bearno mark, which bear too many marks orunreadable marks.This can be done within thepolling station, in which case the scanner functionsas the ballot box and can warn voters if theyhave cast an invalid vote and giving theopportunity to recast a valid vote. Otherwise, thevotes are inserted into ballot boxes (usuallywithout folding and sometimes with a secrecy

folder) and transported to a centralised locationfor counting.

This kind of technology has been used in somemunicipalities in Norway since the early 1990s,and Bosnia and Herzegovina (1998 and 2000),Hong Kong (2000), some polling stations in Russia(1995 & 1996) and Venezuela (since 1999).

Data-input systemsSome vote counting systems use standardpersonal computers and require the data fromthe ballot papers to be entered either manually,or by means of a bar-code reader attachment.These systems do not greatly change theexperience of the voter in that they arepresented with a ballot paper with a list ofcandidates, beside which they make their mark,with a pen or pencil, to express their preference.The only difference with a bar-code system wouldbe the presence of a bar-code next to the nameof each candidate on the ballot paper. At thecount, election staff swipe the bar-code next tothe name of the candidate chosen by the voter,and the vote is tallied by the computer software.

In the case of a voting system which involves acomplicated count, as is the case with someproportional systems, it is possible to count theballots by hand but to enter the totals into acomputer programme to calculate the final result.

Examples of countries which use computerisedcounting in this way are Australia, Bulgaria,Hungary, Lithuania, Russia, South Africa andUkraine.

Information from Pilot Projects

As part of the local election pilot schemes, twolocal authorities experimented with electroniccounting on 4th May 2000.These wereBroxbourne Borough Council and Three RiversDistrict Council. Elections for the London Mayorand Assembly, held on the same day, also pilotedthe use of electronic counting.

Broxbourne (Data-input)The electronic counting pilot in BroxbourneBorough Council took place in four of the twelvewards in which elections were taking place, with


45

one councillor to be elected in each ward.Thefour wards selected (namely Wormley/Turnford,Cheshunt Central, Cheshunt North andHoddesdon North) were those with the largestelectorates.

The system used was a data-input type systemsimilar to that developed by Epping ForestDistrict Council in 1990 and used in parishcouncil and council shire authority elections there.A demonstration of the system was given priorto election day for candidates and their agents inthe relevant wards.

Barcodes were printed onto the ballot papers,with a specific barcode for each candidate,printed to the left of his/her name.Voters markedtheir ballot papers in the usual way, by placing across in a box to the right of the name of thefavoured candidate. Ballot papers were thenfolded and ‘posted’ into sealed ballot boxes in thetraditional manner.There was no evidence tosuggest that the addition of barcodes to theballot paper had in any way confused the votersor caused any ballot papers to be spoilt.

Existing local authority equipment was used forthe count, and consisted of one computer andone barcode reader per ward. BroxbourneBorough Council also designed the software thatwas used.

4,875 votes were cast across the four wards.Within each ward, ballot papers were initiallysorted into piles according to the candidate forwhom the vote had been cast. One member ofstaff per ward then used the barcode reader to‘swipe’ the barcode next to the candidate’s nameselected by the voter. As each ballot paper wasswiped, the name of the party for whom the votehad been cast appeared on a screen, visible tothe counting staff, and other observers.Theresults were cumulatively stored in thecomputer’s memory, and when all ballot papershad been swiped, they were aggregated. As istraditionally the case, candidates and their agentswere entitled to be present in order to observethe counting process.

The count was completed in a similar time-frameto those wards using traditional counting

methods, though staffing levels were lower (4 perpilot ward as opposed to 6 – 8 per ‘traditional’ward). No complaints were received during orafter the count from any of the candidates ortheir agents and no recounts were requested.Though no specific testing was carried outimmediately prior to the count, BroxbourneBorough Council believes that both the hardwareand the software proved reliable.

The principal increase in expenditure was thecost of printing barcodes on the ballot paperswhich raised printing costs by £1,300. Since theequipment used was existing local authorityequipment, no additional expenses were incurredin this area. Staff time had been dedicated to thedevelopment of the software, but other than thetime needed for minor modification in the future,this was a one-off expense. As mentioned, fewerstaff were used in the pilot wards than in wardsusing traditional counting methods, thus creatingsome savings.

In its evaluation, Broxbourne Borough Councilidentified an increase in the number of barcodereaders per ward, from one to two, and aconsequential adjustment to the software, as away of delivering the result more quickly as wellas making extra savings, since it would notnecessarily mean increasing staffing levels.

Three Rivers (Optical scanning)The pilot scheme in Three Rivers District Counciltook place across all 16 wards in which electionstook place, to elect 17 councillors (this includedone by-election).

Electronic counting was conducted in these wardsby means of two optical scanning machines(optical mark readers) and a personal computer,provided by the company Election Systems andSoftware (ES&S).The particular model ofcounting machine used1 is able to process up to350 ballot papers per minute.

Voters came to the polling station to cast theirballots in the usual manner.The ballot paperformat was different from a traditional ballotpaper in that it was slightly larger, and pre-printedwith an official mark, rather than needing to bestamped upon issue. More noticeably, instead of a


1 Election Systems & Software’sModel 550 NetworkableCentral/Regional Super High SpeedBallot Counter and Vote Tabulator

46

blank box beside the name of each candidate,there was a small box within a larger box with across running through both.Voters were requiredto mark their choice of candidate, with a pencil(this was crucial since the scanning machineswere not able to detect ink) by drawing over thecross in the appropriate box.The ballot paperthen had to be posted into the ballot box,preferably without being folded and preferablyface-down to maintain privacy. Although themachines can cope with ballot papers that havebeen folded, processing of ballots that do notneed unfolding, and which are perfectly flat isquicker and more efficient. Ballot papers werealso encoded with information, readable by thecounting machines, about the ward and, whereappropriate, the polling station to which theybelong, so that they can be counted appropriatelyeven if separate batches of ballot papers are fedthrough simultaneously at the count.

The count for all wards took place at a singlevenue, with about 20,000 ballots papers to becounted. Prior to the commencement of thecount, ES&S staff conducted a logic and accuracytest on the counting machines to show that theywere counting correctly, and to prove that novotes were already stored in the machines beforethe count had begun.These tests took place inthe presence of the Returning Officer who signeda declaration stating his satisfaction with themachinery and the software at the beginning ofthe count.

The machines count the total number of ballotpapers (for verification), and the numbers ofvotes cast for individual candidates at the sametime.The postal votes were the first to be fedthrough the machines. Upon arrival of ballotboxes from the polling stations, these ballotpapers were fed through the counting machines.Folded ballot papers did not cause undueproblems or delays, beyond having to unfold andsmooth the papers before feeding them into themachines.

The initial figures given were the total number ofballot papers per polling station, which werecompared with the Presiding Officers’ ballotpaper accounts which, for ease of reference, hadbeen entered into the computer. In almost every

ward, the figures tallied immediately and theReturning Officer was satisfied that the processand been more accurate than when conductedby hand.

Before the totals for each candidate could becalculated, a number of issues had to be dealtwith.These involved ballot papers that had beenrejected as unreadable by the machines because:

p they had been completed in pen rather thanpencil (many though not all of these werepostal votes)

p voters had changed their minds or placed theircross in the wrong place

p they were spoilt (i.e. had too many votes, ornone at all)

The Returning Officer was required to adjudicateon all of these ballot papers, in the presence ofthe candidates and their agents. Due to thenumber of papers rejected by the machines thiswas a lengthy process. After adjudication, theseballot papers had to be fed back through thecounting machines, either as valid votes for aparticular candidate or as spoilt ballots, in orderfor the number of papers to tally.

The counting process eventually took longer thananticipated, and longer than a manual countwould have taken. However, it is important tonote that in the ward which also had a by-election where voters were required to vote fortwo candidates, the Returning Officer felt that incomparison with a hand-count, the machines“…dealt extremely quickly and efficiently with(the) count.”

A considerable number of the problemsencountered, including misplaced andunadjudicated ballot papers, were due toinsufficient or ill-judged logistical arrangements.Some of these arrangements, such as the decisionnot to proceed on a ward-by-ward basis, weretaken in an attempt to speed up the process, butended up causing confusion and mistakes.TheReturning Officer’s report2 and ES&S’s ownevaluation3 clearly identify improvements in theseprocedures, including the need for a rehearsal,


2 Alastair Robertson (ReturningOfficer), Three Rivers CouncilElection, 4th May 2000: Pilot use ofAutomatic Counting Machines, May2000

3 Evidence given to theCommission by Angus Ward,Director of Sales Europe, ES&S,18th July 2000

47

and we will confine our comments to issuesraised specifically by the technology.

Confusion was reported amongst some voters asa result of the format of the ballot papers,particularly the fact that there appeared to becrosses in all of the boxes already. Clearly, this is adesign problem, which the Returning Officersuggests could be easily rectified by using muchfainter crosses, possibly using dotted lines.

The large number of ballot papers rejected by themachines due to being completed in pen ratherthan pencil caused considerable delay, particularlysince these needed to be remarked (with blackdots) before being fed back into the machine.

Similarly, where voters had changed their mindsor put their crosses in the wrong place, theReturning Officer had to judge whether theintention was clear, and if so, either blot out theirrelevant mark (with white dots), or remark theballot paper (with black dots) before feeding theballot papers back into the machine.

Fifteen people were employed at the count,including ES&S employees who were operatingthe counting machines. It was felt by the ReturningOfficer and his staff that the number of electionstaff working alongside ES&S staff was insufficient,and that ideally there would need to be two orthree per machine in order to monitor and assistwith the process. Candidates were entitled to onecounting agent per machine, however the numberof ballot papers on which adjudication wasnecessary left them short-handed in terms ofscrutinising the rest of the process.

The electorate involved in this pilot scheme wasabout 52,000.The total additional cost was£10,245 which included the increased printingcosts of the ballot papers, and employing theservices of ES&S.The optical scanning machineswere provided free of charge on a trial basis.Savings were made by a reduction in staff at thecount, from 50 to 15.

London (Optical scanning)Elections took place on 4th May 2000, the sameday as the local authority pilot schemes, to elect,for the first time a Mayor and Assembly for

London.The electorate was about 5.1 million.

Two different voting systems were used, theAdditional Member System (AMS) to electAssembly members, and the Supplementary Vote(SV) to elect the Mayor. AMS means that everyvoter has two ‘X’ votes, one to elect aconstituency Assembly member and the other toelect an ‘additional’ member on a London-widebasis. SV allowed voters a first and a secondpreference in electing the Mayor (also two ‘X’s). Itwas, then, a large-scale and relatively complexelection, and for reasons of speed and accuracy itwas decided to explore the use of technology.After a tendering process, it was decided thatelectronic counting in constituency centres bestmet the Government Office for London’s criteriaand it was decided that the structure andprocedures surrounding the count should stick asclosely as possible to the traditional methods,simply introducing machines to do the countingand tabulation. It was hoped that results would bedelivered by breakfast-time the following morning,the 5th May.

Data & Research Services (DRS) was thecompany that won the contract.They provided168 optical-scanning machines (optical markreaders) which can count votes, in any orientation,at a rate of up to 2 votes per second4.Themachines use infra-red light to scan the papersand are therefore capable of reading pencil orpen, as long as the ink is not red.Tests wereconducted by two independent consultants, oneemployed by DRS to ensure the reliability of theperformance of the hardware and software, andthe other by the Greater London ReturningOfficer (GLRO) to assess the rigour of the testing.

Training was conducted by the Association ofElectoral Administrators and by DRS for electionstaff and officials in the months prior to theelection. A large-scale rehearsal also took place inHammersmith at the end of February using anotional turnout of 60% in the West Centralconstituency as its basis. As a result, 406,000ballot papers containing up to 1.6 million voteswere processed. 390,000 of these ballot papershad votes laser-printed onto them, and theremaining 16,000 were completed manually byelection staff. Representatives of the London


4 The machines used were Data &Research Services’ CD800 OpticalMark Readers

48

boroughs, the political parties, Government andthe press were invited to attend.The test wentsmoothly, both technically and organisationally, andany issues raised were subsequently addressedbefore election day5.

The machinery and the specific set-up to be usedin each constituency was tested in a warehousetwo days before the election.Tests of the set-upwere also run in the constituency countinglocations during election day.

The format of the ballot papers was largelydictated by the voting systems being used, withone ballot paper for the Assembly, and a separateone for the Mayor. Features that were a directresult of the counting system were:

p barcode serial numbers on each ballot paper, toprevent papers from being counted more thanonce; the barcodes were unique to each ballotpaper (as with serial numbers on traditionalballot papers), but also included an identifier forthe particular contest (i.e. Mayor or Assembly),and an identifier for the Borough; it was the firsttime that barcodes had been used on ballotpapers in this way in a public election

p a faint cross in each box corresponding to thecandidate or political party, to encouragevoters to place their mark in the correct placeto be scanned by the machines

In the usual manner, voters attended their localpolling stations to cast their votes.Voters indicatedtheir voting intention by marking an ‘X’ in theappropriate boxes beside the names of candidatesor parties.Voters were asked to mark the ballotpapers with the supplied pencils and also to puttheir ballot papers into the ballot box unfolded,with the blank side up in order to retain privacy.

On election night, counts were conducted byConstituency Returning Officers (CROs) in acentral location within each of the 14constituencies, with a Calculation and Declarationcentre in Westminster under the control of theGreater London Returning Officer (GLRO). Bothlocal authority election staff and DRS employeeswere present at each count.

The machines were programmed to accept and, asappropriate, count all ballots that had beencorrectly completed, which were blank, or hadblank sections.They were programmed to reject allballot papers where the voter’s intention was, forwhatever reason, unclear, so that the ConstituencyReturning Officer could adjudicate on their validity.(See Appendix 1, p. 61 for detailed breakdown ofaccepted/rejected ballot papers). Each constituencycount had 11 – 14 scanners, 4 – 6 computerworkstations and a server.

When ballot boxes arrived at the constituencycounting centre, the ballot account sheets, ascompleted by the Presiding Officer in eachindividual polling station, were checked and thetotals transcribed onto a scannable checksheet.Any folded or creased ballot papers wereflattened and all ballot papers were stacked, alongwith the scannable checksheet, ready for insertioninto the counting machine feeder trays.This wasthe end of the ‘reception’ process.

The next stage was ‘registration’.The checksheetwith the ballot account totals was scanned andthe data entered into one of the computerworkstations.The checksheet was then placed atthe bottom of the pile of ballot papers.

The ‘reading’ stage of the process consisted ofballot papers and the relevant checksheet beingplaced on the feeder trays of the machines, andfed through for scanning. Ballot papers emergedfrom the machines sorted into two trays: ‘Good’and ‘Rejected’.The ballot papers in the ‘good’ traywere those which were correctly cast or blank;those in the ‘rejected’ tray were those whichneeded to be adjudicated upon because theintention of the voter was unclear. If the machinecould not read a barcode on a particular paper,then the machine would stop in order for theunreadable paper to be removed and put into afolder marked ‘Unidentified’.The raw data fromthis reading was gathered and sent to the server.

Next followed the ‘verification’ stage in whichballot papers in the unidentified folder weremanually entered with a handheld barcodescanner, to ensure that they were genuine ballotpapers.The total number of ballot papers read ineach contest was then displayed and transcribed


5 The First Election of the Mayorand Assembly for London: Reportfor the GLA, Office of the GreaterLondon Returning Officer, July 2000

49

onto the checksheet to compare with the ballotpaper accounts. Rejected ballot papers thenproceeded for adjudication.

The Constituency Returning Officer adjudicatedon ballot papers which the machines hadrejected. Once the status of each vote had beendetermined, the barcodes were manually scannedand the voter’s intention entered directly into thecomputer.

Once all ballot papers from the constituency hadbeen dealt with, the ‘consolidation’ process began.The Returning Officer printed four separatereports from the computer (the constituencyAssembly results, the votes cast for additionalAssembly members, the first and second choicescast for Mayor, and the distribution of secondchoice votes in relation to first choice votes).TheReturning Officer showed the first three of thesereports to the candidates and their agents, andon gaining their agreement that the count wasaccepted, confirmed the result.

Results were then transmitted electronically bymodem over secure lines to the Greater LondonReturning Officer (GLRO), and a hard copy wasalso faxed.The constituency results wereconfirmed, and the other totals were entered intothe computer for consolidation with those fromother constituencies.

Once results had been received from all 14constituencies, the computer finalised theconsolidation of constituency results and madethe necessary London-wide calculations for theMayor and the additional Assembly members.TheGLRO then informed the candidates for Mayorand their agents of the result and sought theiracceptance. Once accepted by all parties, theresult was publicly declared.

The same process took place for the London-wide ‘additional member’ candidates, and theresults were duly declared.

A number of problems occurred during thisprocess which, despite a turnout of just 33.7%,cumulatively caused the final result to be declarednot at breakfast-time but closer to midday onFriday, 5th May. Some of these problems were of

a logistical nature; these are set out in the GLRO’sreport.6 Although these are clearly importantfactors, our examination will focus on the moretechnical issues raised specifically by the use ofelectronic counting.

The counting machines operated considerablymore slowly than anticipated (50% slower insome areas) and considerably more slowly thanat the rehearsal that had taken place in February.

One problem was the unexpectedly high numberof ballot papers that were rejected by themachines for adjudication.These were not onlyballot papers where the voter’s intention wasunclear, but also postal votes (15% of postal votesin Barnet & Camden had to be hand-scanned7),ballot papers with blemishes on them, ballotpapers where the cross extended just beyond theconfines of the box, and some ballot paperswhere the machine was unable to read thebarcode. Although the scanners had beendesigned to read the barcodes on the ballotpapers in any orientation, a late change in thesoftware for the Mayoral election meant that thescanners were not able to read barcodes onballot papers inserted in one particularorientation.These factors meant that the CROsspent significant amounts of time dealing withindividual ballot papers, which then had to beentered manually, all contributing to a slowingdown of the process.

Another problem was double-feeding of ballotpapers, where ballot papers were sticking togetherand passing through simultaneously; this happenedfar more frequently than had been anticipated.Themachines were equipped to detect this, howevereach time it happened, the machine stopped, thepapers had to be retrieved and the machinerestarted, therefore slowing down the process.Ballot papers which had not been cleanly detachedfrom their counterfoils also caused the machines tostop.The CRO for Bexley and Bromley estimatesthat the scanners stopped between 10 – 15 timesfor each polling station batch8.

In several constituencies, many of the countingmachines ceased to function during the count (allbut two of the twelve machines in the Enfield andHaringey constituency).This turned out to be due


6 The First Election of the Mayorand Assembly for London: Reportfor the GLA, Office of the GreaterLondon Returning Officer, July 2000

7 Evidence given by Max Caller,CRO for Barnet & CamdenConstituency, to GLA ElectionsInvestigative Committee, July 2001

8 Evidence given by Walter Million,CRO for Bexley & BromleyConstituency to GLA ElectionsInvestigative Committee,September 2001

50

to fluff from the baize table-cloths interfering withthe machines’ double ballot detectors. Althoughthe same baize table-cloths were used at therehearsal, people had, on that occasion, been onhand to regularly clean the detectors. It was alsosuspected that static resulting from the use ofplastic ballot boxes where voters had to posttheir ballot papers down a plastic chute, hadexacerbated the problem both of ballot paperssticking to each other and of fluff sticking to theballot papers.These ballot boxes had not beenready for use during the rehearsal. Anotherpossibility was that the tinting process that madethe Assembly ballot papers and the Mayoral ballotpapers different colours interfered with theirpassage through the scanners9.

Mistaken attempts to feed ballot papers whichhad already been counted into the machinesdelayed proceedings whilst machines had to bestopped in order to remove the ballot papers.Themachines did, however, recognise the papers ashaving already been read. Other CROs havereported serious discrepancies between ballotpaper accounts and the number of papers verifiedby the machines. In some cases, ballot paperswere rescanned and provided a different total.

There were also anecdotal reports of datainput/software problems which were eventuallyresolved by bypassing the official software –removing the data relating to a set of alreadyscanned-in ballot papers and starting again –thereby arousing serious security concerns.

The budget for the election, excluding theinformation campaign, was £12.5 million.Thiscovered the fees and expenses of the ReturningOfficers, the cost of polling stations, count venuesand staff, as well as the electronic countingcontract, which included the cost of printing ballotpapers and providing ballot boxes and countingequipment.


Florida (Punchcards)Presidential and Legislative elections took placeacross the United States of America on 7thNovember 2000. As has been well-documented,the result of the Presidential election was

extremely close, and due to a combination ofmany factors, the process of deciding who hadwon the Presidency was slow and painful. Althougha result was expected the following day, thecontest was not resolved until mid-December, afterprotracted and contentious legal proceedings.

The closeness of the result served to highlightproblems with many aspects of electoraladministration in the US, institutionally, technicallyand logistically. Our focus will be on the technicallessons that emerged from the election in relationto the use of punchcards, and our discussion willrestrict itself to issues of relevance to elections inthe United Kingdom.

The State of Florida has 8.4 million registeredvoters across 67 counties. Decisions about howto conduct elections are taken at county level,though all equipment must be certified by theState-level Division of Elections. Certificationincludes voluntary standards of accuracy levels, aswell as testing procedures. Four differentmethods of voting or vote-counting were used10,including punchcards in 24 of Florida’s morepopulous counties.

Of the 24 counties using punchcards, 15 usedVotomatic systems and 9 used Datavote, and 7different models of punchcard readers were used.In all cases, ballot papers were transported to acentral location for counting. Across the UnitedStates,Votomatic systems are far more popularthan Datavote systems (17.5% of counties usedVotomatic systems with only 1.7% usingDatavote).This is due to the fact that a singleVotomatic card allows voting locations for a largenumber of elections and issues, whilst Datavotewould require the use of several cards. In manyareas, voters were asked to vote in twentydifferent elections and on twenty referendum-type issues.

Since electoral arrangements and equipmentvaried from county to county, we will not providea detailed description of the conduct of theelection, but will examine the main issues thatarose across the 24 counties.

Although the concept of data processing cardshad existed for many years, it was not applied to


9 Evidence given by Simon Day,Commercial Director, DRS, to GLAElections Investigative Committee,July 2001

10 Punchcards in 24 counties;Optical scanning in 41 counties;Mechanical lever machines in 1county and paper ballots withmanual counting in 1 county.

51

the electoral process until 1964 when it was usedin elections in 5 counties across the UnitedStates11. Its use spread rapidly, to the extent thatten years later it was being used by 10% of the USelectorate, and in the November 2000 elections, itwas used by about a third of the electorate. Oneof the reasons for the rapid spread of thistechnology was that the cards were in a standardformat which was readable by equipmentautomatically provided with business computers atthe time, making it a cost-effective option.12 In mostcases in Florida, punchcard equipment was boughtin the 1980s to replace mechanical lever machines(see chapter on Electronic Machine Voting).

The turnout in Florida was 68.1% and the initialresult was a lead to George Bush of 327 votes outof 5.9 million votes cast (a difference between thetwo leading candidates of less than 0.01% of votescast).The number of ballot papers which were notcounted due to voters casting a vote for morethan one candidate (an ‘overvote’), for nocandidate at all (an ‘undervote’) or for otherreasons that made the ballots uninterpretable, wasjust under 180,000, or 2.93% of the votes cast –550 times more than the difference between thetwo leading candidates.

Such high percentages of overvotes andundervotes are worrying. Although some voterswould take the trouble of going to the votingstation and deliberately spoiling their ballot paper,or voting for other offices but not for President, itseems unlikely that as many as 3% would do so. Ithas been estimated that of the 2% of uncountedballot papers across the United States, that only0.5% made a deliberate choice not to vote forPresident, and that the remaining 1.5% (1.5 – 2million voters) believed that they had cast a validvote but had been inadvertently frustrated by theballot paper layout, or by the voting or vote-counting equipment13. Statistics from across theUnited States reveal that counties using punchcardsystems for the 2000 Presidential election had ahigher average rate of uncounted ballot papers(2.5%) than the average for counties using anyother form of voting or vote-counting equipment.In Florida the rate of uncounted ballot papers withpunchcard systems was 3.93%, as opposed to the2.93% State average.Three counties haduncounted ballot rates of over 9%14.

A problem from which both Votomatic andDatavote systems suffer is that the ballot cardsneed to be inserted into a frame before a votercan cast a vote; this is exacerbated in the case ofVotomatic systems, in that a booklet showing thechoice of candidate is also inserted and alsoneeds to be properly aligned. If the ballot paper isnot inserted absolutely correctly, then the votercould end up punching holes in places other thanthose in which they intended, and thereby votingfor another candidate, or not casting a valid voteat all. At least with a Datavote system it is mucheasier for a voter to see if they have made amistake, as long as the ballot layout is clear,because the names of the candidates are printedon the ballot paper.This could account for someof the uncounted ballot papers.

Even without the claims of technical andadministrative irregularities, the results in Floridawere close enough to make recounting alikelihood. Florida Statutes require that a recounttake place if the winner’s margin is 0.5% or less,unless the losing party specifically requestsotherwise in writing. In the case of unclear ballots,voting officials are required to attempt to judgethe voter’s intention. However there are no State-wide standards for interpreting voter intent inrelation to the range of possibilities that canoccur with punchcard systems.

Four counties came under scrutiny initially withrequests from the Democrats for recounts. Ofthese four, three (Broward, Miami-Dade and PalmBeach) used Votomatic punchcard systems andhad percentages of uncounted ballots of 2.6%,4.6% and 7.0% respectively.The calls for recountssubsequently spread to other counties.

Legal wrangling continued for almost another fiveweeks over whether or not recounting should bepermitted, and whether it should be conducted bymachine or by hand. During this period, variousrecounts were begun, if not all completed, and theyrevealed further problems with the punchcardsystem, particularly the Votomatic versions.

Electronic recounts are fairly quick and easy.However, in this case, they drew attention to twomajor problems, both involving the infamous‘chads’.Votomatic punchcards are standard-sized


11 Fulton and De Kalb counties inGeorgia; Lane county in Oregon;San Joaquin and Monterey countiesin California

12 Roy G. Saltman, ‘Voting Systems’,The Bell,Vol.1, No.1, May 2000

13 Voting:What is and what couldbe? California Institute ofTechnology and MassachusettsInstitute of Technology VotingTechnology Project, July 2001

14 Duval county (Votomatic),9.23%; Glades county (Datavote),9.59% and Jefferson county(Datavote), 9.19%

52

pieces of card on which are printed a series ofboxes and numbers. Each box represents apotential voting location and is pre-scored toassist the voter in punching a standard sized andshaped hole.The pre-scored areas are the chads.

However, for some time now, voices have beenraised against the use of pre-scored punchcards15,on the basis that they actually make it moredifficult for the voter’s intention to be correctlyregistered and can impede the punching of aclean hole, resulting in dimpled chads, hangingchads, trapdoor chads etc. High-profile problemswith punchcard systems in the election of 1968caused IBM to withdraw from the market.Wisconsin and Massachusetts abandonedpunchcard technology and banned the use ofpre-scored punchcards after problems in 1993and 1996 respectively.

Firstly, there is a danger that improperly punchedballot papers are incorrectly read by theelectronic counting machine. For example, ahanging chad which is loose, but still attached tothe ballot paper can be squeezed back over thehole whilst being stacked to go through thecounting machine, and therefore be read as ablank ballot, an undervote.

Secondly, when ballot papers are fed through themachines, particularly if this is done several times,it may be that some chads get knocked out ofthe ballot papers, thereby creating an overvote.

In either case, votes get spoilt or mistakenlyrecorded as being spoilt, and for this reason, theresults in Florida were different each time theballots were run through the counting machines.

Datavote cards are not pre-scored and thereforedo not tend to suffer from these problems.

In terms of manual recounting,Votomatic ballotpapers are extremely difficult to count becausethe holes are so small, there are so many differentvoting locations on each ballot paper, and no pre-printed names of candidates.There is a significantmargin of interpretation as to what constitutes avalid ballot and what the voter’s intention mighthave been.The same problems apply to hand-counting as to electronic counting in terms of

knocking chads out accidentally with excessivehandling. Once again, these are problems relatingto the layout and the pre-scoring of theVotomatic ballots and do not apply to theDatavote ballots.

Over-ruling a decision made by the FloridaSupreme Court, the US Supreme Court ruled on12th December that there were to be no morerecounts of any kind.This meant that GeorgeBush won the electoral votes of Florida andthereby won the Electoral College vote for thePresidency of the United States.

Oslo (Optical scanning)Public elections take place every two years inNorway with local elections alternating withnational ones. Election day is a Monday. Eachmunicipality decides which mechanisms it will useto count its votes, though most count manually.Data-input type systems using existing softwaredo not need approval, however, optical scanningsystems require approval from The Ministry ofLocal Government and Regional Development.

The Municipality of Oslo has used optical-scanning since about 1980, both in local andnational elections.The machines are provided byResearch Election and Expert Services (R-2E) andEphorma, based in Norway along with Data &Research Services (DRS)16 based in the UnitedKingdom. Optical-scanning has been used on atrial basis in other Norwegian cities such asTrondheim and Kristiansand, but never regularly asin Oslo.

Norway uses a party list system, where eachpolitical party provides a list of candidates in aspecific order.The voting experience in areasusing electronic counting is exactly the same as inareas where votes are counted manually.Thevoter attends the polling station and registers inthe usual way.Voters are given a range of ballotpapers, each of which represents one party list;the voter chooses the ballot paper of the partyfor whom s/he wishes to cast a vote and, withoutmaking any marks on the ballot paper, places it inan envelope and puts it into the ballot box.Thismeans that a voter has cast a vote for a particularparty, and that s/he has accepted the order inwhich the candidates appear on that party’s list.


15 Accuracy, Integrity and Securityin Computerized Vote-Tallying, RoyG. Saltman, Institute for ComputerSciences and Technology, NationalBureau of Standards, August 1988

16 The company that provided theoptical scanning machines for theelections for London Mayor andAssembly in May 2000. In this case,the machines used are the CD850model.

53

However, voters also have the option ofamending the ballot paper if they wish to vote fora party but do not agree with the order in whichcandidates appear on the list. Striking through thenames of candidates has the effect of demotingthose candidates within the party list. In localelections, voters have further options foramending the ballot paper, which are to put across beside the names of candidates in order topromote them on the list, or to write in thenames of additional candidates selected fromother party lists. Once any marks have beenmade, the ballot paper is inserted into anenvelope and put in the ballot box.

At close of poll at 9pm on Monday evening, apreliminary manual count takes place in eachpolling station.These preliminary results, as well asthe results of early voting17, are given to themedia.The ballot boxes are then sent from thepolling stations to the Town Hall for a centralisedcount which begins as soon as ballot boxes arriveat about 11pm.The final ballot boxes usuallyarrive at the Town Hall just after midnight.

Ballot papers are fed through the scanners whichscan both sides of the ballot paper at a potentialspeed of up to 9000 per machine per hour.Barcodes make each ballot paper specific to aparticular election in a particular municipality, andalso reveal the party to which that ballot paperbelongs. Each barcode is unique, so that themachines can detect ballot papers that havealready been scanned.

The scanners also detect any other marks such ascrosses, deletions or ‘write-ins’. Crosses anddeletions are read and counted by the machines;‘write-ins’ are detected and usually separated inorder for the details to be transcribed by hand.Where newer Intelligent Mark Readingtechnology is in place, the scanner can usuallyinterpret the handwriting. Results are stored onthe scanners’ hard disk and are then uploadedonto a server for aggregation.

The scanning process continues throughout thenight and is usually finished by the following(Tuesday) morning. Although the overall result isfairly clear by this stage, final results are notdeclared until a 24-hour grace period for the late

arrival of postal votes has elapsed. Final results aredeclared on Wednesday afternoon.

The machines are hired for each election.Theelectronic counting operation costs Osloapproximately 2 million Norwegian Krone(about £160,000).

The last national election was in 1997 and Osloused 22 machines to count 300,000 votes(turnout of 79.2%).The use of IntelligentCharacter Recognition (ICR) for reading the‘write-in’ section of the ballot papers will beextended for elections in 2003.

Australia (Data-inputting)On Saturday, 3rd October 1998, Federal Electionswere held across Australia to elect the House ofRepresentatives and half of the Senate. It was thefirst election in which the Australian ElectoralCommission (AEC) was entitled to usecomputers to assist with the counting of theSenate election18.

In the election to the House of Representatives,one Representative was being elected from eachof the 148 divisions in the country.The systemused to elect the Representatives is theAlternative Vote, where voters number thecandidates in order of preference and a candidateneeds 50% of the vote to win.

In the Senate election, six Senators were beingelected from each of the six States, and two fromboth of the Territories – 40 Senators in total.Theelectoral system used is also a preferential system.Voters have the option of voting ‘above the line’,which means that they cast a single preference(‘1’) for a group ticket, usually a political party(where the group/party fixes the subsequentorder of preferences); or they can vote ‘below theline’ in which case they number all of thecandidates in order of preference. Candidatesneed to win a certain proportion of the vote,known as a ‘quota’ to win.

Because Representatives are elected from single-member constituencies, it is possible to gain afairly accurate picture of the final results onelection night. However, since Senators areelected in multi-member constituencies, the


17 Early voting comprises about25% of the total votes cast.

18 This was due to amendmentsmade to the ‘CommonwealthElectoral Act 1918’ by the ‘Electoraland Referendum Amendment Act1998’ which received Royal Assenton 17th July 1998

54

results from the entire State or Territory have tobe received (including postal votes, the deadlinefor which is 13 days after Election Day) beforethe quota can be calculated and furthercalculations can be completed, meaning that onlya general impression of the final results can begained. ‘Above the line’ votes for groups arecounted on election night, as are ‘below the line’first preference votes for Independent candidates.

Under the circumstances, as the Joint StandingCommittee on Electoral Matters has pointed out,“The delivery of federal election results is ofnecessity a complex and time-consuming process”19.

Counting takes place in two stages, on electionnight to give an initial indication of the results andafter election night to examine all votes and allpreferences in detail, and to complete allnecessary calculations.

In each State and Territory there is an AustralianElectoral Officer (AEO) who is the returningofficer for the Senate election in that jurisdiction.There is a Divisional Returning Officer (DRO) ineach of the 148 divisions responsible for theelection to the House of Representatives inhis/her jurisdiction.

The AEC used a national computerised systembased in each of the Divisional Offices.Thetelecommunications network was provided forthe AEC by Computer Sciences Corporation andunderwent significant testing prior to the election.It is a system which provides an audit trail byproducing reports for inspection by election staffand scrutineers at every stage. It had also beencertified by the Australian National Audit Office.

The system linked each Divisional Office to aNational Tally Room in Canberra which provideda central point for the progressive display ofelection results throughout the evening. About700 representatives from the media, 100 politicalparty activists, 30 overseas observers and 2,500members of the public visited the National TallyRoom during the evening. Results were also feddirectly to a computer terminal for the PrimeMinister in Sydney, to another for the leader ofthe opposition in Perth as well as to the majortelevision networks and the Australian Associated

Press.Within minutes of reaching the NationalTally Room, the results were also displayed on theAEC’s website.

The AEC had put in place various back-up plansin case of technical difficulties.Thecommunications links were duplicated throughalternative networks and exchanges, in case ofnetwork failure.There was also a duplicate powersource in case of electricity problems. A systemusing telephones and faxes had also been put inplace in case of problems with the computers. Allof these systems were tested at a rehearsal twodays before the election.

Political parties and others involved in theelectoral process had been fully briefed by theAEC about the introduction of the computerisedsystem during the year leading up to the election.

Voting was compulsory and the resultant turnoutof 95% meant that 11.5 million voters cast a votein two separate contests.

Counting began immediately on close of poll at6pm on Election Day, within each polling station.Votes for the House of Representatives tookprecedence.The main tasks on election night were:

p to count the total numbers of first preferencescast for each candidate in the House ofRepresentatives election

p to conduct a ‘two-candidate preferred’ count ofthe House of Representatives ballot papers, inorder to provide an indication of the likelyresult in each Division on election night

p to count the first preferences in the Senateelection

Candidates were able to appoint scrutineers whoare entitled to be present at every stage of thecount. Scrutineers are also provided with reportson the progress of the computerised counting.

Upon completion in the polling station of each ofthe three tasks listed above, the totals werecommunicated to the Divisional Office bytelephone.The DRO entered the figures for eachpolling station in the Division into the Australian


19 1998 Federal Election: Reportof the inquiry into the 1998Federal Election and MattersRelated Thereto, Joint StandingCommittee on Electoral Matters,Parliament of Australia, June 2000

55

Electoral Commission’s computerised ElectionNight System.The figures were then transmittedelectronically to the National Tally Room inCanberra where they were displayed oncomputer terminals and a National Tally Board.

Election results were sent to the National TallyRoom from 6.30pm until midnight and wereprogressively updated and displayed throughout thistime. No difficulties were encountered on electionnight and with the information provided from theNational Tally Room the television networks wereable to predict the final result by 8pm.

Election day was a Saturday, and the second partof the count began on the following Monday inthe Divisional Offices. Ordinary votes for theHouse of Representatives were recounted,including all of the preferences expressed, and finaldecisions made about spoilt ballots. Declarationvotes (early votes, postal votes, absent votes20 andprovisional votes21) were checked and counted.

The Senate votes were counted using thecomputerised ‘Senate Scrutiny System’. All ballotsmarked ‘above the line’ were dealt with manuallyin the Divisional Offices.

Ballot papers marked ‘below the line’ wereforwarded to a centralised count in the capitalcity of the State or Territory where largenumbers of computers were installed.Thepreferences for each candidate were entered intocomputers, which could differentiate between avalid and a spoilt ballot paper.

The ‘above the line’ totals from each Divisionwere also entered into the computer andcombined with the ‘below the line’ votes.Thecomputer then calculated the quota, distributedall of the preferences and produced the results.

In the past, it has taken up to two months toreach a final result in larger States. In the previousFederal elections in 1996, it had taken six weeks.In 1998, with the computerised counting system,all results were finalised within three weeks.

The total cost of the election was A$94,925,371,or A$5.06 per elector22, slightly less than theA$5.08 at the previous Federal Election in 1996.

Evaluation of Electronic Counting

The ballot paperPunchcard and optical scanning systems bothrequire ballot papers of very precisespecifications. If they are to be correctly read bythe scanners, the ballot papers need to beprinted on paper within a specific weight-range, ofspecific dimensions, and with precisely alignedprinting. Perforations, for example where theballot paper is detached from its stub, also needto be precise and clean. Ballot papers need to bekept dry and flat at all stages of the process toavoid swelling or folding. If any of theserequirements are not met, then problems arelikely to ensue23.These are problems which donot affect data-inputting systems.

There are also questions around the layout of theballot paper, as were well-illustrated by the‘butterfly’ ballot controversy in the United Stateselections in 2000. However, the issue of ballotpaper layout exists to some extent for everyelection whatever the method of counting, andwith electronic counting of all kinds there wouldseem to be no reason to deviate too far from themodel of ballot paper as traditionally used in theUnited Kingdom.

Ballot paper security is as important with asystem of electronic counting as it is at present.The use of barcodes on ballot papers wouldrequire specific security measures as well as astandardised format for each type of election.

The voting experienceEach of the three methods of electronic countingexamined in this chapter provides a slightlydifferent voting experience.

Punchcards seem to be the most user-unfriendly,in that the ballot paper needs to be correctlyinserted into a frame and then the voter uses astylus to punch a hole in the desired spot.WithVotomatic ballot papers, the voter is furtherdependent on the correct insertion of a bookletinto the frame in order to reveal which votinglocation relates to which candidate. Furthermore,Votomatic systems as currently used in theUnited States do not seem to guarantee that thevoter makes a clear hole in the card, even if they


20 Votes cast in a polling stationother than the voter’s own, butwithin his/her State or Territory

21 Votes cast in circumstanceswhere an elector’s name cannot befound on the roll or the name hasalready been marked off the roll

22 Excluding public funding.Thefigure including public funding isA$7.87

23 The 1998 elections in theAutonomous Region in MuslimMindanao,The Philippines,encountered misaligned printing onthe ballot papers, therebypreventing the optical scannersfrom reading the votes for aparticular candidate. Ballot papersbeing transported from a pollingstation to the central count got wetand had to be ironed dry beforebeing scanned.

56

have chosen the correct location in which toattempt to do so. Punchcard systems are notideal for postal voters.

Optical scanning systems deviate less dramaticallyfrom the familiar method of marking a ballotpaper, except that they require more precision inmaking a mark (whether by shading or marking across) than at present, which could potentially beproblematic for elderly, disabled and partiallysighted electors.The US-based NationalOrganisation on Disability24 has said of bothpunchcard and optical scanning:

“This system is not accessible to voters who areblind or visually impaired. It is also inaccessible tovoters who have hand, upper body and/or upperarm strength and /or dexterity limitations.”

Some optical-scanning systems also require theballot paper to be completed in pencil – a factorrelatively easy to control within the polling station,but less so for postal votes cast elsewhere.Therequirement for voters not to fold their ballotpapers is also potentially problematic, firstlybecause it can be difficult to enforce, even withinthe polling station, and secondly because with anunfolded ballot paper there is a small risk that thevoted ballot paper will be exposed between thepolling booth and the ballot box.

Data-input systems would not necessarily changethe voting experience for the voter at all.

Centralised versus localised countingIn the United Kingdom, counting is conducted in acentralised location which may be a particularpolling station, a town hall or suchlike. Ballotboxes are brought from polling stations in thesurrounding area and are counted together. AHome Office report in 1994 specifically rejectedproposals to change this arrangement25. Anychange would have an impact on the currentrequirement for the contents of ballot boxesfrom more than one polling district to be mixed.

Punchcards and optical scanning both offeropportunities for counting to take place in acentral location, as at present, or in each individualpolling-station. Counting in each individual polling-station involves a scanner acting as a ballot box,

so that when the vote has been cast, the voterinserts the ballot paper directly into the scanner.The advantage of this is that voters canimmediately be alerted to a spoilt ballot paperand can be given the opportunity to rectify anymistakes or to cast a fresh ballot. It also savestime, particularly in rural areas, on ballot papersbeing transported to a centralised countinglocation. Results from each polling station arethen communicated to a central location fortallying with other results in the area.

Polling-station based counting already happens insome areas of the United States, and is beinglooked upon increasingly favourably in the light of the number of uncounted votes in the 2000election.

It would however, require the expense of ascanner in each polling station, and necessitate thepresence in each polling station of an individualwith sufficient expertise in the use andmaintenance of the machine. It would also makethe process of scrutinising the count morecomplicated and diffuse.

ScrutinyBy removing an open, visual element of theprocess, scanning ballot papers rather thancounting them manually almost certainly reducesthe extent of scrutiny during the count, in thatobservers cannot see decisions being made abouteach individual ballot paper. It also more or lesseliminates the current election night culture,where candidates and their agents are able togain a picture of how the votes have been cast bywatching the piles of votes for each candidatemounting up during the course of the evening.The lack of visual evidence about the likely resultof the election can also leave agents short of timeto make decisions about recounts etc.

However, it does not prevent observers frombeing present, watching the counting operation orraising questions about procedure or theadjudication of questionable ballots. Data-inputsystems allow observers to scrutinise to a similardegree as at present.

Most forms of electronic counting are able toproduce reports of the number of votes counted,


24 Voting System AccessibilityComparison, National Organisationon Disability, August 2001

25 Report of the Working Groupon Automated Vote Counting,Home Office, February 1994

57

and the running totals at any stage during theprocess (though this is an option which has notas yet been taken advantage of in the UnitedKingdom in an attempt to keep procedures assimilar as possible to the traditional system), andcertainly detailed reports can be produced oncompletion of counting.

Also, the existence of paper ballots recording thevoters’ preferences provide a paper audit trail anda method of checking an electronic tally against amanual one.

Accuracy and recountsProlonged use of punchcard (especiallyVotomatic) systems in the United States has donevery little to prove the reliability of scanning holesin pieces of card. Electronic recounts are relativelyeasy to carry out, however the evidence suggeststhat they are less than reliable.The fact thatrecounts produce different results each time isdeeply worrying, and the fact that the voter’sintention can get lost with the handling of ballotpapers means that there is no longer a reliableaudit trail.Though manual recounts with aDatavote system are not too difficult,Votomaticsystems are eye-straining and error-prone.

Optical scanning systems seem to be considerablymore reliable, particularly if the ballot paper hasbeen marked in the appropriate way. One concernhowever is with ballot papers that have not beencorrectly marked. Most of these will be rejected bythe machines for manual adjudication. However,with a system where a valid vote is one where amark is made in a target area, there is a possibilitythat a voter who makes a mark in the target areato indicate disapproval of a candidate will havehis/her vote recorded as a vote in favour of thatcandidate. Similarly, if a scanner is programmed toaccept blank ballot papers (interpreting themsimply as spoilt), there is a danger that a voterwhose mark was placed outside the target areawill have his/her vote ignored.

There is no chance of the voter’s mark beingchanged by handling (as with punchcards), nomatter how many recounts are conducted.However, there could, as with the cases describedabove, be a difference between the interpretationof some papers in an electronic count and in a

manual recount.The physical existence of thepaper ballots provides opportunities for relativelystraight-forward electronic or manual recounting,if necessary.

Data-inputting, whilst open to some level ofhuman error (swiping the wrong barcode, orentering the vote for the wrong candidate) is nomore prone to this than hand counting and isequally as open to scrutiny. As with optical-scanning, voters’ intentions remain intact and,should the need arise, could be entered into thecomputer a second time or counted manually.

There is evidence to suggest that in some cases,trust in vote-counting machinery is so high thatthere can be reluctance on the part ofReturning Officers to hold a recount, even if themargin of victory is very narrow.This kind ofcounting does not even provide for a curtailedrecounting procedure, such as a bundle-check ina manual count, which can at least highlightmajor discrepancies.

SpeedThe main reason for the use of electroniccounting equipment is to reduce the amount oftime taken to deliver the result of an election.Factors affecting the speed of an electronic countare similar to those that determine the speed of amanual count, namely, the size of the electorateand the level of turnout, the complexity of thevoting system, the number of places to be electedwithin a single area (i.e. single or multi-memberwards or constituencies), the time taken totransport ballot papers to the counting centre, thespeed and accuracy of the verification process, thenumber of counting staff/machines, the number ofdoubtful ballots needing adjudication.

Whilst scanners are no doubt quicker at countingthan humans, as Colin Rallings and MichaelThrasher have pointed out, “Estimating savings incount time is not simply a matter of dividing thenumber of papers a machine can count perminute by the number of ballot papers expectedor received.”26

The areas in which electronic counting can bringthe greatest time-savings are in elections for morethan one place, or elections with more


26 Elections – the 21st centurymodel: an evaluation of May 2000local electoral pilots, Colin Rallingsand Michael Thrasher, LocalGovernment Association, 2000, p.36

58

complicated voting systems. Examples of thesewould be multi-member local authority wards,where the manual counting process is morecumbersome and therefore slower and moreerror-prone than a single-member election, andthe London election where voters were castingup to four votes or preferences in two separateelections using two different voting systems.

Scanners tend to lose some of the time gained atthe verification and counting stages because ofthe number of unreadable or rejected ballotswhich then need to be adjudicated manually orremarked and fed through the machines again. Asdiscussed above, this can happen for a largevariety of reasons.

Unforeseen technical problems can also be afactor in slowing down an electronic scanningcount, as the experience in London has shown.This could potentially be the case for data-inputting too, though fewer moving parts makethis less likely.

Data-inputting is not necessarily significantlyquicker in terms of actually counting the numbersof votes cast, but can be greatly time and labour-saving if the election requires calculations to bemade on the basis of the votes cast, as was thecase in the Australian example above.

CostThe cost of introducing a punchcard systemwould have to incorporate not only the scanningmachines but also the holders and styluses forevery polling station.There would also be anincrease in cost in producing the ballot papers tothe strict specifications.

Optical scanning would require spending onscanning machines and software and onspecialised ballot papers.

Data-inputting, depending on the nature andscale of the operation can be relatively cheap inthat existing hardware, in the form ofcomputers, can be used. Software would needto be developed or purchased. Ballot papers canbe as at present in terms of quality and format,though the addition of a barcode would raisecosts somewhat.

The cost of involving a technology-providingcompany in the electoral process will necessarilyincrease the cost of the election initially, thoughover time the reduction in staff hours could helpto offset the cost. If the equipment was to bepurchased rather than hired, then storage,maintenance and upgrading would also contributeto costs.

In summaryPunchcards: There would be little use for theVotomatic punchcard, with all its associatedproblems, in the United Kingdom, as each voterseldom has to cast more than one or two, orvery occasionally three or four, votes at any onetime. Although the use of Datavote punchcardswould be more appropriate in that candidates’names are written on the ballot paper, and in thatit is less problematic in several respects, thedifficulties of placing the ballot paper in a frameand punching a hole in the appropriate locationdoes not encourage the Commission torecommend a move in this direction. Punchcardsystems are unable to deal with preferentialvoting systems as used for local, Assembly andEuropean elections in Northern Ireland.

Optical-scanning: Evidence shows that optical-scanning is capable of speeding up large-scale andcomplex counts, because it speeds up both thephysical counting of marks on ballot papers aswell as any subsequent calculations that arenecessary to reach a result. As regards smaller,simpler elections, the benefits are not aspersuasive, particularly in relation to the expense.Although scanning provides good auditabilitybecause of the existence of paper ballots, aconcern remains regarding the interface betweenhuman-made marks, and automatic machinereading of those inevitably somewhatunpredictable marks; this is the case even withsimple ‘X’ voting but all the more so forpreferential voting.

Data-inputting: Data-inputting also seems to bemost advantageous in a complex election,whether it is large or small-scale, though in alarge-scale election it could be significantly slowerthan an effective optical-scanning operation. Data-inputting cannot greatly improve the speed atwhich the votes are counted, however it can help


59

to significantly reduce the time needed to makeany calculations necessary to arrive at a finalresult. In a single-member first-past-the-postelection, any time-gains achieved in comparisonwith a manual count would be likely to bemarginal. Auditability is good as is the opportunityfor scrutiny. Costs are relatively low and theproblem of machine/human interface andinterpretation does not occur even if the voting ispreferential.


If the United Kingdom were to find itself in aposition where elections or referendums werebeing conducted with increasing frequency, withgreater numbers of choices, and with morecomplicated electoral systems, then theCommission believes that electronic countingcould solve some problems.

However, based on our terms of reference, weconclude that there is insufficient evidence infavour of the use of electronic or mechanisedvote-counting at the current time.We believethat other available solutions would be moreappropriate and more effective. It is ourconsidered view that electronic counting doesnot increase turnout or boost public confidencein the electoral process, nor will it make votingeasier. Increases in efficiency can be achieved butat great financial cost.

Without significant changes to the electoralcalendar, or to the technology available, we donot see great value in further pilots of electroniccounting.

We recommend as follows:

1. If investment were to be made in this sort oftechnology, then the Commission recommendsthat an optical scanner in lieu of a ballot box ineach polling station (instead of centralisedcounting) would present a way forward.Thissolution would at least have the merit ofpreventing the unintentional spoiling of ballotpapers.

2. a) Recounts must be available.The possibility ofmanual recounting must not be ruled out.

b) Guidelines must be in place regarding thecircumstances in which requests for recounts may be granted and whether therecount would be electronic or manual.

c) This information must be available to allconcerned.

3. There must be rigorous and realistic testing ofhardware and software in situ by the ReturningOfficer – with candidates and agents given theopportunity to be present.


60

Appendix 1 – Ballot papers accepted and rejected by counting machines at LondonMayor and Assembly elections

Mayoral ballot papers

First Choice Second Choice Destination First Choice Second Choice Stack Classification Classification

Single mark Single mark Accept Good vote Good vote

Single mark Multi-mark Reject Good vote Multiple vote

Single mark No mark Accept Good vote Uncertain or blank

Multi-mark Single mark Reject Multiple voting No valid first choice

Multi-mark Multi-mark Reject Multiple voting No valid first choice

Multi-mark No mark Reject Multiple voting No valid first choice

No mark Single mark Accept Uncertain or blank No valid first choice

No mark Multi-mark Accept Uncertain or blank No valid first choice

No mark No mark Accept Uncertain or blank No valid first choice


61

London Assembly ballot papers

Constituency London vote Destination Constituency Vote London Vote Vote Stack classification classification

Single mark Single mark Accept Good vote Good vote

Single mark Multi-mark Reject Good vote Multiple vote

Single mark No mark Accept Good vote Uncertain or blank

Multi-mark Single mark Reject Multiple vote Good vote

Multi-mark Multi-mark Reject Multiple vote Multiple vote

Multi-mark No mark Reject Multiple vote Uncertain or blank

No mark Single mark Accept Uncertain or blank Good vote

No mark Multi-mark Reject Uncertain or blank Multiple vote

No mark No mark Accept Uncertain or blank Uncertain or blank

Source: Election of the Mayor of London and the London Assembly: Count Rehearsal at Hammersmith

Town Hall, 22nd February 2000


62

Electronic machine voting has had very limiteduse in public elections in the United Kingdomhaving been used only in three local authoritypilot schemes to date. It has however been usedin one form or another in countries such as theUnited States,The Netherlands and Brazil formany years. It is a polling-station basedtechnology, and therefore does not aim to haveany effect on levels of voter turnout.The mostsignificant difference between this kind oftechnology and the current methods, or any ofthe methods discussed thus far, is that voters nolonger cast their vote on a paper ballot, butrather directly onto a machine which stores therecorded vote and tallies the records.

Electronic Voting Machines The roots of the current models of votingmachines lie in the mechanical lever machines thathave been used in the United States since 1892(see Appendix 1, p. 72).These machines wereseen as a way of reducing fraud by eliminating thepossibility of ballot paper manipulation or ballotbox stuffing.Though mechanical lever machineswere used by 17.8% of voters in the US electionsin 2000, their use is declining rapidly in favour ofnewer technology.

Electronic voting machines, also commonly knownas Direct Recording Electronic machines (DREs)consist of physical apparatus and dedicatedelection software. Prior to election day, thesoftware is installed on a central computer anddata about the political parties, the candidates andthe constituencies/wards and polling districts isentered.The relevant data is loaded on a moduleinto each voting machine.

The machines are positioned within pollingstations and usually unfold to form a self-contained, polling-booth type cubicle.The voter

goes to the polling station and once his/herentitlement to vote has been established, s/heproceeds to a polling booth in the usual way.

The terminal is usually activated by inserting acard or a number.The choice of candidates is thendisplayed on the screen, usually in the format of aballot paper.Voters select their favouredcandidate(s) by touching the appropriate sectionof the screen, either by hand or with a special pen,or by pushing a button.Voters have theopportunity to make changes before confirmingtheir vote.

The vote is electronically stored on a memorycartridge or a disk.At close of poll, the results canbe retrieved by printing vote totals from theindividual machines, or by removing the memorycartridges and sending them to a central location, orthe machines can be equipped with modems whichtransmit the totals to a central counting location.

Some voting machines have audio facilities (usingheadphones), larger font or braille facilities for blindand partially-sighted voters.

Machines of this sort are used in mostmunicipalities in The Netherlands, and in severalcities in Germany.

Belgium started to experiment with electronicvoting in the early 1990s, and in the local electionsin 2000 it was available in every canton and wasused by about 44% of voters.The system uses light-pens on a touch-screen personal computer, andrecords the vote on a magnetic card as well as oncomputer disk.The magnetic cards are depositedinto a ballot box and can be used as a back-up.

The Republic of Ireland intends to pilot this kindof technology in several constituencies in the

Elections in the 21st Century: from paper ballot to e-voting Electronic Machine Voting

63

Electronic MachineVoting

2002 General Elections, and then, if successful, toextend it to the rest of the country for local andEuropean elections in 2004.

Information from the Pilot Schemes

On 4th May 2000, three local authoritiesconducted pilot schemes using electronic machinevoting.They were:

Authority Number Electorateof wards

Bury Metropolitan 1 7000Borough Council

Salford City Council 1 7324

Stratford-on-Avon 18 61,000District Council (+10 parish

elections)

All three authorities used touch-screen systems;these were provided for Bury and Salford byTrilogy Information Systems1 and for Stratford-on-Avon by Nedap/Powervote2.

In all cases staff were involved in a trainingprocess, and the voters were informed of thepilot scheme both directly and through localand/or national media coverage. In Bury andSalford a mock election was held during themonth prior to the election, to give voters theopportunity to come and try the equipment.Similarly, Stratford organised a series of 30 – 40‘road show’ type events for people to familiarisethemselves with the equipment, and also made amachine available in the District Councilheadquarters for people to try out.

Detailed back-up strategies were put in place. InBury and Salford, the machines themselves wereprovided with a source of back-up power for upto eight hours. In Stratford, car batteries werebrought in to provide back-up provision. Sparemachines were on hand in case of mechanicalfailure, and in Bury ballot papers and ballot boxeswere available in case of complete system failure.

In Bury and Salford, the system was used in fiveand four polling stations respectively, each

equipped with two voting machines.Technical staffprovided by the company were present, alongsidepolling staff, throughout the day at each pollingstation. In Stratford, there were 80 polling stationswith one machine each, and several mobiletechnical teams, able to reach any polling stationwithin 20 minutes. Machines underwent a pre-opening readiness test to ensure correctoperation, and election staff also produced a‘Zero Report’ from each machine to verify thatno votes were stored in the machine’s memory.In Bury and Salford, this report was signed by thePresiding Officer and poll clerk and the machineswere then activated at the opening of the poll bythe insertion by the Presiding Officer of anadministration card and a Personal IdentificationNumber.The card was then removed.

In Bury and Salford, the electoral register washeld on a laptop computer in addition to thepaper register.Voters arrived at the polling stationand stated their name and address and had theirnames marked off the paper register in the usualway.The polling staff inserted a smartcard into avalidation device attached to the laptop computerand entered the voter’s electoral number into thecomputer. Once validated, the card was given tothe voter to take to the polling booth containingthe voting machine.The card was then insertedinto the machine thereby initiating the votingprocess. In this way, a link between the vote castand the voter’s registration number is created.

In Stratford, voters presented themselves in theusual manner and were then given a ticket.Thenumber on the ticket was written next to thevoter’s name on the electoral register.The ticketwas then presented to the Presiding Officer whoactivated the machine on behalf of the voter.Theticket numbers were sequential, and so a ticket ofthe wrong colour or at the wrong point in thesequence would alert the Presiding Officer to avoter not entitled to vote at that machine (forexample in a double polling station). Since themachine numerically records the sequence inwhich votes are cast, the link between the voterand the vote cast is maintained by referring backto the marked register.

A ballot paper appeared on the screen (twoballot papers in wards in Stratford with a district


1 In association with Global ElectionSystems Inc, based in the USA.Themodel used was the AccuVote TSmodel

2 The model used was theNedap/Powervote ESD-1 model

64

and a parish election) and the voter had to touchthe name of the candidate(s) for whom s/hewished to cast a vote. By doing this, an ‘X’appeared in the appropriate place on the ballotpaper. A message appeared asking the voter toconfirm the choices that s/he had made andgiving the opportunity to make changes.Whenthe voter confirmed his/her choice, the vote wasaccepted and could no longer be altered.Themachines were programmed to make itimpossible to vote for more than the requisitenumber of candidates, however a blank ballotcould be cast upon confirmation from the voter.

In Bury and Salford, the system recorded the factthat a particular voter had cast a vote on aninternal audit roll; the voting card was thenremoved by the voter and returned to thepolling staff for re-validation and re-use byanother voter.

At close of poll, the Presiding Officer accessedthe machines (in Bury and Salford this was doneby inserting the administration card and PersonalIdentification Number).The results were printedout from the machine and the audit roll showingthe numbers of votes cast and the disk containingall the data was removed. In Bury, each machinewas connected to a telephone line and the datawas transmitted to a server at the central count,where results were displayed on a large screen;the audit roll and disk were delivered to thecentral count for verification. In Salford, resultswere downloaded from the machines to a disk,and were delivered to the central count alongwith the printed results and the Zero Report.

In Stratford, the presiding officer completed aballot account form and printed out the resultfrom the machine.The ballot modules wereremoved from the machines along with theprintout and ballot account, and delivered to oneof four collection points in the authority. Fromthere, they were taken to the central count inStratford-upon-Avon. At the central count, therewere three computers with reading units.Theballot modules were inserted into the readingunits and the ballot accounts verified.The votescast in each ward were collated and thendownloaded onto a disk so that the results couldbe displayed in the Council Chamber.

Both Bury and Salford experienced problems ingetting the machines up and running in time forthe opening of the poll; in several polling stationsone or both of the machines were not ready foruse until 40-50 minutes after the opening of polls.In two cases, paper ballots were issued to voterswho were in a hurry. In another polling station,the delivery of the wrong machine meant thatonly one machine was in use up until 1.20pm,when the correct machine was delivered.Thesedelays meant that in Bury not all pre-polling testswere carried out.There was some dissatisfactionwith the ability of the technical staff to resolveproblems when they arose and with the speed ofthe back-up response. Stratford did not suffer anytechnical problems.

Salford issued tendered ballots to several voterswho had inadvertently skipped through the ballotpaper screen by pressing the screen twice.Theywere unable to return to the necessary screenand therefore unable to cast a vote on themachine. Salford also suffered several mechanicalbreakdowns during the day, though no singlepolling station was ever without a functioningmachine. It was thought that voters trying toretrieve the smartcard from the machine tooquickly, or trying to insert the card into the wrongmachine had been the cause of these problems.

In Bury, one polling station had difficulty intransmitting the results to the central countingcentre, and a disk with the results had to bedelivered in person. Bury had hoped to deliverthe result within 30 minutes (10 minutes for thepolling station totals and a further 20 minutes forpostal votes and verification). It actually took 55minutes to declare the result, but it was still thefirst of the sixteen wards in the authority todeclare. Salford’s result came within 35 minutes ofclose of poll, the first result to be declared inGreater Manchester.

In Stratford the results were declared in 2 hours.This was about an hour slower than had beenhoped, though significantly quicker than usual.Themain difficulty was the distance that needed to becovered in transporting the ballot modules to thecounting centre, in some cases a 45-minutejourney away. It was suggested that in futureelections, four counting centres would be used (as


65

is usually the case), and the results transmitted toa central point for tallying.

Authority Total cost of Total cost of pilot/Cost normal election per elector /Cost per elector

Bury Metropolitan £21,410/£3.06 £3,000/£0.43 Borough Council

Salford City Council £5,700 / £0.783 £2,022 / £0.28

Stratford-on-Avon £45,000/ £1.964 £45,000/ £1.96 District Council

Both Bury and Stratford felt that validating thesmartcards took too long and created queues,and that a simplification of the ballot interface onthe screen was necessary to aid voters infollowing the correct procedures.

Salford also suggested the use of three ratherthan two machines per polling station andhaving a supply of ballot papers in case of totalsystem failure.


The NetherlandsFor a period,The Netherlands experimented withthe use of mechanical voting machines. Problemswith the performance of this equipment led tothe introduction of electronic voting machines inthe 1980s and they have been in use ever since.

The Elections Act of 1989 made provision for theapproval by the Minister of the Interior ofsystems of voting ‘…other than by means ofballot papers’5, based on criteria such as themaintenance of secrecy and ease of use forvoters. All equipment is tested against fixedregulations6 by a non-governmental researchorganisation on behalf of the Ministry of theInterior. Even after initial approval, manufacturersof voting machines must submit their machinesfor re-inspection every four years.

With approval from the Ministry, the decisionabout whether or not to use voting machines istaken at municipal level. Around 90% of Dutchmunicipalities, and 85 – 90% of the Dutchelectorate (about 12 million), now use electronicvoting machines, using equipment provided by

two companies, Nedap/Powervote7 and SduUitgevers. One machine is used per pollingdistrict, which covers roughly 1500 voters.

Elections in The Netherlands use an open partylist system where the whole country, province ormunicipality is a single constituency depending onthe type of election8.Voters cast a vote for aparticular candidate within a party list.

Voters enter the polling station and register byshowing their voter identity/polling cards. Amember of staff controls access to the machineand enables it for each voter in turn. A ballot ofthe same format as a paper ballot is displayed onthe screen. If there are several elections then,depending on the model of the machine, either allballots are shown on the screen at the same time(but clearly separated), or they are displayed in asequence and voters, prompted by messages,work their way through the series of screens.

The voter touches the name of a candidate withinone of the party lists on the screen, or otherwisepushes a button to cast an abstention.Voters havethe opportunity to correct their choices and whenthey are satisfied, they are required to push a redbutton to confirm that the choices made arethose that they intended to make. No furtherchanges can be made. Once the voter hasconfirmed the vote, a blocking mechanism is inplace to prevent any further votes from beingcast, until a member of the polling station staffenables the machine for the next voter.

Close of poll is at 8pm and at this stage, reportsare printed from the voting machines in eachpolling station.These show the total number ofvotes cast, the number of votes cast per list, percandidate and the total number of abstentions.Theinformation is then transcribed onto the pollingstation’s official report and signed and dated by thePresiding Officer and other polling station staff.Thisinformation is sent, often by fax or e-mail to thecentral count in the electoral area.

The ballot module, which contains all the dataabout votes cast, is extracted from the machineand physically transported along with the pollingstation report to the central count for reading.When all ballot modules across the country have


3 Due to financial constraints at theplanning stage,Trilogy InformationSystems offered Salford a specialrate in order to be able to goahead with the pilot.The cost of thescheme as originally envisagedwould have been around £18,000.

4 Stratford rented the machinesfor the election. Stratford estimatesthat purchasing the votingmachines would increase the costper elector to £3.87 for the firstseven years of use.

5 Elections Act, 28th September1989, Section J 33

6 Regulations for the conditionsand approval of voting machines,The State Secretary for InternalAffairs, July 1997

7 The Nedap/Powervote machinesuse integrated election softwareprovided by Groenendaal bv.

8 In National and Provincialelections, these large constituenciesmay be further divided intoelectoral districts, but only for thepurpose of selecting a geographicalrange of candidates, and not for thepurpose electing representatives

66

been read, and all the votes aggregated, the finalresults can be calculated.

The Netherlands News Agency (ANP) receivesresults directly from municipalities on electionnight and makes an unofficial calculation of thefinal result.These unofficial results are availableabout two hours after close of poll.

Official detailed results are available two days afterclose of poll, with the final result declared threedays later.This timetable has remained unchangeddespite the transition from paper ballots toelectronic machine voting and is largely a matterof procedure and tradition.The technology meansthat final results could be declared much sooner.

Machines are purchased rather than hired, and thecost is borne by the municipalities.The initial costof equipping 9000 polling stations with electronicvoting machines was about £22.5 million. Beforethe introduction of electronic machine voting,elections cost about 2 guilders (£0.60) per electorwhereas now they cost 0.80 guilders (£0.23).Nedap estimates that the initial outlay can berecouped within 4 or 5 uses of the machines.

To date, there have not been any challenges toelection results in The Netherlands relating to theelectronic voting machines or software.

In an attempt to gain greater flexibility than withdedicated election equipment,The Netherlandshas been experimenting with the use of ordinarypersonal computers in the voting process.Thesehave been used in several elections on a pilotbasis in 7 municipalities since 1998, and piloting isdue to expand to include up to 50 municipalitiesin elections between 2002 – 2004. Othercountries have shied away from using ordinarypersonal computers because of the potential forincreased software security risks.

The Netherlands is also developing a system ofonline voting for public elections which it isintending to put into practice on a pilot basis inthe provincial elections of March 2003.

BrazilElectronic voting has been used in Brazil since19969. After 21 years of military rule, democracy

in Brazil was restored in 1985, but suffered fromallegations of electoral fraud.The ‘TribunalSuperior Eleitoral’ which oversees elections inBrazil decided to introduce electronic voting as away of attempting to reduce the problem. It wasalso hoped that it would make voting easier forilliterate voters (up to 20% of the population)and that it would speed up the counting processwhich was taking about 30 days.

Brazil has an electorate of around 109 million.Voting is allowed from the age of 16, and iscompulsory for literate people aged between 18 –70.The voting system is a two-round system, wherea candidate needs 50% of the vote to win. If nocandidate obtains 50% in the first round, the twocandidates with the highest number of votes go toa second round of voting to decide the winner.

Technology was first introduced into the electoralprocess in Brazil on a limited scale in 1990, forthe transmission of results in one state to the‘Tribunal Superior Eleitoral’.Voting however, wasdone by hand in the traditional manner. In 1994,this system was extended across the country.

The mayoral and county elections in 1996 gaveBrazilian voters their first opportunity to castvotes electronically. 33 million voters (nearly athird of the electorate) did so from pollingstations in 52 of Brazil’s largest cities.

In local, state and national elections inOctober/November1998, over 60 million votes(57% of the electorate) were cast electronically onmachines provided by Procomp Amazonia IndústriaElectrônica with a security system designed byModulo Security Solutions and Microsoft.

In the municipal elections of October 2000,electronic voting was extended across the entirecountry for the first time.The first round of theelections was held on 1st October. Once again,machines manufactured by Procomp10 were used.The machines have a personal computerarchitecture but stripped down to essentials inorder to reduce costs and to save energyconsumption. Many of the 326,000 polling stationsin which they were installed lacked reliableelectricity supplies, so the machines were designedto run for 12 hours on a rechargeable battery.


9 Electoral legislation makingprovision for electronic voting waspassed in October 1995

10 In the year prior to theelections, Procomp was bought byDiebold Inc, a US company, andbecame a subsidiary

67

The voting equipment consisted of a unit with anumerical keypad (which in addition to the tennumber-keys had a green, an orange and a whitekey) and a display monitor. Attached to this unitwas a separate keypad.The keypads were alsomarked in Braille, and the units had audiocapabilities for blind and visually impaired voters.Voters had to enter their registration numberinto the attached keypad, thus checking theireligibility to vote and enabling the machine. Eachcandidate was assigned a specific number, and thevoter typed the code number of their favouredcandidate into the numeric keypad on the votingunit.This approach was chosen, because it wasfelt that numbers would be easier to deal withthan letters for illiterate voters.When thenumber had been entered, a picture of theselected candidate appeared on the screen alongwith his/her name and political party. Using thecoloured keys, the voter could then confirm thechoice (green), change it (orange) or cast anabstention (white).

The votes were stored on the computer’s harddisk, using encryption. At close of poll, the datawas downloaded onto a diskette and alsoprinted off.The data was then transmitted overa telephone line to a central location wherevotes were tallied.

Results from all of the major cities were completeby the following morning. Results from theAmazon region were slower, but all results wereavailable by that evening (within 30 hours of closeof poll).

Of Brazil’s 26 states, 15 elections wereconclusive on 1st October ; the other 11proceeded to a second round of voting on 29thof the same month. Results from all of these‘run-off ’ contests were available within 6 hoursof close of poll.

Brazil is lending 152 of its voting machines toParaguay for a pilot in 5 districts during itsmunicipal elections in November 2001.The‘Tribunal Superior Eleitoral’ is planning to use420,000 voting terminals in its elections in 2002 and is currently commissioning aninspection of the software with a view toimproving security mechanisms.

Evaluation

The Voting ExperienceElectronic machine voting as discussed in thischapter remains a polling-station based solution,so is unlikely to have any significant effect on thenumber of people who go to cast a vote.

However, voting by means of touch-screenmachines, can assist the voter in casting aneffective and secret vote once they have reachedthe polling station.Touching a screen rather thanusing pencil and paper requires less manualdexterity and precision, which could make votingeasier for voters with certain disabilities as well asolder voters. Audio and braille facilities as well asadjustable font-sizes, can assist voters with visualimpairments who have traditionally needed toseek assistance in the polling booth, to voteindependently and in secret.The US-based‘National Organisation on Disability’ has said ofelectronic voting machines “This system isaccessible for the overwhelming majority ofpeople with disabilities.”11

Electronic voting machines can also bring peopleinto the voting process because of theopportunity they provide for eliminating spoiltballot papers. Machines can be programmed notto accept attempts to vote for more than therequired number of parties or candidates and toprovide a warning message for any voter whoattempts to do so; equally they can beprogrammed not to accept ballots with fewerthan the requisite number of votes, or to requirethat the voter confirms that this is really whatthey intend to do. In this way the problem ofunintentionally spoilt votes could be almostentirely eliminated, and it would be a decision forthe appropriate authorities whether to allow thedeliberate casting of abstentions.

“I’d be very sorry to go back to the traditionalmethod of ballot papers because the electoratetook to it so well…The 40-50 age-group said“we’re worried about the elderly, they’re notgoing to like it”, but they had no problemswhatsoever.”

Bernadette O’Hare, Stratford-on-Avon District Council


11 Voting System AccessibilityComparison, National Organisationon Disability, August 2001

68

This kind of technology can deal equally as wellwith any type of voting system, whether ‘X’ voting,preferential voting, single ballot paper or multipleballot papers.

CountingThe counting process using an electronic systemis undoubtedly far quicker than manual counting,and the larger and more complex the electionthe greater the time savings. Effectively, the resultfrom each polling station can be known withinminutes of close of poll and communicated to acentral counting location, or otherwise can beknown and aggregated with the results of otherpolling stations within minutes of being physicallytransported to a central counting location.

Human error can be almost completelyeliminated from the counting process, as long asall results are communicated/transported to thecentral counting centre and all results areaggregated, as would be the case with ballotboxes and ballot papers.

Recounting with an electronic system is largelyredundant if it is simply a re-tallying of the sameelectronic records that were tallied the first timearound; the result of a recount will inevitably bethe same as the original result. Clearly, this raisesissues about the ability of candidates and theiragents to challenge results.

Scrutiny and SecurityWhilst the use of electronic machine voting goesa long way to eliminating human error in boththe voting and the counting processes, there is nodoubt that it introduces new problems in termsof potential software and hardware error whichare more difficult to detect.

i. Software designWith any use of software in the electoralprocess, concerns arise about the security ofthat software.The problems could be errors inthe software design or pieces of hidden codeintroduced maliciously.These could have aneffect on the final result of an election.

This is a concern made all the stronger by thefact that with any direct-recording votingmethods there is no paper record of the

voter’s intention.There is a fear that softwareproblems, whatever their cause, would not beimmediately evident or be easily traceable.

Although paper records are not perfect atcapturing voters’ intentions, there is a feelingthat in the case of problems with or challengesto the system, the existence of a paper recordof votes cast would provide the opportunityfor recounting, without having to start from thebeginning and hold the whole election again.

Various methods are available which attemptto address this problem, most of which usesome form of ‘redundant’ recording of the data,i.e. a second record of each vote cast whichcan be referred to if necessary. Some systemsdo this internally by storing an electronic ballotimage of each vote or by recording votes on aroll of paper inside the machine. Others do itexternally by printing the vote cast on a receiptfor the voter to check and deposit in a ballotbox, or encoding the vote on a magnetic cardto put in a ballot box.

All of these solutions using redundantrecording are useful in case of corruption ofone of the sources of data within the system,in that they provide a separate back-up versionof all votes cast. However, they do very little toaddress the problem of software reliability, i.e. isthe machine correctly recording the voters’intentions? As U.S. academic Rebecca Mercuripoints out:

“Fully electronic systems do not provide anyway that the voter can truly verify that theballot cast corresponds to that being recorded,transmitted, or tabulated. Any programmer canwrite code that displays one thing on a screen,records something else, and prints yet anotherresult.”12

It is much easier to monitor the software ofmachines which are designed and usedspecifically for elections, rather than generalpurpose personal computer type technology.This is because the software can be kept muchsimpler, and because it is the only softwareinstalled on the machine.


12 Rebecca Mercuri’s Statement onElectronic Voting, 2001http://mainline.brynmawr.edu/~rmercuri/notable/RMStatement.html

69

Some experts believe that making the softwarecode publicly accessible and therefore totallyopen to scrutiny would enhance security.However others believe that this would simplymake the software more vulnerable totampering.There is little disagreement on thefact that an officially designated third partyneeds to have access to the software code inorder to check it thoroughly with a view tocertifying it for use in public elections.

ii. Hardware designIssues relating to power supply are amongstthe most important in terms of hardwaredesign considerations.The key areas are:

p the need for battery back-up in case of amomentary or more lengthy power cuts, sothat voting can continue

p the need for votes to be recorded on apermanent medium that can not be affectedby a power cut or power surge

p the need for certainty about whether theinterruption of a voting transaction by apower cut meant that the vote had beenrecorded or not

A user-friendly interface is also absolutelycrucial, both in terms of the layout of the ballotand the buttons or areas of the screen thatneed to be pressed in order to cast a vote.

In view of these software and hardwareconsiderations and the fact that key areas ofthe administration of the election take place inthe micro-circuitry of an electronic machine,the role of scrutiny in the election processneeds to be somewhat different from thetraditional role. If all key players in the electionprocess are to be satisfied as to the securityand integrity of the election, then rigoroustesting of the equipment needs to beundertaken.

CostThere is no doubt, that the introduction ofelectronic voting would be costly. An electronicvoting machine currently costs about £2500.Thealternative is to rent machines for individual

elections. Stratford took this approach and it cost£312.50 per machine.

“We will certainly continue to look at touch-screen technology because we can use it forsurveys, referenda – all the things we have tobuild into best-value plans and community plans.But it is expensive – we can’t afford to buythese machines and simply have them hangingaround for elections.”

Stan Monaghan, Bury Metropolitan BoroughCouncil

Clearly, there are also savings to be made.Principally these are ballot paper printing, hiringcount venues and counting staff. In Stratford,these savings counterbalanced the cost of rentingthe machines so that the overall cost camewithin the usual budget. It should be notedhowever, that provision would still need to bemade for the printing, delivery and counting ofpostal ballots.

If the machines are purchased rather than rented,then savings can be cumulative over a period oftime, potentially to a point at which electronicelections become cheaper than traditional paper-and-manual-counting ones13.

Conclusions and RecommendationsThe introduction of electronic machine votingwould be a fundamental change to the votingprocess – it opens up many possibilities butneeds substantial resources to enable theprocess to be properly planned andimplemented. If electronic voting turned out tobe a transitional stage in the modernisation ofthe voting process, it would be a costly transition.

Although we do not believe that electronicmachine voting would have a significant effect onlevels of turnout, it would have the benefit ofpreventing inadvertently spoilt ballot papers,greatly hastening the delivery of results andpotentially improving their accuracy.

Our principal concerns lie with the reduced levelof transparency and diminished possibilities forscrutiny that an electronic voting system is ableto provide.


13 Nedap/Powervote estimates thatsavings begin to be made after the4th or 5th round of elections.

70


1. Pilots of electronic machine voting are appropriateat all public elections (including Parliamentary by-elections) other than a General Election

2. For security reasons, the equipment used shouldbe specifically election-dedicated votingequipment

3. Testing of hardware and software must:

a) be rigorous and realistic, preferably conductedin situ; test modes should not be allowed, andtest votes should be entered by hand, not aspre-prepared data

b) include the Returning Officer – withcandidates and agents also given theopportunity to be present

4. Voting data should be recorded and stored induplicate, in case of damage to data

5. Voting machines should be programmed to allowvoters to cast a blank ballot; however voters mustbe warned that this constitutes a spoilt ballot andbe given the opportunity to cast a valid votebefore confirming their action

6. The voting machines should log all events, byvoters and administrators, in order to maketampering, or attempts at tampering, evident

7. The security of the machines and cartridges onelection day and leading up to it must be as highas ballot paper security

8. During election day, the percentage turnout bypolling station should be made available onrequest to candidates and their agents

9. A detailed breakdown of voting, by polling station,should be made available to candidates and theiragents as soon as possible after the declaration ofthe result.This would require a change to currentelectoral law.

10.In the case of a challenge to the result of anelection, all election data must be made availableto a court.


71

Appendix 1 – Mechanical Lever Machines

These machines were first used in the UnitedStates in 1892 and were being used by almosthalf of the US electorate during the 1960s.Theywere seen as a way of reducing fraud byeliminating the possibility of ballot papermanipulation or ballot box stuffing.Their use isdeclining now, because they stopped beingmanufactured almost twenty years ago, and spareparts are very difficult to find. However,mechanical lever machines were used by 17.8% ofvoters in the US elections in 2000.

The voter activates the machine by pulling ahandle which also closes the privacy curtain of thepolling booth.The machine is equipped with anarray of switches, each marked with the name of acandidate or a referendum issue.Voters flick theswitches of their preferred choices and then pulldown a lever to register their choices.There is aninterlocking mechanism to prevent voters fromcasting more than the requisite number of votes.

Inside the machine is a set of interconnected cogsfor each candidate and issue; the cogs function ascounter wheels.There is a cog each to count thehundreds, tens and units of votes cast. As thelever is returned to its upright position, it turnsthe units counter one tenth of a rotation. Eachtime the units counter makes a complete rotationit turns the tens counter one tenth of a rotation.The tens counter does the same for thehundreds counter. At close of poll, electionofficials open the back of each machine and thepositions of the counter wheels for eachcandidate reveal the votes cast.

Since this kind of equipment is close to becomingobsolete and since its disadvantages have beenwidely discussed and accepted in the UnitedStates14, we have not examined it in this report.


14 For example, Roy G. Saltman,Accuracy, Integrity and Security inComputerized Vote-Tallying,Institute for Computer Sciences andTechnology, National Bureau ofStandards, 1988, 3.3.1-2

72

For a number of years now, telephones havebeen used as a medium for voting in privateelections both in the United Kingdom1 andoverseas, generally as an alternative to voting bypost.The increasing use since the 1980s of touch-tone technology which is quicker, more reliableand user-friendly than its ‘pulse’ predecessor hasfacilitated this development.

Telephone voting has not yet been used for apublic election in this country. However it hasbeen used to elect advisory committees forpublic consultation purposes in Walsall2 and wasused for the first time in a local referendum inMilton Keynes in February 19993 and has beenused as an option in several other localreferendums since then.The piloting of telephonevoting in local elections is now being activelyencouraged by the Department for Transport,Local Government and the Regions.

Voting by phone removes the need for voters tovisit a polling station, and therefore, like postalvoting, has the capacity to increase turnout byincreasing convenience for the voter. Likeelectronic voting, it removes the need for a ballotpaper and allows voters to cast an electronicballot, thereby allowing for speedier andpotentially more accurate counting. As with postaland electronic voting respectively, issues ofsecrecy and security arise.

Generally, voting by phone proceeds as follows:

1. Electors receive a package of informationwhich includes a list of the candidates/options,a telephone number (usually a freephonenumber), a security number and some generalinstructions.

2. Having called the freephone number, the voter,

prompted by a computer-synthesized voiceenters his/her security number on the keypadof the telephone, and the system checks thevoter’s eligibility to vote. If the voter is eligible,the system allows him/her to proceed and casta vote.

3. Prompted by the voice, the voter presses theappropriate numbers on the keypad to indicatehis/her preferred option(s).

4. The voter is asked to confirm the choice and isgiven the opportunity to go back and correctany mistakes; once the voter has confirmed thechoice the transaction is complete.

5. Votes are stored electronically and can betallied very quickly upon close of poll.

Voting can be from a land-line or a mobile phone.For diagram of the voting process, please refer topage 81.

UK experience

Telephone voting has not yet been used for apublic election in the United Kingdom. It hashowever been used to elect advisory committeesfor public consultation purposes in Walsall andmore recently as a component in several localreferendums.The first of these was a referendumto decide on the level of council tax in MiltonKeynes Borough Council in February 19994,where telephone voting was an option alongsidea postal vote. Since then, similar referendums havetaken place in Bristol City Council and theLondon Borough of Croydon (February 2001),once again to advise on council tax and rentlevels; these gave voters the option of voting bypost, by telephone or over the internet.Thesethree referenda were conducted by Electoral

Elections in the 21st Century: from paper ballot to e-voting Telephone Voting

1 For example the elections to theNational Executive Committee ofthe Labour Party; in 2001 thesewere conducted by Election.com(formerly Unity Security Balloting)and gave voters the option ofvoting by post, by phone or overthe internet.

2 Walsall Metropolitan BoroughCouncil has established ‘advisorycommittees’ for some of its local

communities. In November 1998, incollaboration with British Telecom,Walsall MBC held the committeeelections in the Rushall area bytelephone.The voting period was 2weeks, and the service was availablein six languages. BT generated 12-digit ‘Secure Voter IdentificationNumbers’ which Walsall MBCapplied to the electoral register toenable electors to get access to thevoting system.The turnout was

25.3%, slightly less than the localelection turnout in the same areathe previous year.

3 This was a referendum todetermine the level of Council Tax.It was conducted by ElectoralReform Services, in conjunctionwith Milton Keynes BoroughCouncil

4 This referendum won the LGC’sInnovation of the Year Award 2000

73

Telephone Voting

Reform Services in conjunction with therespective councils.

London Borough of CroydonIn July 2000, the London Borough of Croydondecided that it would hold a referendum to giveits 230,000 electors the opportunity to influencethe level of council taxes for the year 20001/2. Itwas also decided that Croydon’s 17,000 counciltenants would be given the opportunity to voteon their rent levels. In both cases, residents weregiven three options5, and were asked to express afirst and a second preference6.

These referendums took place in February 2001.Ballot packages were sent by post to residentsand tenants at the end of January and in earlyFebruary, thereby giving voters at least ten days inwhich to vote. Electors were given the option ofreturning their ballot paper by post, or by votingover the telephone or the internet.They did nothave to register for any one of these options inadvance. Instructions regarding each option wereincluded in the ballot package.The instructionsalso noted that voters were entitled to vote onlyonce, and that any attempts to vote more thanonce would be detected and would not becounted7. A helpline was established for theduration of the voting period.

A computer telephony system equipped withspecialised software was used to provide callerswith an automated, interactive session.The systemconfiguration allowed for the use of over 1,000lines.The telephone voting facility was available 24hours a day until close of poll at 5pm on 14thFebruary.

The instructions for telephone voting included thefreephone number that the voter would have todial, and a sentence drawing the elector’sattention to the ballot number and security codeat the top of the ballot paper.The security codeswere generated internally by Electoral ReformServices, and were also printed onto the ballotpapers internally.

Voters called the number and heard a welcomemessage and a description of the voting process.Prompted by the voice, voters entered theirballot number and security number using the

telephone keypad, thus ensuring their eligibility tovote.Voters then entered their voting intention,again using the telephone keypad.The choicesthat they had indicated were read back to themand they were asked to confirm these choices, orto cancel them and start again. Any mistakes wereautomatically identified and explained to thevoter, and an opportunity was given to return andre-enter the numbers. A closing message let thevoter know that the transaction had beensuccessfully completed.

The turnout in the Council Tax referendum was35.1%.This was just slightly lower than the level ofturnout in the previous local elections in May1998, and was far higher than the response toany previous public consultation exercises carriedout by Croydon. Of those who voted, 4.88%(3,865 voters) chose to do so by phone8, and thehighest percentage of telephone voters in anyone ward was 6.28% in Broad Green.

In the Council Rent referendum, the turnoutwas 24%, with 3.36% (137 voters) opting tovote by phone9.

Close of poll was 5pm on 14th February and theresults of both referendums were declared at11am on 16th February.

The cost of the entire process was £151,532, whichincluded a breakdown of the results by ward.

Overseas experience

As in the UK, the use of telephone voting inpublic elections overseas has been limited.Therehave been small-scale pilots but very few large-scale public elections.

Canada is one of the few countries to have usedtelephone voting for public elections andreferendums. Changes made to the MunicipalElections Act in the Province of Ontario in 1996allowed municipalities for the first time to “…usean alternative voting method, such as voting bymail or by telephone, that does not requireelectors to attend at a voting place in order tovote.”10 Until then, only paper ballots had beenpermitted along with the use of somemechanised counting equipment, which was


5 In the Council Tax referendum,the options were increases of 2%,3.5% or 5%; in the Rent referendumthe options were increases of 0%,1% or 2%.

6 It was intended that the secondpreferences would come into playonly if no option gained a majorityof first preferences.

7 The use of ballot numbers madethis checking facility possible.

8 91.72% (72,679 voters) chose tovote by post; 3.4% (2,693 voters)chose to vote over the internet.

9 95.73% (3,900 voters) voted bypost; 0.91% (37 voters) voted overthe internet.

10 Municipal Elections Act, 1996,c.32, Sched., s.42 (1) (b)

74

regulated by the Province of Ontario.Consequently, 1997 saw both a referendum in the(former) City of North York11 and municipalelections in the Town of Gravenhurst, theTownships of Severn and Tiny conducted bytelephone only.

The Town of Gravenhurst and the Townships ofSevern and Tiny, Ontario, Canada Following the change in Provincial Legislationmentioned above, the Gravenhurst Councilpassed a by-law in August 1997 to allow allelectors to vote by telephone only in themunicipal elections.The aim was to increaseturnout by making voting more convenient forlocal residents, and particularly for seasonalresidents (almost half of Gravenhurst’s residentslive elsewhere in November), and also to reducecosts and gain quicker results.

Bell Canada provided the equipment and servicesfor the municipality, and the Townships of Severnand Tiny also joined the team in an effortmaximise economies of scale and make theexercise more cost-effective.

Bell Canada generated a list of 5-digit IdentificationNumbers, which was handed over to themunicipality.The municipality assigned one of thesenumbers to every elector on the electoral register.This list was kept by the municipality, and BellCanada had no record of which number had beenassigned to which elector.

Bell Canada then generated lists of 8-digit VoterIdentification Numbers and matching 2-digitConfirmation Numbers, one of each of whichwas assigned to every elector.These numberswere kept by Bell Canada, and the municipalityhad no record of which numbers had beenassigned to which electors. In this way, neither BellCanada nor the municipality had the capability tomatch a vote to an individual voter, in accordancewith Canadian electoral law.

The new Municipal Elections Act made it astatutory responsibility that the municipal clerkestablish procedures for the use of new votingmethods or equipment.The municipal clerk wasalso required to provide all candidates (and anymembers of the public who requested it) with

a copy of the procedures at least 30 daysbefore the day of the election, in order toincrease transparency and provideopportunities for scrutiny.

A significant voter education programme wasundertaken which included media work and visitsto civic groups to explain the new technologyand voting methods.

Voter information packages were sent out duringthe week of 20th October.These contained theelector’s 8-digit Voter Identification Number and2-digit Confirmation Number, the telephonenumbers to call, instructions about how to vote, alist of candidates and information about how tocontact the Helpline.There were differenttelephone numbers for electors with touch-tonetelephones and those with dial telephones. Localcalls were free of charge, but voters outside themunicipality had to pay for the phone-call.

The municipality compiled a computerisedelectoral register, which allowed staff at the HelpCentre to answer register-related queries, andalso allowed for the compilation of a markedregister, based on the list of IdentificationNumbers handed over to the municipality eachday by Bell Canada. Scrutineers were entitled toview the latest copy of the marked registerthroughout the voting period, excluding electionday itself.

A network of ‘Revision Centres’ was established,in order to make additions, deletions andcorrections to the electoral register, and toprovide replacement Voter Information Packagesfor voters who had not received them or whowere experiencing difficulty in using theIdentification numbers. Electors were required toproduce proof of identity for all of thesetransactions. Scrutineers were entitled to bepresent at the Revision Centres. Only fourelection officials, with security codes, had accessto the administration of the telephone votingsystem and control of the Identification Numbers.

The telephone lines opened on 30th Octoberand stayed open 24 hours a day until 10thNovember.Voters could call from anywhere inthe world at any time of day, using any kind of


11 This referendum took place inFebruary – March 1997, and askedelectors whether or not they werein favour of amalgamating with theCity of Toronto.

75

phone.The majority of those who voted did soon election day.

The phone-call consisted of an interactive voicesystem.Voters were first asked to enter theirVoter Identification Number and ConfirmationNumber, which the system then attempted tovalidate. If this information was invalid after twoattempts, the interactive voice system explainedthe situation and asked the elector to contacthis/her local municipal office for assistance.

The identification numbers encoded informationabout the voter’s ward and so, once validated, thesystem automatically moved on to a list of thecandidates in the appropriate ward and thenumbers to press to vote for each candidate.Thetelephone system allowed voters to cast anabstention.Votes cast were repeated back tovoters and they were asked to confirm thechoices. Once voters had confirmed their choices,they were not able to enter the system again.

The results were delivered quickly, and the wholeelection remained within the allocated budget,which was the same as the budget for a paperballot election.The staff found the election “theeasiest (they) had ever administered.”12 The onedisappointment was the failure to increaseturnout – which, at 30.1%, remained the same –though this was partly accounted for by the factthat there was not a mayoral contest in 1997.

The referendum in the City of North Yorkproceeded along similar lines.The City Clerk,Denis Kelly estimated that the cost ($350,000)was less than half the price of a normal election13.The telephone system gave voters the option offive different languages, and provided a‘Teletypewriter’ line (equivalent to Minicom14 inthe UK) for hearing-impaired voters. It achieved aturnout of 40.5%, as compared with 31.1% at theprevious municipal elections.The results weredeclared half an hour after close of poll.

In the elections of 2000, Gravenhurst usedtelephone voting once again, and was able tomake savings on its 1997 budget, because most ofthe policies and procedures were already in place.The turnout (with a mayoral race) was alsoslightly higher, at 34%.

However, in 2000 18 rather than 3municipalities used telephone voting systems, allusing the same service provider. Possibly as aresult of this increase in numbers, someproblems were experienced in that some votershad difficulty getting through to the system,either having to stay on the line for a lengthyperiod, or having to try several times, indicatingthat there were not sufficient telephone lines. In2000, Gravenhurst received its results withinhalf an hour, though other municipalities had towait up to two hours.

In 1999, Bell Canada quoted the following pricesfor telephone elections15:

Size of electorate Price per voter

Less than 30,000 $8 – $15

30,000 – 49,000 $5 – $8

50,000 – 74,999 $3.50 – $5

75,000+ $2 – $3.50

Evaluation

The Voting ExperienceTelephone voting has the potential to makeparticipating in elections far more convenient forelectors by removing the necessity to visit aparticular voting site and by extending the votingperiod from a single day to several days orweeks.This is of particular interest in view of theElectoral Commission’s survey after the 2001General Election16 showing that 37% of non-voters said that they had not voted becausethey couldn’t get to the polling station becauseit was too inconvenient, or because they wereaway on election day.The lack of use oftelephone voting in public elections means thatthere is little empirical data regarding its effecton levels of turnout, though the same surveyfound that 66% of non-voters claimed that theywould have been more likely to have voted ifthey had had the opportunity to do so byphone or mobile phone.


12 ‘An alternative voting method:telephone voting’ Cindy AnneMaher, Clerk,Town of Gravenhurst,p.7

13 ‘Phone in system lauded’ , BruceDeMara and Paul Moloney,TheToronto Star, 20th March 1997

14 A minicom is a machine thatenables a hearing-impaired personto use the telephone. Both the

person making the call and theperson receiving the call need aminicom.The minicom has a smallkeyboard and a small screen. Eachuser takes it in turn to type in whatthey want to say.The messages arereceived down the telephone lineand appear on both minicomscreens.

15 ‘Televote: IVR solution’ – apresentation by Bell Canada to theAnnual General Meeting of theAssociation of Municipal Managers,Clerks and Treasurers of Ontario(AMCTO)

16 Attitudes to Voting and thePolitical Process, a surveyconducted by MORI on behalf ofthe Electoral Commission, 4th July2001

76

Access to telephones is generally good. Mosthouseholds and workplaces have a telephone, andthe level of mobile phone ownership is also high(71%)17. All mobile phones and phone booths usetouch-tone technology, as do most home-usersand nearly all workplaces. In any case, it ispossible, as in Canada, to make provision, ifnecessary, for those with dial phones. However,concern remains about those without immediateaccess to a telephone at home or at work. Fullyfunctioning public telephone booths are notnecessarily easy to find, and it would beunacceptable to exclude from the electoralprocess, by reasons of access, those who mayalready to some extent be socially excluded.

Telephone voting can be set up to provide a multi-lingual service, and special Minicom and TextDirectlines can be provided for voters who are deaf orhard of hearing. It can be used for both ‘X’ votingand preferential voting. However, the morecomplex the election (the more elections takingplace at the same time, and the greater thenumber of candidates), the less attractive thetechnology becomes. As with postal voting,telephone voting may be more difficult for thosewith low levels of literacy since the instructions arewritten and there is no immediate human support.

As with other forms of remote voting such aspostal voting, and online voting, if an electionwere to be held entirely by telephone, withoutpolling stations or a specific election day, politicalparties would have to adjust the timing of theircampaigning activities, and some electors wouldundoubtedly regret the loss of the social aspectthat election day can provide.

It is also possible that some electors would feeluneasy about using the interactive telephonesystem, and that they could find it confusing,though it should be noted that interactivephone systems are becoming increasinglyprevalent and accepted. Attention would needto be paid to making the system as user-friendlyas possible, and allowing the instructions andchoices to be repeated at any stage. On theother hand, telephone voting can preventpeople from inadvertently spoiling their vote byexplaining mistakes and giving the voter theopportunity to try again.

Congested telephone lines could causeconsiderable frustration for people attempting tocast a vote, and could potentially deter themfrom doing so.The capacity of the system wouldneed to be sufficient to deal with peak periods.

Secrecy and SecurityWith regard to the secrecy which a voter canexpect whilst casting a vote, similar problemsapply to telephone voting as to other forms ofremote voting such as postal voting and onlinevoting. No form of remote voting can assurethe level of secrecy that can be guaranteed bycasting a vote in a polling booth within a pollingstation. With telephone voting, it is not possibleto know whether a voter has cast his/her votein secret.

There are also issues that arise in relation to thesecurity of remote methods of voting.Votersmust have access to some form ofdocumentation or information that allows themaccess to the voting process.With traditionalvoting methods, it is sufficient that voters presentthemselves at the correct polling station andprovide their name and address.With a remotevoting system, there must be some other ‘key’ tothe process; with postal voting the key is theballot paper and declaration of identity that issent by post; with telephone voting it is aPersonal Identification Number.

Firstly, these PINs are generated. Securitysurrounding the generation of these numbersneeds to be as high as the security surroundingballot papers and their serial numbers.The entirePIN needs to be generated randomly; a PINbased on an existing number relating to anelector, such as an electoral registration number,even if combined with extra randomly generateddigits, runs the risk on enabling the code to be‘cracked’.The numbers must also be sufficientlylengthy to make it virtually impossible to enterthe system with a random number.The use of asupplementary ‘confirmation number’ or ballotnumber makes such random entry considerablyless likely.

Secondly, the PINs are applied to the electoralregister, and allocated to electors. Securitysurrounding the allocation of these numbers


17 The e-MORI Technology Tracker,based on a nationally representativesample of adults, aged 15+ acrossGreat Britain, September 2001

77

needs to be as high as the security surroundingballot paper counterfoils which combine anelectoral registration number with a ballot paperserial number.

However whilst levels of security surroundingtraditional voting methods are largely a questionof physical security (i.e. putting the sensitivematerials in physically secure conditions), with anelectronic system, it is a question of safe-guardingelectronic information.This could be done eitherby using encryption, or, as in Canada, by makingseparate agencies responsible for different partsof the process. For example, the people whogenerate the PINs and count the votes could bequite separate from those who apply the PINs tothe electoral register.This would create an‘information gap’ which would make it impossiblefor either agency to find out who had cast whichvote. However, it would allow for the possibility ofthe two agencies comparing their records to findan individual vote, if they were required to do soby a court order.

Thirdly, the PINs must be communicated to theelectors.The most practical and cost-effectiveway of doing this at the moment is by post,which arouses the same security problems as thedistribution of postal votes – how can one besure that the correct person has received thePIN and thereby gained access to the votingsystem? The only way to protect against the useof PINs by unauthorised individuals is to requirethe PIN to be used in conjunction with anadditional piece of information, such as a date ofbirth or, in countries where they are used, anidentity card number.Without a secondaryidentifier, a voting system is no more secure thanthe way in which PINs (or postal votes) arecommunicated to the electors.

During the interactive phone-call, the system canbe designed to create a record for every aspectof the call, if desired. Records almost certainlyneed to be created for :

p the fact that a call has been made with aparticular PIN (before a vote has been cast)

p the option(s) selected (to be repeated back tothe voter before s/he confirms)

p the confirmed vote

Other information that could be stored is:

p the date and time of the call

p each digit selected by the voter during thecourse of the call (whether or not these turnout to be confirmed)

p the phone number from which the call ismade, if available

Under UK electoral law, where a link between avoter and a vote must be maintained, theconfirmed vote and the voter’s PIN would needto be stored as a single record.

Voters who have already registered a vote wouldneed to be barred from entering the system asecond time in order to avoid disruption tocallers who have not yet cast a vote.

Any disruptions to postal or telephone servicescould be severely problematic for the smoothrunning of a telephone election.

As with electronic voting, where the recordingand the counting of votes takes place withinmicro-circuitry, rather than in ballot boxes andcounting halls, the opportunities for traditionalmethods of scrutiny are much reduced.Theremust be certainty about the fact that the systemis recording each vote correctly, and tallying thevotes correctly. However, opportunities forscrutiny could be created at different – probablyearlier – stages of the electoral process, with thepossibility of extended provision of informationfor scrutineers during the voting period.Thesecould include involving candidates, agents andpotentially members of the public in the testingof the equipment, and providing agents withtotal numbers of people who have voted atvarious stages during the polling period.

Unlike online voting, where the computer systemis vulnerable to external hackers, it is not possibleto use a telephone to sabotage the voting system.Though the telephone voting system is itself acomputer, and therefore potentially vulnerable toexternal hacking, it is possible to ensure that it is


78

not visible to the internet, and that if it is on aninternal network, that the system has adequatesecurity in the form of firewalls. It would also bepossible to use encryption so that anyunauthorised access would not permit deletion,addition or modification of data.

Some people fear the possibility of surveillance.Since this would almost certainly have to be doneon an official level, it is not clear that there is anytechnical way of ruling out such a situation withtelephone voting.

Accuracy and RecountsAs with electronic machine voting, the counting ofvotes with a telephone system is an electronictallying of electronic records.Whilst minimising theopportunities for fraud, it also minimises thepossibility of human error, thus increasing accuracy.Recounts are not meaningful, as inevitably – aslong as all the requisite data has been included –the same result will be produced.

Accuracy is dependent on the hardware andsoftware being correct and fully functioning.Everybody involved in the election process needsto be confident that this is the case, and rigoroustesting and certification procedures clearly needto be in place.

Speed and CostResults can be produced extremely quickly, alongwith detailed statistics, if desired.

As with an all-postal election, savings can be madeon polling station sites and staff, ballot boxes anddelivery of polling station stationery, though therequirement to post a PIN to every electormeans that costs in this area are likely to besimilar to an all-postal election. However, withtelephone voting, further savings can be made oncounting sites and staff.

Beyond a threshold of a few thousand voters, theuse of a telephone voting system is likely to beno more costly than traditional methods of votingor an all-postal election.


The Commission believes that telephone voting

has the potential to increase voter turnout(though there is not yet sufficient evidence of itsuse in public elections to be certain) at amanageable cost. It has the ability to prevent theinadvertent spoiling of ballots, and can makevoting easier for people with certain disabilities.We believe that results could be delivered withgreater speed and accuracy.

However, problems surrounding the secrecy inwhich the vote is cast and the diminishedpossibilities for scrutiny must be addressed.Weare also aware of concerns regarding access tothe technology.


1. We recommend further pilots in telephonevoting. However we recommend that thesepilots initially avoid combined elections andelections with more than three vacancies.

2. Elector cards should be introduced in order toincrease the security of telephone voting.Voters would be required to use a PIN inconjunction with their Elector card number,making personation far more difficult.

A second option would be to make changes tothe way in which electoral registration functions(as has been proposed in the Electoral Fraud(Northern Ireland) Bill).Then voters would berequired to use their PIN in conjunction withtheir date of birth in order to increase thesecurity of the system.This would require thecollection of additional information when theannual canvass takes place or on completion ofthe registration application.

If neither of the above were acceptable, the onlyway in which to increase security would be tosend the ballot number and the PIN to theelectors separately.

3. The voter’s identity must be stripped from thevote and stored separately. No individual orindividual agency should have the capability tomatch the two sets of records.The two sets ofrecords must only be matched if a court orderrequires such action to be taken.


79

4. All telephone calls from within the UnitedKingdom must be free of charge.

5. PINs must be generated randomly, and not onthe basis of electoral registration numbers orany other existing number that may beallocated to the elector.

6. Buying or selling PINs must be made anoffence; the penalties should be widelypublicised.

7. The interactive voice system must be as user-friendly as possible, particularly in terms ofusing ‘plain English’.Voters must have theoption of having instructions and optionsrepeated at any stage.Voters must be given theopportunity to review their choices beforeconfirming their vote.

8. It is essential that there be sufficient telephonelines that at no point do electors have difficultygetting through to the voting system.

9. The system should log all aspects of the call, soas to increase security.

10.Voters should be allowed to cast a blank ballot;however, voters must be warned that thisconstitutes a spoilt ballot and be given theopportunity to cast a valid vote beforeconfirming their action.

11.If the Voter Pack including the PIN number hasnot been delivered and there is not enough timeto apply for a replacement, the voter should beentitled to attend a polling station in the relevantconstituency and be issued with a tenderedballot paper after answering the statutoryquestions (suitably amended).This would avoidelectors being unnecessarily disenfranchised.

12.A detailed breakdown of voting, by the smallestappropriate polling division (equivalent to apolling station), should be made available tocandidates and their agents as soon as possibleafter the declaration of the results.

13.In the case of a challenge to the result of anelection, all election data must be madeavailable to a court.


80

Appendix 1 – Voting by phone: a template (based on Bell Canada’s Televote Callflow Template)


81

Thank you for calling the (year) (type of election) system. Please enter 0 to begin

For security purposes, you will be asked to enter you 8-digit personal identification number as well as you 2-digit confirmation number located in the Voter Information Pack that you received

in the post. Please have this information in front of you before you continue.

Using the keys on you touch tone keypad, enter the 8-digit personal identification number that appears in your Voter Information Pack

Now please enter you 2-digit Confirmation Number

Your identification numbers have been validated.The number shows that you are a voter in the (name) ward/constituency

The (date) (type of election) voting system will ask you to vote for your (office, e.g. MP).You willvote by using the keys on your touch tone keypad. Once you have made your selection it will berepeated and you will have the opportunity to confirm or change you answer before continuing.

Once you have cast your vote, you will not be able to call again.

You may vote for (number candidate(s)). Please listen carefully to all the choices before making yourselection. Then select (number) candidate(s).To vote for (candidate name + party) press 1.To vote

for (candidate name + party) press 2 etc. To hear this list again press star.

You have voted for (candidate name). If this is correct press 7. To re-enter your vote press 9.

You have completed your vote.Thank you and goodbye

97

1 2 3 4

Please hold while we validate your voterinformation

If the voter information is invalid after two attemps,the caller will receive the

following message:I’m sorry the personal Identification

Number you entered is not valid. if you needassistance, please call (telephone number).

Goodbye


82

Online systems are the newest form oftechnology to be applied to the electoral process,and have yet to be extensively used in publicelections anywhere in the world.

There has been a surge in the usage of onlinetechnologies such as e-mail, the internet,interactive digital television and text-messaging inrecent years. Increasingly, business and bankingtransactions are being conducted by thesemethods and governments have also grasped thetrend.The UK government aims to put allgovernment services online by 2005.Voting isbeing seen as one such ‘government service’1.

There have been a number of online voting trialsaround the world (though not as yet in publicelections), and more are planned. In the UK, asmall number of local government referendumshave allowed votes to be cast via the internet, andthe Department of Transport, Local Governmentand the Regions is actively pursuing the possibilityof pilots using online methods for local elections.

Online voting allows voters to cast their ballots viaexisting infrastructures, such as the internet orcable networks, using equipment which need notbe in a polling station such as a personalcomputer, a digital television or a mobile phone.There has been much discussion of theconvenience of these methods of voting and theirconsequent potential for increasing turnout.

Online voting combines the convenience of postalvoting with the accurate and instant count of anelectronic system.Yet, it raises issues of voteraccess to the technology and unites the concernsabout secrecy associated with remote voting withtechnological security issues. Security problemsrelating to e-businesses are regular news itemsand any online voting system would have to

guarantee ballot integrity and convince the publicof its reliability.

Whilst online voting can be based in a pollingstation or in some sort of public kiosk, most of theissues that are raised by using the technology inthat way have been covered in the chapter onElectronic Machine Voting.We shall concentrate inthis chapter on remote online voting.

An online voting system would consist of thefollowing general stages:

p the voter is sent the requisite information andinstructions to gain access to the voting site,including security numbers

p the voter enters the voting site by means of acomputer, a digital television or a mobile phoneand enters the required security information toprove his/her identity

p the voter is presented with a ballot in a similarformat to a paper ballot

p the voter makes the required choice(s),confirms the choice and submits the vote

p the vote is received at the election site, and isstored electronically

p at close of poll all the records of votes cast aretabulated

UK Experience

Online voting in public elections has been verylimited in the United Kingdom, as elsewhere. Nouse has been made in public elections as yet ofvoting via digital television, e-mail or by textmessaging. However, local referendums held by

Elections in the 21st Century: from paper ballot to e-voting Online Voting

1 Speech delivered by DouglasAlexander MP at Wilton Park,October 2001

83

Online Voting

Bristol City Council and the London Boroughsof Croydon and Islington in 2000 and 2001offered internet voting as one of several possiblevoting methods.

Bristol and CroydonIn February 2001, electors in Bristol and Croydonwere given the chance to take part in localreferendums. Both councils worked alongsideElectoral Reform Services to run the ballots.Around 287,000 voters in Bristol and 230,000voters in Croydon were asked to decide the levelof Council Tax in 2001/2. In addition, 17,000council tenants in Croydon were asked to set thelevel of their rent for 2001/2. In each case voterswere given a range of tax/rent level options fromwhich to choose.

Ballots were issued to voters early enough to givemost voters 10 days in which to cast their votes,by post, by telephone or over the internet. Aserial number and a unique security code wereprinted on each ballot. By using the number andthe code voters could gain access to the internetvoting page (and the telephone voting system).Instructions that directed voters to the electionwebsite were provided on the ballot paper.

The overall turnout in the Bristol and Croydonreferendums was 40.2% and 35.1% respectively.Within that, the take-up of internet voting was 2.7%in the Bristol tax referendum and 3.4% in Croydon.In the Croydon rent referendum, take-up waslower, at 0.9%.

There is evidence to suggest a ‘digital divide’.Therewas a significant difference in the take-up of internetvoting for the tax (3.4%) and the rent (0.9%) ballotsin Croydon.The rent ballot was only open tocouncil tenants – less affluent electors who wereless likely to have convenient internet access.Further evidence of this divide is provided byfigures from the Bristol referendum. In theprosperous Bristol ward of Clifton East 5.69% ofvotes were cast via the Internet whereas in the lessaffluent Filwood ward that figure was only 1.12%. InCroydon, the pattern was repeated, with a moreaffluent ward achieving the highest internet turnout.

IslingtonIn May/June 2001 the London Borough of Islington

held a referendum to determine its Councilstructure and to resolve a local education issue2.Once again, internet voting was offered in additionto postal and telephone voting.Voters were givenfive language options on the ballots and the pollswere open for three weeks up to the 4th June.

The overall turnout was 22.3%, with 2.4% ofthose votes being cast over the internet. Althoughpostal voting was once again by far the mostpopular option, the percentage of internet votes(2.4%) just exceeded the number of telephonevotes (2.3%).

International Experience

The US provides some interesting examples ofonline voting, once again using the internet.Theseoccur for the most part not in public elections assuch but in candidate selection processes andprimary elections.

Apart from the examples given below, other stateshave also experimented with internet voting.Washington primary voters in one county castballots in a mock presidential primary in February2000. Prior to that, in November 1999, two Iowacounties held similar non-binding trials. In January2000, Republicans in Alaska held a ‘straw poll’.TheUS Reform Party presidential candidate electionwas also partly held over the Internet.

Federal Voting Assistance ProgramIn the United States a specific law governs voting inpublic elections for citizens living overseas (TheUniformed and Overseas Citizens Absentee VotingAct).This Act is administered on behalf of theDepartment of Defense by the Federal VotingAssistance Program (FVAP).

In response to research that it had carried outshowing that the difficulties involved in castingpostal ballots from overseas were a significantdisincentive to voting, the Federal Voting AssistanceProgram conducted a trial during the US GeneralElection in November 2000, that enabled overseascitizens to vote over the internet3.The project useda voting system that had been specifically tailoredfor this particular election and which was designedto mirror the existing postal voting procedures asclosely as possible.


2 The London Borough of Islingtonworked alongside Electoral ReformServices to conduct the ballot.

3 ‘Voting Over the Internet PilotProject Assessment Report’, June2001, http://www.fvap.ncr.gov

84

The online ballot was open to overseas citizensregistered in particular counties within the statesof Texas, Florida, and Utah, and in the whole ofSouth Carolina.The voters were all volunteersand were almost entirely from the militaryservices. 91 people used the system to registerfor an ‘absent’ vote, and of those 84 used it toactually cast their vote. It was the first time thatelectors had been able to vote over the internetin public elections for positions in all levels ofgovernment.The system had to comply with USelectoral law.

Voters used personal computers with internetaccess from home or from work over a period ofabout 30 days.The FVAP supplied each voterwith a CD-ROM which included software (abrowser plug-in) that electors needed to install inorder to vote. It also included the high-encryptionversion of the browser program for thoseelectors who did not already have it installed ontheir computers.

Voters were issued with digital certificates as themechanism for authenticating their identities.Thiswas done through the Department of Defense’scertificate distribution infrastructure, at no cost toelectors who were all military personnel and theirdependents. In order to receive the digitalcertificate, they were required to appear inperson before an issuing authority withphotographic identification.Voters also had toassign a password to their certificates.

The registered electors were contacted by e-mailto inform them the polling period had begun.They logged on to the FVAP server via theinternet, using their digital signature andpassword.The system logged the time and datethat these actions were taken. Having madehis/her choice on the ballot paper screen usingthe mouse and/or keyboard, the voter had theopportunity to review the choice before clickingon the ‘submit’ button which encrypted the vote.The digital signature (along with the password)was applied to the vote, and the signed,encrypted vote was sent back to the FVAPserver. Once again, the server logged the timeand date that the ballot was submitted,authenticated the digital signature and forwardedit to the server in the appropriate local election

office. Upon receipt in the local election office,the server sent a message to the voteracknowledging the vote.Votes were then storedin encrypted format on the server.Voters alsohad the option of logging on to a website whichinformed them of the status of their vote, i.e.whether it had reached the FVAP server, or thelocal election office server.

Once the election period had ended, twopeople, with privacy keys and passwords, in eachlocal election office removed the signatures,decrypted, randomised and printed the ballots.These ballots were transcribed ontoconventional paper ballots and the results werecounted along with all the other ballot papers inthe relevant jurisdiction.

The FVAP conducted a post-election survey andfound that most respondents did not have anydifficulty in installing the necessary software.However some difficulties arose as a result of alack of familiarity with digital certificates. Aproblem arose when a couple tried to use thesame computer to obtain their digital certificates.In order to do this successfully, the computerneeded to be re-configured – a process that wasfound to be difficult and discouraged one personfrom voting.

FVAP Director Polli Brunelli has stated that “Theelectronic ballots worked well and weexperienced no problems with them”.Theassessment report found that:

p the integrity of the electoral process had beenmaintained4

p the authentication processes had ensuredthat only those entitled to vote had voted(and that they had only voted once)

p the system provided greater protectionagainst alteration of ballots than postalvoting

p the system enhanced the enfranchisement ofthe overseas electors

p inaccurate, incomplete and unclear voteswere eliminated


4 Though it acknowledged that thiswould be far more difficult withgreater numbers of voters, and non-military/Department of Defensevoters.

85

p all votes cast were received and counted

p overseas votes were included in theordinary counting processes, and notdelayed until a later date

The entire cost of the project was $6.2 million.

Arizona Democratic Party PrimaryThe US Democratic Party’s Presidential Primaryelection in the State of Arizona gave voters thechance to cast their ballots over the internet. Insome states, primary elections are controlled bystate election officials, this election however wasrun privately by the party and the ballotingcompany,Votation.com (now known asElection.com).

This primary election took place in March 2000and gave 843,000 registered Democrats thechoice of voting at a polling station, by post orover the internet. In order to vote via theinternet, voters had to request a registration formand provide some personal information that wassubsequently used to verify their identities. Allelectors were provided with a randomlygenerated PIN, which was sent by post.

The internet ballot was open for 4 days from 7th– 11th March, with 124 polling stations open on12th March. Some polling stations provided theoption of internet voting alongside the traditionalpaper ballots. In order to vote over the internet,voters were required to visit the ArizonaDemocratic Party’s website, or Votation.com’swebsite. Once at the site, voters were required toenter their PIN, followed by their date of birth orsocial security number and finally their address, inorder to verify their identity.Voters were theninvited to click on one of four buttons on thescreen, 3 candidates and a ‘No Preference’ option.Having made their selection, the voters wereasked to confirm the choice.The vote wasencrypted and transmitted over the internet tothe election server using Secure Socket Layer(SSL) technology.Those who confirmed werepresented with a screen confirming that theirvote had been cast.

Around 35,000 electors chose to cast their votesvia the internet – about 4.2% of registered

Arizona Democrats. Approximately the samenumber voted by post, with around 18,000people voting at polling stations on election day.The number of people voting at polling stationswas lower than expected, because of the last-minute withdrawal of one of the candidates (afteronline and postal votes had already been cast).Online votes were cast early with nearly half ofthe final total having been cast by noon on thesecond day of voting.

Democrat officials hoped that utilising theinternet would encourage more people to vote.Indeed, the overall turnout increased hugely from12,800 (about 1.4%) in 1996 to 85,970 (about10.2%) in 2000. However, unlike in 2000, the1996 primary was uncontested (Bill Clinton wasunopposed). Furthermore, as well as introducinginternet voting, the 2000 election was the first inwhich postal voting on request was offered.

Whilst more people were encouraged to votethan in the past, there were some technicaldifficulties. It was reported that downloading theelection web page onto voters’ screens was aslow process and in some cases took so longthat voters’ connections timed out5, probably dueto insufficient server capacity.The website wasout of action for an hour on the first day ofvoting, once again most likely due to the fact thatthe servers could not cope with the quantities oftraffic. It was also reported that some electorswere presented with a blank on-screen ballotpaper, rather than one with the four availableoptions6. Also, people using Macintosh computersand older browsers had difficulty getting accessto the site7.

The election was challenged by the VotingIntegrity Project, a pressure group that supportsfurther development of internet voting systemsbut advocates caution until current securityconcerns are addressed. It took out a legal actionin an attempt to stop the election and later tohave the results declared void.The basis of itscase was the debate about voter access to thetechnology: voting opportunities available to somewere not available to all.The Voting IntegrityProject argued that rich, generally white votershad greater access to the internet and that theelection was illegal in discriminating against the


5 Wired News,http://wired.com.news/politics/0,1283,34830,00.html

6 Australian Electoral Commission,http://www.aec.gov.au/about/action_plan.htm

7 ‘Is Internet Voting Safe?’, DeborahM. Phillips & David Jefferson,VotingIntegrity Project Report, July 2000

86

poor and the non-white.This racial dimension tothe debate hinges on evidence that in the USAfrican-Americans and Hispanic-Americans are40% less likely than white-Americans to haveaccess to the internet at home8. In response tothe Voting Integrity Project’s lawsuit the judgerecognised that there was a ‘digital divide’ butallowed the election to proceed.

Contra CostaIn the Californian county of Contra Costa a mockinternet election was held in parallel with the USGeneral Election.This trial, contracted out toSafevote Inc. by the Secretary of State, aimed totest polling-station-based internet voting. Duringthe five-day voting period, 307 people voted.Thecost per voter, as estimated by Safevote, wasbetween approximately US$1 and US$1.709.Although the result of this election was notbinding, it provides an important demonstrationof security issues.Voters logged onto a computerusing a unique ‘Digital Voting Certificate’ (DVC),which is “a cryptographic combination of sixletters and numbers” provided by electionofficials. No computer disk was required as in thecase of some such cryptographic keys.The DVCconfirmed the identity of the voter and his/hereligibility to vote, as well as ensuring the correctballot paper was displayed. It also encrypted andcertified the authenticity of the ballot datasubmitted by the voter.

The technology employed placed voter’scomputers in a ‘stealth mode’ on the internet, sothat voters could see and mark their ballots butcould not be seen by others on the internet.Safevote invited computer hackers to attack theelectronic system “to try to penetrate or cripplethe system”. Safevote states that “No attacker wasable to compromise the network in any way” 10.

Plans for the future

Trials of online voting in public elections aroundthe world have been limited and only now isaccess to the technology becoming sufficientlywide for governments to give it seriousconsideration. Amongst others, Costa Rica,Estonia, the European Union, the Netherlands, theUnited Kingdom and the United States haveexplored the possibility of online voting.

The United KingdomThe Government has commissioned a researchproject into “the practical requirements for theimplementation of remote electronic voting”.Thestudy will attempt to identify technological andsocial obstacles to successfully introducing onlinevoting and ways in which they can besurmounted.The specific concern in theintroduction of such a system is whether it shouldoperate “in parallel with existing arrangements” orwhether “a comprehensive modernisation inwhich polling station and out of station methodsare integrated into one system” is required11.TheDepartment for Transport, Local Governmentand the Regions is actively seeking localauthorities to pilot online voting via the internet,digital televisions or mobile phones – possiblyincluding text messaging.

In Scotland, the Highland Council is proposing toevaluate online voting as a method of increasingvoter participation in this remote location. It plansto investigate public attitudes to and the feasibilityof online voting from personal computers, digitaltelevisions and public kiosks.

European UnionIn September 2000, the EU Commissionestablished a ‘Cybervote Project’ – a research anddevelopment programme that is designing asystem to produce “fully verifiable on-lineelections guaranteeing absolute privacy of thevotes and using fixed and mobile internetterminals”, with a view to increasing voterparticipation.The system is to be piloted in localelections in Kista Borough (Stockholm), Issy-les-Moulineaux (Paris) and in Bremen.The 3000voters involved in the trials will have the optionto vote via the internet or by more conventionalmethods.The project has a budget of 3.2 millionEuros and the Commission believes that asubstantial cost reduction is likely to beachieved12.The project is due to be complete byMarch 2003.

The NetherlandsThe Netherlands has established its own distinct‘Remote Voting Project’. It operates on roughlythe same time-line as the Cybervote.The RemoteVoting Project will experiment with internetvoting from any polling station of the voter’s


8 SecurePoll.com Internet VotingUpdates 3rd March 2000

9 Answer given in response toquestions asked by the Commission,July 2001

10 Safevote’s Contra Costa Report,http://www.safevote.com/contracosta/ccc-fr.zip & Safevote’s Technology,http://www.safevote.com/tech.htm

11 Dept. of the Environment,Transport and the Regions,Summary Research Specification:The Implementation of ElectronicVoting, 2001

12 EU Commission Press Release,13th October 2000,http://www.eucybervote.org

87

choice and from special public ‘voting columns’ inthe Provincial Council elections on 11 March2003. In the run up to this election a number ofpilots will be held that will enable any necessaryadjustments to be made to the system. Inaddition, research into remote online voting willbe undertaken.The aim is to make voting in allpublic elections less “location dependent” and, inthe long-term, “enable every citizen to vote fromany location of his or her choice” 13.

United StatesThe United States Federal Election Commission(FEC) produces guidelines for the running of USelections called ‘Voting System Standards’ (VSS).Independent designers and testers use thesestandards to evaluate voting system hardware andsoftware across the United States.The FEC hasbeen unwilling to permit remote internet voting.The 2001 Voting System Standards provideguidelines which state that:

“At this time it is widely recognized that thetechnologies now used to develop internet-basedbusiness systems do not fully address therequirements and risks associated with votingover the internet. Consequently, the VSS do notpromote Internet voting.14”

However, they also state that:

“The Standards allow for internet voting systemsoperated in parallel with another voting system,and do not address or allow for a stand-aloneInternet Voting System.15”

The California Secretary of State’s Internet VotingTaskforce found electronic fraud was too much ofa risk to permit remote internet voting, althoughit did leave the door open to internet voting frompolling stations. Another US report thought it“appropriate for experiments to be conducted”to test “poll site internet voting”. However, itfound remote voting to be “a significant risk tothe integrity of the voting process, and should notbe fielded for use in public election untilsubstantial technical and social science issues areaddressed”16.

Whilst the failings of the electoral system in the2000 presidential election have spurred on

research into new voting methods, the decision-makers have been unwilling to underwrite therisks in using such bold technology. Security is themajor concern; most have been unable to satisfythemselves of the privacy and secrecy aspects oftransmitting votes from remote sites over theinternet. As discussed above, the US did, however,permit some US citizens overseas to cast ballotsvia the internet in the 2000 presidential election,and there are plans for a similar pilot in the mid-term elections in 2002. Furthermore, at least 12states have been considering legislation to providefor internet voting17, including the possibility of apilot using digital television in a primary electionin Los Angeles18.

EstoniaFollowing significant steps in the field of e-government, Estonia has changed its laws to allowinternet voting in local, parliamentary andpresidential elections. It has also passed legislationthat recognises electronic signatures.The Ministryof Justice is preparing a system that it plans topilot in one municipality in the 2002 localgovernment elections. Should this provesuccessful, the system will be rolled out for theparliamentary elections in 2003.

Reuters reported that 35% of the 1.4mpopulation is now connected to the internet19,and the figure may in fact be as high as 45%20. Inaddition, 90 Public Internet Access Points havebeen established throughout the country. Estonia’slocal authorities are now competing to be thefirst to hold an election online.

Evaluation

AccessLevels of access to the kind of technology thatallows for online voting vary across society.TheCommission surveyed members of theAssociation of Electoral Administrators (AEA) ontheir attitudes towards the introduction ofremote online voting. All respondents agreed thatdifferences in voter access to the technologywould be a disadvantage.

Figures from the Office of National Statistics showa strong link between levels of income and use ofthe internet, with lower usage amongst lower


13 Ministry of the Interior of theNetherlands, 29th May, 2001http://www.minbzk.nl/international/documents/pub2781.htm

14 US FEC VSS, 2001, Section 1.5.4,3rd paragraph

15 Ibid., last paragraph

16 Internet Policy Institute, Reportof the National Workshop onInternet Voting: Issues and ResearchAgenda, March 2001, p.34

17 DM Phillips and D Jefferson,VIPReports, Is Internet voting Safe?,http://www.voting-integrity.org/text/2000/internetsafe.shtml

18 ‘California Firm Prepares to testvoting via Digital TV’, KevinFeatherly, Newsbytes, 29thNovember 2001

19 Reuters, New Media Age,5/7/2001, pp. 35-39

20 Estonia Today, InformationSociety, Fact-sheet 23/1/2001

88

income groups21 (see Appendix 1, p. 96), whichpoints towards something of a ‘digital divide’.

Many people have access to the internet at work,or via public libraries and internet cafes (seeAppendix 2, p. 96).The Government has set up1500 UK Online centres across the country,based in village halls, school, shops etc. Surveyshave found that 51% of British adults haveaccessed the internet at some time22.

That said, access to online technology is growingall the time, from 9% with access from home in1998 to 35% in 2001 (see Appendix 3, p. 97).And it is not simply a question of the internet; inJuly 2001, 6% of adults who had ever used theinternet reported that they had done so using adigital television, and 8% from a mobile phone23.

It may be that the growth in access to digitaltelevision will outstrip the growth in access topersonal computers (for e-mail and the internet),particularly with the predicted switching off ofanalogue television in about 201024.

For many, the problem is not access to thetechnology, but the know-how necessary to use it.

The issue of access arose in relation to theArizona Presidential Primary election in 2000.TheVoting Integrity Project took legal actioncontending that the election violated the 1965Voting Rights Act because it allowed online votersfour days during which to cast their votes, whilstthose voting in person at a polling station wereonly allowed one day. It was also argued thatonline voting increased access to the electoralprocess amongst those with internet access athome (predominantly the wealthy), whilstdiscriminating against those without(predominantly minorities and the poor).

The voting experienceRemote online voting is able to make geographicallocation largely irrelevant.The ElectoralCommission’s report on the 2001General Electionidentified a need for greater flexibility about whenand how we vote.Their survey found that 21% ofnon-voters at the 2001 General Election say thatthey did not vote because “I couldn’t get to thepolling station because it was too inconvenient”25.

Indeed, the Chair of the Society of Local AuthorityChief Executives (SOLACE)’s Electoral MattersPanel, David Monks believes “The increase in postalvotes indicates a demand for more convenientways of voting.Technological developments offerreal-time voting via the internet or email, enablingpeople to vote from any location without the timedelays of the postal system.”

The Commission’s survey of electoraladministrators also found unanimous agreementthat convenience would be an importantpotential benefit of remote online voting.However, this did not necessarily translate intoconfidence about its effects on levels of turnout,with 40% of the respondents undecided aboutthe impact of remote online voting on turnout.

The ballot period can be substantially longerthan the single day allowed for polling-stationvoting.This would also have the effect ofspreading out the level of service demandsplaced on the infrastructure. It would also affectthe way in which political parties run theirelection campaigns.

Online voting via the internet and digital televisioncould offer increased accessibility to electors withvisual impairments. Adjustable font-size and ascreen-reading system with head-phones wouldallow voters who traditionally need assistance, tocast a vote independently and privately.

Certain forms of online voting, such as thoseusing the internet and digital television could alsoallow for the provision of further informationabout the candidates standing for election, thusenabling voters to make a better informedchoice. However, such measures raise numerousand complex issues which fall beyond thisCommission’s remit. However, we consider theissue worthy of further consideration.

A survey conducted by this Commission ofcompanies proposing online voting systemsshowed that almost all proposed systems givevoters the opportunity of confirming the choicesthat they have made before actually casting theballot.The systems can also be configured toprevent voters from spoiling their ballot papers,or at least to warn them that if they confirm their


21 National Statistics, ‘Internetaccess’, March 2001,www.statistics.gov.uk/pdfdir/int0301.pdf

22 National Statistics, ‘Internetaccess: Household and Individuals’,September 2001,www.statistics.gov.uk/pdfdir/int0901.pdf

23 National Statistics, ‘Internetaccess: Household and Individuals’,September 2001,www.statistics.gov.uk/pdfdir/int0901.pdf

24 However, at present only digitaltelevision via a cable network iscapable of interactivity. Neitherterrestrial nor satellite connectionscan currently support the kind ofinteractivity that would benecessary for voting.

25 Electoral Commission,Voterattitudes surveys for the 2001general electionhttp://www.electoralcommission.gov.uk/moripoll.htm

89

current choice it will be a spoilt ballot. It wouldalso be easy to provide a ‘none of the above’option for voters wishing to expressdissatisfaction with all candidates, shouldlegislation so require.

Most systems incorporate a mechanism whichconfirms the fact that the vote has been cast andreceived by the system, as a reassurance to thevoter. Some do this by means of an almostimmediate on-screen message. Others allow thevoter to visit a secure website to check whethertheir name appears, or – for added security –whether there is a confirmation number next totheir name which matches a confirmation numberthat was given to them on completion of theirballot. None of these systems reveals the way inwhich the vote has been cast.

Secrecy and SecurityAlthough increasing numbers of financialtransactions are being conducted online, andalthough many people believe that this meansthat online voting is safe, the security andprivacy issues involved are very different. Forinstance, financial fraud on the internet is notuncommon, and companies are happy tounderwrite this to a certain extent; this is notacceptable in an election. With financialtransactions, customers can be issued with areceipt which confirms exactly what happenedand when; in order to maintain secrecy andprotect the voter from undue pressure, this isnot possible with voting. Customers identities’are intrinsically bound to financial transactions;with a vote, the two must (at least to someextent) be separated.

The following are the main secrecy and securityissues which apply to voting systems:

a) Voter authentication: In order to enter theonline voting system, voters would have toproduce some form of identification which wouldenable the system to check their entitlement tovote.Three methods have been proposed:biometrics, usernames/PINs, digital signatures or acombination of the three.

Biometric identification is perhaps the mostsecure method.This uses personal information

such as voice, finger and retinal data to establishidentities.This technology is still under trial andexpensive. It would also require an enormous(and probably intrusive) amount of data to becollected from the electorate.

Alternatively, voters might be sent a voting packcontaining the necessary security information,probably in the form of security numbers or PINs,much as discussed in the chapter on telephonevoting.Where identity cards26 or smartcards exist, itwould make sense to require that the PIN be usedin conjunction with the number on the voter’scard, in order to further ensure that the PIN wasbeing used by the individual it was intended for.Alternatively, a PIN in conjunction with a date ofbirth or some similar piece of personal informationalready known to the voter could provide similarsecurity. Both of these measures would requirelegislative changes in the United Kingdom.

Security is clearly of utmost importance.However, since one of the main arguments infavour of online voting is convenience to thevoter, one should bear in mind that a system ofvoter authentication which involves too manyadministrative steps will counteract thesepotential benefits. All the more so if the processhas to be repeated for each new election.Security measures should not make it difficult fortwo people to vote from the same terminal.

It is likely that as more and more governmentservices go online, that increasing numbers ofpeople will be issued with some form ofsmartcard (or a series of separate smartcards)that will incorporate a digital certificate27. In orderto be issued with a digital certificate, an individualmust prove his/her identity, by producing adocument such as a passport – the process isstrong enough to put the identity of the individualbeyond reasonable doubt. A smartcard of thissort, along with a PIN, could be one method ofauthenticating voters’ identities and ascertainingtheir entitlement to vote. Specific kinds ofsmartcards can also be used with WAP phones28.

These sorts of digital certificates would also helpin terms of linking the vote to an individual voter,as required by electoral law.The ElectronicCommunications Act (2000) recognised the


26 Finland and Italy have started tointroduce electronic identity cards,with the purpose of facilitatingaccess to electronic governmentservices, and possibly eventually toonline voting

27 Smartcards of this sort couldfacilitate online voter registration

28 Mobile phones which access theinternet via Wireless ApplicationProtocol (WAP)

90

validity of electronic signatures in establishing theauthenticity and the integrity of anycommunication or data. Clearly, the vote and thesignature must be separated before the votes arecounted, and would need to be stored inseparate databases, probably connected by a key.A trusted party, such as the ElectoralCommission, would need to hold the key toensure that the two databases cannot bematched up unless authorised.The two databasesact as an ‘audit trail’ of the election.

The voting companies surveyed by thisCommission showed that most are currentlyworking on the basis of usernames with a PINand/or a password, with the possibility – whereavailable – of a personal identifier (such as a dateof birth or a National Insurance number) foradded security. One system requires voters to re-enter their PIN before confirming their vote inorder to ensure that it is the same personcompleting the process as began it. Although onlyone of the respondents has used a ‘Digital VoteCertificate’, most envisage electronic signaturesand digital certificates as the way forwards.

b) Undue influence: With any form of remotevoting, undue influence is difficult to address.Clearly, bribery and vote-buying and selling iseasier from remote sites than from supervisedpolling stations. One approach (the current one)is to have penalties and to publicise the penaltiesprominently.

This Commission’s survey of online votingcompanies similarly revealed a lack of methodsfor dealing with this issue.The only originalproposal was one which allows multiple entryto the system and multiple votes, with theknowledge that the final vote would replace anypreviously cast. In this way, it was suggested, itwould be impossible for someone attempting toexert undue influence to know that the votethey had seen being cast was actually the finalvote cast. However this would also open up thepossibility of undue pressure being brought tobear on those who had already cast a vote insecret to recast their vote. It would also allowthose casting a vote online the opportunity tochange their mind, which would not be the case for those voting by other methods. It

would be a major departure from thetraditional view of casting a ballot being adecisive, unchangeable action.

Clearly, it would be essential to ensure thathaving cast the vote, there was no way ofreturning to or recreating the ballot paper screenor page as completed and submitted by thevoter.This applies equally to a home computer, acomputer based in a workplace or in a publicplace. It is possible that changes would need tobe made to existing employment law in order toensure the privacy of those wishing to voteonline from their workplace.

Other security issues relate more to thetechnology itself than to the physical location ofthe voter. ‘End to end’ security is essential tosafeguarding the integrity of the ballot. Securitythreats could come from both within the system(source code reliability, election officials) andoutside (hacker attacks). Standards, certificationand testing procedures need to be in place toensure the correct functioning and reliability ofthe system. And clearly, election procedures needto be governed by strict security guidelines toprevent unauthorised people using election sitecomputers.Trust should be divided so that anumber of officials must concur before decisivesteps are taken in the ballot.

The more serious and unpredictable threats areexternal ones.

c) Protecting the voter: With internet voting via apersonal computer, attacks on voters’ machines, inthe form of viruses represent a significant risk, sincetheir machines are unlikely to have the same levelsof protection as the election site computers andwill not be scrutinised by officials. Since PersonalComputers are designed to carry out multipletasks, and are able to download new software(which itself is likely to have vulnerabilities), it isextremely difficult to make them secure.Virusestake many different forms, such as Trojan Horseviruses which hide within another program, wormswhich spread through networks and time bombswhich take effect on a specific date.Viruses couldbe spread by e-mail in the run-up to an electionand could be specifically aimed at damaging thesoftware used by the elector to cast a vote; they


91

could be specifically designed to alter the candidatefor whom the vote has been cast, without thevoter’s knowledge.Virus protection software –already installed on most machines – is the onlyway to prevent this kind of attack.This problemdoes not as yet apply to digital television whichcurrently has far less flexibility and therefore far lessvulnerability than a PC.

One of the e-voting companies surveyedsuggested the possibility of getting every voter todownload a free firewall. Special secure browsersanti-virus programs could also be distributed tovoters. Any such additional software must becompatible with as many types of computers andbe simple to install. It should be noted that votersin the United States Federal Voting AssistanceProject pilot encountered some difficulties indealing with this. Furthermore, care must betaken that necessary security measures do notmake online voting so inconvenient as todiscourage people from casting a vote.

d) Protecting the communication link: Anotherdanger is unauthorised interception or reading ofballots between the vote being cast and beingreceived by the system. Encryption can be used tohelp solve this problem. Furthermore, encrypteddata can be bound with an electronic signature toshow that it has not been altered en route.Thiskind of technology is becoming increasinglycommon for business transactions, as is SecureSocket Layer (SSL) technology29, and is currentlybeing developed specifically for e-governmentservices.The equivalent of SSL technology formobile phones which can access the internet viaWireless Application Technology (WAP), isWireless Layer Transport Security (WTLS)30.

Encryption can also be used at the election sitefor storing all votes cast. If it is considerednecessary it is also possible for a physical recordof each vote cast to be created, for example by‘burning’ each one onto a write-once CD-ROM.This would provide a back-up in case ofcorruption of or interference with data.

e) Protecting the election site: ‘Websitespoofing’ is where somebody creates a websitethat looks like the election website, with a verysimilar web-address, in order to intercept data

and prevent electors from casting a vote.Votersmistakenly visiting this site may have no reason tosuspect that they have not cast their vote on thevalid website.

There are several measures that can be taken toprotect against this kind of scenario. Firstly,election authorities need to be vigilant for suchspoof websites, which tend to be high profile andeasy to spot. Once they have been identifiedsteps can be taken to have them shut downrapidly. Secondly, a website verification code couldbe issued to voters as part of their voting pack.When a voter types in their security ID the realwebsite would show the correct verification code;a spoof website would be incapable of doing so.Thirdly, the ability to log on to a website to checkthat the voting system has received one’s votecould alert a voter to the fact that their vote hasnot been received and that something is amiss.

Denial of service attacks present anotherproblem.They are designed to hurl masses ofinformation at the website servers, or the routers,by making many spurious queries or, attemptingto access the site many times over simultaneously.The aim is to slow the system down andultimately to prevent it from functioning at all.Thisis a very basic form of attack and although theyhave caused well-publicised problems31 there arenow strong methods have been developed fordealing with the problem.The effect of such anattack can be absorbed if a sufficient number ofredundant computer servers are used, and oncethe attack has been detected it is also possible toswitch over to different servers altogether. Suchan attack can also be detected, and before itseffects strike the website, the election sitecomputers can be programmed to refuse furtheraccess to the source of the attack.

A similar form of attack bombards the websitewith potential passwords in an attempt todiscover valid access information through aprocess of trial-and-error. Again, similar defencesto a denial of service attack can be employedalong with ‘firewalling’, which defends againstunauthorised access attempts by creating extralayers of security. One type of firewalling uses aproxy server.This server intercepts all messagespassing through the site, effectively hiding the


29 Secure Socket Layer technologyprovides a secure a secure networkconnection session between a clientand a server, usually between a webbrowser and a web server

30 Wireless layer Transport Securityprovides a secure networkconnection session between a WAPdevice and a WAP gateway

31 Such as the shutting down ofYahoo in February 2000

92

cyber-location of the true computer server.

More pernicious attacks have greater subtlety andact on the source code of the website. Moresubtle attacks might attempt to use ‘gaps’ (flaws)in the programming to gain access to secretinformation.The type of programming employedby software developers is clearly an issue here.There is an ongoing debate about the relativemerits of open-source and closed-sourcecomputer programming. In any case, the securityof the system should not rely on maintaining thesecrecy of its inner structure, since this may notbe possible.What is clear is that the source codeof the system must be made available on anofficial basis to those who have the qualificationsto ascertain just how secure the program really is.

By running the election at many server sites therisks to the integrity of the ballot would beminimised. For example, it is unlikely that aGeneral Election would be run from a singlesystem, but would be far more likely to be basedin local authorities.This would ensure that asuccessful computer attack on an online electionwould have less destructive potential in thattargets would be spread out and susceptible onlyto localised problems should defensive measuresfail. It would also mean that the electoral register,although online, could remain on an authority-by-authority basis, as is presently the case, ratherthan as one centralised electoral register.

The system should not be totally dependent onany one element or connection.That is to say ifone part of the system fails, this should notprevent the system as a whole from functioning.Therefore, the design must incorporate back-upfacilities, such as redundant servers, alternativepower supplies and connections, as appropriate.Any hackers or fraudsters responsible forattempts to corrupt the integrity of an electionmust face suitable penalties.

ScrutinyThe purpose of scrutiny in our current system, isto make the voting and counting processes astransparent as possible. Candidates and theiragents are able to see that the electoral processis being carried out in accordance with the rulesspecifically designed to protect elections from

corruption of any sort. If they see that theprocess is not being carried out properly, of ifbecause of their observation, they believe amistake to have been made, they are able tomake a challenge. Because of this transparency,there is general confidence in the process and inthe result.

Scrutiny of the voting and the counting processesis not possible with remote online voting to nearlythe extent that is possible under the currentsystem. Many of the processes are just not visiblein the same way – one cannot see the votersthemselves, nor their names being marked on theregister and the ballot papers issued, nor the factthat they are casting votes free from undueinfluence or bribery. One cannot see the ballotboxes being sealed nor the individual ballot papersbeing counted into piles for each candidate.

Trust in an online voting system means havingconfidence in the machinery and infrastructure,rather than simply in the physical andadministrative processes. In order to gain thatconfidence, all systems must conform to certaincriteria which ensure that they maintain thesecurity of the electoral process – standards mustbe in place by which to judge each system. Acertification process needs to be in place for allsystems which meet those standards.Furthermore, once a system has been certified,there must be strict protocols about the testingprocedures during the period leading up to theelection and on election day itself – for example,the practice of showing that the ballot box isempty before polls open could be replicated by aprocedure to show that no votes are stored onthe system. Candidates and their agents wouldneed to have the opportunity to be presentduring these tests.

Another measure which could to some extentincrease confidence in the system is to takeadvantage of the fact that online systems are ableto provide more information during the course ofthe election. Most proposed online systems areable to provide turnout statistics at any pointduring the voting process – either for electionofficials only, or on a website.The provision ofsuch information could help to reassurecandidates and agents that the process is working


93

as intended, and it could also help with theircampaigning.

Speed and AccuracyThe counting process is an electronic tallying ofelectronically stored records. It is potentially farquicker and more accurate than a hand count.

The Commission’s survey of electoraladministrators found that all respondents agreedthat fast and accurate vote counting was one ofthe most important potential advantages ofinternet voting, particularly where more complexvoting systems are involved.

This was borne out by the survey of online votingcompanies all of which claimed that severalthousand votes could be counted within minutesof all data being available.

As with electronic machine voting, recounts arenot meaningful as they would simply give thesame total again. Although 80% of electoraladministrators surveyed did not see this as aparticular problem, we acknowledge thatcandidates and their agents are more likely tohave reservations, particularly when there is avery small winning margin.

CostThe cost of online voting would vary enormouslydepending on the type of system employed andthe type of security used (passwords, software,biometric identification).

Setting up a public infrastructure for internetvoting, in the form of personal computers forpolling stations, other supervised sites andelection offices would be costly. Once this cost ismet however, the remainder of the costs wouldbe similar to those for postal voting. One of thee-voting companies surveyed by this Commissionestimated that the cost of remote internet votingused on its own would currently be US$0.20 toUS$0.50 (roughly £0.14 – £0.36) per voter.Another company estimated that after initial setup costs, an online voting system would costabout A$0.5 (about £0.19) per voter.

In addition, there will be costs incurred by using atelephone line to vote via the internet, which

would be publicly funded in the case ofsupervised sites and would be paid for by votersin the case of remote sites.


The Commission regards the internet as atransformative technology which is having and willcontinue to have a profound effect on publiccommunication, including the electoral process.Whatever role the internet may play in relationto elections, it is vital that public confidence in thesystem is maintained.

The Commission believes that by increasingconvenience for the voter, online voting has thepotential to increase turnout (though there isnot yet sufficient evidence of its use in publicelections to be certain), particularly amongstsections of the population which have recentlybecome disengaged from the electoral process,such as young people. It also has the ability toprevent voters from unintentionally spoiling theirballot papers and to deliver results with greaterspeed and accuracy than our current methods.

Our principal concerns lie with issuessurrounding differential access to the technology,the reduced possibilities for scrutiny that anonline voting system is able to provide, and thesecurity of the systems from interference.Webelieve that solutions to the problems of voterauthentication can be found.The electionwebsite, the servers, and the communication ofthe vote from one place to another can be maderelatively secure. However, if remote voting fromhome computers is used, the Commissionremains concerned about the lack of securitysurrounding these machines.


1. Before any pilots using online voting are run, aTechnology Taskforce should be established inorder to evaluate and challenge the system.This would be in addition to any otherevaluations that might be taking place.

2. At present, there is a case for online voting asan addition to existing voting methods and notas a replacement.


94

3. Electors who already have digital certificates(issued by a recognised authority) and thenecessary software must be allowed to usethem as a form of voter authentication.

Electors who do not have a digital certificate,or who prefer not to use it for voting, shouldhave the option of being issued with an ElectorCard by their local authority.Voters would berequired to use their PIN in conjunction withtheir Elector Card number, making personationfar more difficult.

A second option would be to make changes tothe way in which electoral registrationfunctions (as has been proposed in theElectoral Fraud (Northern Ireland) Bill).Thenvoters would be required to use their PIN inconjunction with their date of birth in order toincrease the security of the system.

If neither of the above were acceptable, theonly way in which to increase security wouldbe to send the ballot number and the PIN tothe electors separately.

4. The voter’s identity must be stripped from thevote and both must be sealed and storedseparately. No individual or individual agencyshould have the capability to match the twosets of records.The two sets of records mustbe matched only if a court order requires suchaction to be taken.

5. PINs must be generated randomly, and not onthe basis of electoral registration numbers orany other existing number that may beallocated to the elector.

6. Buying or selling PINs must be made anoffence; the penalties should be widelypublicised.

7. It is essential for there to be sufficient serverswith adequate capacity so that at no point doelectors have difficulty connecting with thevoting system.

8. Voters should be allowed to cast a blank ballot;however, voters must be warned that thisconstitutes a spoilt ballot and be given the

opportunity to cast a valid vote beforeconfirming their action.

9. When an elector has not received the requisitesecurity information in the post, and there isnot enough time to apply for a replacement,the voter should be entitled to attend a pollingstation in the relevant constituency and beissued with a tendered ballot paper afteranswering the statutory questions (suitablyamended).This would avoid electors beingunnecessarily disenfranchised.

10.A detailed breakdown of voting, by the smallestappropriate polling division (equivalent to apolling station), should be made available tocandidates and their agents as soon as possibleafter the declaration of the results. It could beargued that it would logical to extend this to allmethods of voting.

11.In the case of a challenge to the result of anelection, all election data must be madeavailable to a court.


95

Appendix 1

Proportion of adults who have used the internet by social class

Source: ‘Internet Access’, National Statistics, March 2001

Appendix 2

Locations adults have used to access the internet (for personal use)

Source: ‘Internet Access: Household and Individuals’, National Statistics, September 2001


96

0

90

80

70

60

50

40

30

20

10

0

10 20 30 40

Per cent50 60 70 80 90

Respondent’s own home

Respondent’s workplace

Another person’s home

An educationalinstitution

A public library

An internet caféor shop

Somewhereelse

2

78

88

24

3336

Unskilled 27

Partly skilled 33

Skilled manual 37

Skilled non manual 52

Intermediate 65

Professional 78

All adults 51

Appendix 3

Proportion of UK households with home access to the internet

Source: ‘Internet Access’, National Statistics, September 2001


97

0 05 10 15 20

Per cent

25 30 35 40 45

April-June ’98 9

April-June ’99 14

April-June ’00 26

April-June ’01 35


98

(names in bold indicate individuals who have givenoral evidence to the Commission)

Administration and Cost of Elections Project,www.aceproject.org

Australian Electoral Commission, www.aec.gov.au

Jim Adler, President and CEO,VoteHere.net

Manuel Amago, Systems Developer, Entranet

Janet Andrews, City of Toronto

Brenda Armstrong Johnston, Association ofMunicipal Managers, Clerks and Treasurers ofOntario

Ausland Asbjørn, Municipality of Oslo, Norway

John Bambrook, Electoral Consultant,TheydonBois, Essex

Tony Bennett, Harlow, Essex

Todd Biggs,VoteHere.net

Steve Billington, Broxbourne Borough Council

Gerard Boon, Ministry of the Interior andKingdom Affairs,The Netherlands

Ian Brown, Computer Security Researcher, andMember of Advisory Board, Foundation forInformation Policy Research

Max Caller, Constituency Returning Officer, Barnetand Camden

John S.Cartwright, Croydon

Eric Chalker, member, Electoral Reform Society

Norman A.Critchley, Councillor, BoltonMetropolitan Borough Council

Simon Day, Commercial Director, Data &Research Services plc

Paul Doherty, Electoral Registration and ReturningOfficer, Swindon Borough Council

Scott W. Flood, President, iBallot

Bev Forbes, Australian Electoral Commission

Greg Forsythe, Global Election Systems, Canada

Ed Gerck, CEO, SafeVote

George Gill, Leader of the Council, GatesheadMetropolitan Borough Council

Jan Groenendaal, Nedap/Powervote

Des Grogan,Assistant Director of CentralServices (Civic and Administrative), BoltonMetropolitan Borough Council

Terje Gusdal, Ephorma, Norway

Katie Hanson, Parliamentary and Public PolicyOfficer, SENSE

Brian Hayes, Elections Administrator,WiganMetropolitan Borough Council

R.L.Howarth, Leader of the Council, BoltonMetropolitan Borough Council

Elizabeth H. Jarvis


99

Submissions and otherInformation

George Johson, ITV Technology Centre

Kim Joyce, Political Assistant to the ConservativeGroup, Swindon Borough Council

Stephen Judson, Electoral Policy Manager, ElectoralCommission

Mark Jurejko, Electoral Services Unit, DoncasterMetropolitan Borough Council

Jason Kitcat, Co-ordinator, the FREE e-democracyproject

Councillor John Lancaster, City of Salford

Roy Loudon, Nedap/Powervote

Tony Machin, Head of Administration and Elections,Doncaster Metropolitan Borough Council

Alice Maders, Chairman, Bury South ConservativeAssociation, Manchester

Cindy Anne Maher, Chief Administrative Officer,Town of Gravenhurst, Ontario

Stan Monaghan, Deputy Returning Officer, BuryMetropolitan Borough Council

Mikael Nordfors, General Manager,Vivarto VotingSystems

Bernadette O’Hare, Electoral Services Manager,Stratford-on-Avon District Council

Robin Parker, Leader of the Opposition,Stevenage Borough Council

Lynn Parsons, Electoral Reform Services

Robert Peden, Chief Electoral Office, NewZealand

Andrew Pinder, UK Government’s E-Envoy

Debbie Proctor, Elections Officer, BuryMetropolitan Borough Council

Christopher Quinton, Halarose of Oxford

Paul Regan, Constitutional Unit,The Home Office

Jonathan Rew,Assistant Director, Chief Executiveand Legal and Administrative Services Department,Gateshead Metropolitan Borough Council

Noel Rippeth, Leader of the Opposition,Gateshead Metropolitan Borough Council

Herman Ruddijs, Publisher Voting Systems, SduUitgevers,The Hague,The Netherlands

Peter-John Sidebottom, Ministry of MunicipalAffairs and Housing, Ontario, Canada

Rob Steel, Salisbury

Jan Morten Sundeid,The Royal Ministry of LocalGovernment and Regional Development, Norway

Merete Astrup Svartveit,The Royal Ministry of LocalGovernment and Regional Development, Norway

Micheline Twigger, Brazilian Embassy in London

Pål Vigostad, Central Information Office,Norwegian Parliament

Joe Wadsworth, Electoral Reform Services

Angus Ward, Director of Sales, Europe, ElectionSystems & Software

Brian Wichmann,Woking, Surrey

Terry Wilde, Councillor, Doncaster MetropolitanBorough Council

Ian Williams, Liberal Democrat Election Agent,Norwich City Council

Nick Williams, Leader of the Council, NorwichCity Council

Mike Williamson, Chief Electoral Office,New Zealand

Elwyn Wilson, Democratic Services Manager,Three Rivers District Council

Rob Wint, Proposition Manager, Entranet

Elections in the 21st Century: from paper ballot to e-voting Submissions and other Information

100

General

Geoffrey Alderman, British Elections: Myth andReality, B.T.Batsford, 1978

Stephen Ansolabehere (MIT), TechnologyObjectives, Constitution Project, Election ReformInitiative,Working Group on Voting Technology,April 2001

Robert Blackburn, The Electoral System inBritain, St. Martin’s Press, 1995

David Butler and Dennis Kavanagh, The BritishGeneral Election of 1997, Macmillan Press Ltd, 1997

David Butler, Howard R.Penniman & AustinRanney (editors), Democracy at the Polls:AComparative Study of Competitive NationalElections, American Enterprise Institute for PublicPolicy Research, 1981

Richard Clayton, Parker’s Law and Conduct ofElections, Butterworths, 1996 – 2001

Lorrie Faith Cranor, Voting after Florida: No EasyAnswers, December 2000, Revised March 2001

Eric A. Fischer, Voting Technologies in the UnitedStates, Updated January 2001

Oonagh Gay & Chris Randall, ParliamentaryElection Timetables, House of Commons Library,2001

Douglas W. Jones, Evaluating Voting Technology,Testimony before the United States Civil RightsCommission,Tallahassee, Florida, January 2001

Dick Leonard & Roger Mortimore, Elections inBritain:A Voter’s Guide, Palgrave 2001

R.B.McCallum & Alison Readman, The BritishGeneral Election of 1945, Oxford UniversityPress, 1947

H.G.Nicholas, The British General Election of1950, Macmillan & Co. Ltd, 1951

Colin Rallings and Michael Thrasher, LocalElections Handbook 2000, Local GovernmentChronicle Elections Centre, 2000

Roy G.Saltman, Accuracy, Integrity, and Securityin Computerized Vote-Tallying, Institute forComputer Sciences and Technology, NationalBureau of Standards, August 1998

Roy G.Saltman, Conducting Elections in Maryland:Considerations for the Future, A Presentation toThe Special Committee on Voting Systems andElection Procedures in Maryland, January 2001

Attitudes to Voting and the Political Process, asurvey conducted by MORI on behalf of theElectoral Commission, 4th July 2001

Ballot Secrecy, Electoral Reform Society &Liberty, 1997

Building Consensus on Election Reform,TheConstitution Project, August 2001

Election Expenses,The Stationery Office, 1999

Elections – the 21st century model, LocalGovernment Association, 2000

Electoral Arrangement Pilots, Local GovernmentInformation Unit, 2000

Electoral pilots 2002 – the implications fordeafblind people, SENSE Campaigns and Policy


101

Bibliography

Team, November 2001

Modernising Democracy Prospectus forElectoral Pilots – Local Elections 2002,Department for Transport, Local Governmentand the Regions, October 2001

Piloting New Electoral Arrangements, LocalGovernment Association, 2000

Representation of the People Act 1985

Representation of the People Act 2000

Residual Votes Attributable to Technology:AnAssessment of the Reliability of Existing VotingTechnology,The Caltech/MIT Voting TechnologyProject,Version 2: March, 2001

Statistical Analysis of Factors That AffectedUncounted Votes in the 2000 PresidentialElection, Report to the Ranking Minority Member,Committee on Government Reform, House ofRepresentatives, United States GeneralAccounting Office, October 2001

Technology and the Voting Process: Final Report,KPMG/Sussex Circle on behalf of ElectionsCanada, June 1998

To Assure Pride and Confidence in the ElectoralProcess,The National Commission on FederalElection Reform, August 2001

Turnout at Local Elections: Influences on levelsof voter registration and electoral participation,Department of the Environment,Transport andthe Regions, 3rd May 2000

Survey Of Attitudes During The 2001 GeneralElection Campaign, MORI research for theElectoral Commission, July 2001

Voting System Accessibility Comparison,National Organisation on Disability, August 2001

Voting – What Is,What Could Be, Caltech/MITVoting Technology Project, July 2001

Working Party on Electoral Procedures: FinalReport, Home Office, 1999

All-Postal Voting

Leslie Elton, Evaluation of Pilot Scheme,Gateshead Metropolitan Borough Council, 2000

Bernard Knight, Evaluation Report of the PilotSchemes into All Postal Ballots, BoltonMetropolitan Borough Council, 2000

Priscilla L. Southwell, Survey of Vote-By-MailSenate Election: Final Report,Vote-by-MailCitizen Commission, 1996

John Turner, Evaluation Report on Pilot Schemesfor May 2000 Elections, Norwich City Council,2000

1997 Constitutional Convention ElectionReport & Statistics, Australian ElectoralCommission

Doncaster Borough Council All Postal BallotPilot Scheme: Evaluation Report, DoncasterMetropolitan Borough Council, 2000

Evaluation of Pilot Scheme – All Postal Voting,Swindon Borough Council, 2000

Evaluation Report on an Innovative ElectoralProcedure,Wigan Metropolitan Borough Council,2000

Oregon Elections Task Force: Report, BillBradbury (Oregon Secretary of State) & DanaJenkins (President, Oregon Association of CountyClerks)

Oregon’s Vote by Mail Procedures, OregonSecretary of State in association with the OregonAssociation of County Clerks, 2000

Pilot Scheme for Postal Voting at the BoroughCouncil Elections in May 2000: EvaluationReport, Stevenage Borough Council, 2000

Review of the Compulsory Retirement SavingsScheme Referendum, Chief Electoral Office (NewZealand)

Voting by Mail, International Institute forDemocracy and Electoral Assistance, 1999

Elections in the 21st Century: from paper ballot to e-voting Bibliography

102

Electronic Counting

Alastair Robertson, Three Rivers CouncilElection, 4th May 2000: Pilot use of AutomaticCounting Machines, May 2000

Roy G.Saltman, Filling the Institutional Void inElection Administration, Revised January 2001

Roy G.Saltman, Voting Systems,The Bell,Vol.1No.1, May 2000

The Automation of the Elections:A SuccessStory in the ARMM and the Philippines’ ProudLegacy of Democracy to the Rest of the World,National Citizens’ Movement for Free Elections,September, 1999

Evaluation of Pilot Scheme for InnovativeElectoral Procedures: Borough of BroxbourneSemi-Automated Counting Scheme, BroxbourneBorough Council, 2000

Alternative Voting Methods (Part 1) – Overview,Association of Municipal Managers, Clerks andTreasurers of Ontario

Election of the Mayor of London and theLondon Assembly: Count Rehearsal atHammersmith Town Hall, 22nd February 2000

The First Election of the Mayor and Assemblyfor London: Report for the GLA, Office of theGreater London Returning Officer, July 2000

The Local elections and elections for a LondonMayor and Assembly, House of CommonsLibrary Research Paper, May 2000

Report on the conduct of the September 09,1996 Elections in the Autonomous Region inMuslim Mindanao, Commission on Elections,ThePhilippines, May 1997

Report of the Working Group on AutomatedVote Counting, Home Office, February 1994

Review of the 2000 elections for London Mayorand Assembly, Electoral Reform Society, August2000

Revitalising Democracy in Florida,The Governor’sSelect Task Force on Election Procedures,Standards and Technology, March 2001


I.Prosser, Evaluation Report on Electronic Votingand Counting Pilot (Parts A & B), Stratford onAvon District Council, July 2000

Roy G.Saltman, Adopting Computerized Voting inDeveloping Countries: Comparisons with theUS Experience, Computer Professionals for SocialResponsibility, 1998

Michael Ian Shamos, Electronic Voting –Evaluating the Threat, Computer Professionals forSocial Responsibility, 1993

Michael Stanton, The importance of recountingvotes, Agência O Estado de São Paulo, November2000

Electronic Voting and Counting Pilot Scheme:Evaluation Report, Metropolitan Borough of Bury,July 2000

Electronic Voting and Counting Scheme:Evaluation Report, City of Salford, August 2000

Elections in the Netherlands, ElectoralCouncil/Netherlands Ministry of the Interior,TheHague

Telephone Voting

Cindy Anne Maher (Clerk,Town of Gravenhurst),An Alternative Voting Method:Telephone Voting,November 1997

Michael Ian Shamos, Electronic Voting –Evaluating the Threat, Computer Professionals forSocial Responsibility, 1993

Alternative Voting Methods (Part 1) – Overview,Association of Municipal Managers, Clerks andTreasurers of Ontario

Have your say on what you pay, CroydonBorough Council, April 2001

Elections in the 21st Century: from paper ballot to e-voting Bibliography

103

North York Referendum Telephone InformationOfficer’s Handbook, City of North York

Procedure for City of North York TelephoneReferendum – March 3 1997, City of North York,January 1997

Televote: IVR Solution, Bell Canada, January 2000

The Year 2000 Ontario Municipal Elections:Survey of Results – Summary Report,Association of Municipal Managers, Clerks andTreasurers of Ontario (AMCTO)

Online Voting

Tim Brigham, The Internet and the ElectoralProcess, International IDEA, March 1998

Derek Dictson & Dan Ray, Internet Voting:Essential Knowledge for Policymakers,Securepoll.com, April 2000

Derek Dictson & Dan Ray, The ModernDemocratic Revolution:An Objective Survey ofInternet-Based Elections, Securepoll.com, January2000

David M. Elliott (Assistant Director of Elections,State of Washington), Examining Internet Votingin Washington

Ed Gerck, Voting System Requirements, FinancialCryptography ‘01, February 2001

Deborah M. Phillips, Are We Ready for InternetVoting?,The Voting Integrity Project, 1999

Deborah M.Phillips & David Jefferson, Is InternetVoting Safe?,Voting Integrity Project, July 2000

Avi Rubin, Security Considerations for RemoteElectronic Voting over the Internet, AT&T Labs

Berry Schoenmakers, Fully Auditable ElectronicSecret Ballot Elections

Paul Waller, Peter Livesey, Karin Edin, e-Government in the Service of Democracy,International Council for Information Technologyin Government Administration, June 2001

California Internet Voting Task Force, A Reporton the Feasibility of Internet Voting, January 200

‘Internet Voting’ excerpt from Final Report of theElection Technique 2000 Commission, SwedishGovernment Official Reports,

Making our Voices Heard, An Energis WhitePaper on eVoting, 2001

Report of the National Workshop on InternetVoting: Issues and Research Agenda, InternetPolicy Institute, March 2001

Voting Over the Internet Pilot ProjectAssessment Report, Department of Defense,Washington Headquarter Services, Federal VotingAssistance Program, June 2001

What is Electronic Voting? A Whitepaper,Baltimore Technologies, February 2001


104

Electoral Reform Society6 Chancel StreetLondonSE1 0UU

£25

Telephone020 7928 1622Fax020 7401 7789

ISBN 0 903291 22 3

[email protected]

Published January 2002

Date post:	18-Jul-2020
Category:	Documents
Upload:	others
View:	3 times
Download:	0 times

Elections in the 21st Century: from paper ballot to e-voting · partnership with an online voting...

Documents