SECURE SIMPLE EFFICIENT
BIG DATA, ECONOMICS AND CONFIDENTIALITY - ON OPTIMIZING THE MANY SMALL DECISIONS
Kurt Nielsen, CEO Partisia
Agenda • Big Data - economics and private info • Case 1: Towards an automated middleman • Case 2: Confidential benchmarking
SECURE SIMPLE EFFICIENT
BIG DATA Economics and private info
More Information …
More detailed production More individual preferences Positions, images, samples, yours and others …
Clickstreams, previous decisions, yours and others …
Many Optimal Small Decisions
Without big data With big data
Strategic • Top objectives
Tactical • Tactical prior ”rules of thumb”
Operational • Operational prior “rules of thumb”
Strategic • Top objectives
Tactical • Linking the many operational decisions …
Operational • Optimal small scale decisions • Automated decisions
The challenge
Private Info is Valuable – Of Course
E.g. buying • What is it worth to you and
others? • Is anyone cheating? • …
E.g. hiring • Before hiring – who are
you? • After hiring – what did you
do? • … Sensitive personal
info Sensitive business info
Regulation (law/rules) Typically more
strong regulation E.g. rules to
enhance competition
Risk concerns (cost of leakages) Greater diversity in
perception … Mostly incentive
driven …
Great insight from 1979 • Holmstrom (1979) Informativeness principle
– Use any measure of performance to promote incentive
• Myerson (1979) Revelation principle – The best solution is to tell the truth to a trusted
third party acting on your behave
• Shamir (1979) How to share a secret – The ideal (technical) foundation for a trusted third
party
pk
Partisia (2009), “Secure Multiparty Computation goes live” Financial Crypto 2009 and commercial use …
Cited by 7302
Cited by 1841 (Nobel prize)
Cited by 8793
SECURE SIMPLE EFFICIENT
TOWARDS AN AUTOMATED MIDDLEMAN
Many optimal small decisions
The Many Optimal Small Decisions Switching supplier (automatically) when optimal!
Customers
Traders
Exchange
Production Power plants
Nordpool
Lack of competition
Consumers Small firms
Fair competition
Large firms
Supply
Demand
Our primary focus
Professional Procurement – made simple and secure
4. Finding best price
3. Submit bids
1. Request auction
2. Learning your type
5. Accept = contract
1. Request auction 1.
3.
2. 4.
5.
STRONG SECURITY AND AUTOMATION
By the way – we also address cloud security
SECURE SIMPLE EFFICIENT
CONFIDENTIAL BENCHMARKING
Big data through security Danish strategic research www.cfem.dk
European strategic research www.practice-project.eu
Benchmarking to Assist Credit Rating
I need to invest to stay competitive!
Will he go bankrupt?
How competitive
is he?
Their debt: (Mostly known)
BUT too many with too much debt …
Farmers:
Secure Benchmarking to Assist Credit Ra6ng -‐ Based on more and richer data …
VFL
Data
Farmers
Data
RES
Confidential: • Customers id • Valuation of assets
Confidential: • Individual accounts and
production data
Controlled by VFL
Controlled by banks
TTP
The consultancy house
… did I mention that we also address cloud security
Richer Data Set and New Results
ID Bank Score 1 Bank A 0,8 2 Bank A 0,76 3 Bank B 0,9 4 Bank B 0,87 5 Bank B 0,89
Known by VFL (a consultancy house)
Known by Bank A
Known by Bank B
Unique results!
An Example (Portfolio Benchmarking)
Portfolio benchmarking Portfolio benchmarking with bank specific values on land
My bank The other banks
Best Worst Best Worst
My bank
The other banks
SECURE SIMPLE EFFICIENT
CONCLUDING REMARKS
Concluding Remarks • Harvesting the many small gains requires:
– Lots of “true” and relevant information – Simplicity and automation
• As always: – Good models! – Good data!
• Commit to compute directly on encrypted data ! a tool to access better data