BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
BCS-FACS and Central London Branch
Subjects are not Objects
Bernie Cohen, CEng, FBCS, MIEE
Hon Visiting Professor, City University [email protected]
www.asymmetricdesign.com
www.soi.city.ac.uk/~bernie
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
Introduction
The Modelling Relation
from Science and Engineering
via the Frame Problem,
Pragmatics,
Anticipatory and
Sociotechnical Systems to
Projective Analysis (PAN)
elicitating clients' models of their ecosystems,
analysing the sustainability of their strategy and
illustrating the risks to which they are exposed.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
3
The Modelling Relation: Scientific Knowledge
interpret
interpret
commute
Natural System
stimulus
entailment
response
Formal System
precondition
entailment
post-condition
Ontological Commitment: To be is to be the value of a variable [W. V. O. Quine]
Material
cause
Formal
cause
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
4
The Modelling Relation: Engineering
component
component
component
composition by
physical connections
component
model
component
model
component
model
stimulus
response
executeobserved behaviour
parameter
result
calculatetransfer function
commute
interpret
interpretcomposition by
symbolic bindings
Engineered System Formal Specification
Design engineers
identify as
components
just those closed
systems whose models
provide
computationally
effective composition
calculi.
Engineering disciplines
are characterised both by
their component models
and by their collections of
compositional operators.
Design
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
5
Closed Systems
and the Frame Problem* The Frame Problem:
in an OPEN SYSTEM, one cannot identify the set of all state components
that are NOT altered by the occurrence of an event.
Science and engineering make the simplifying assumption that the natural systems they observe are closed,
that is, immune to disturbance from all stimuli that the operative model does not account for.
This distinction between what is and what is not accounted for by the observer’s knowledge is the observer’s
Cartesian cut: What you see is not what you get
*first identified in McCarthy and Hayes, Some Philosophical Problems from the Standpoint of Artificial Intelligence, 1969
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
6
Sociotechnical Systems
* see “Architecting Principles for Systems of Systems”, by Mark W. Maier, University of Alabama.
http://www.infoed.com/Open/PAPERS/systems.htm
Healthcare, military, government and, in fact,
every enterprise currently of interest,
are all sociotechnical systems.
Every sociotechnical system is a System of Systems* (SoS)
whose component systems are autonomous
and collaborate with each other more or less voluntarily.
Each component system must therefore be modelled
as an open system in continuous interaction
with its enclosing ecosystem.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
7
The Orchestrating User
Hence, “emergent behaviour”:
the actual composite behaviour of a SoS
differs from the observer’s composition of its parts
because its composition is not fully understood (e.g. feature interaction in telecoms etc.)
This difference between those system behaviours
that can be predicted by users independent of their use of it,
and those that cannot, is the user’s
Heisenberg cut:
What you get depends on how you use it
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
8
Anticipatory System
The clients for
composed services
(the composite functionality
delivered by an SoS)
are actors.
entailment
prerequisite
circumstance
situation
Final cause:
in the system’s context-of-use,
the situation entails the circumstance,
because the user’s model of demand
contains the counterfactual assertion:
if the prerequisite circumstance
were not to pertain,
then the situation would not occur.
Actors are anticipatory systems*
who derive their
demands for services
from their formulation of
themselves as context-of-use
(e.g. as an enterprise)
for those services.
Anticipative systems are complex
adaptive systems
* see Life Itself by Robert Rosen
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
9
The Modelling Relation: Pragmatics
anticipation
anticipation
modelling the organisation of demand
Model of being driven prerequisite
circumstance
entailment
demand
situation
Experience
experience of
partial satisfaction
Experienced
circumstance
Experienced
situation
Final
cause
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
10
The Frame Problem Revisited Understanding the client to be an anticipatory system,
and given that anticipatory systems are open systems,
then modelling the needs of the client
also suffers from the Frame Problem. But the client’s modelling of her need
in the form of an organisation of demand
constitutes a pragmatics of use, which can be modelled.
Although the client cannot know her needs directly,
she can know them indirectly through her experience of their effects.
This difference between what can and cannot be known
directly by the client about her needs is the client’s
‘endo-exo’ cut
What I want is never exactly what I ask for.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
11
The Modelling Relation: SoS Design
Engineered System Formal Specification
stimulus
precondition interpret
interpret
entailment
entailment
response
post-condition
demand-side
ontology
supply-side
ontology
satisfy?
Experienced
circumstance
Client’s Experience
experience
of
partial
satisfaction
entailment
prerequisite
circumstance
demand situation
Model of being driven
Experienced
situation
anticipate
anticipate
sem
anti
c do
mai
n
as r
eper
toir
e o
f co
mp
osi
tio
nal
app
roac
hes
Compose Direct
orchestration
(user's choice)
There is no universal ontology
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
12
Modeling Sociotechnical Systems
In order to model a client enterprise as a socio-technical system,
it is necessary to situate the observer’s perspective:
The Domain of relevance: What are relevant interactions? i.e. in terms of what kinds of behavior is it defined?
The What: What does the enterprise do? i.e. what is the material nature of the work it does?
The Who for Whom: Who is it being for its customers? i.e. who is your client enterprise serving (and what are the economics of this)
The How: What are the critical identity-defining characteristics of how your client enterprise works? i.e. what are those aspects of how they are organized that are identity defining?
The Why: what makes those identity-defining characteristics of the client enterprise of value within the larger ecosystem of which it is part, particularly in relation to its customers?
i.e. what are the drivers in the larger ecosystem that makes what your client enterprise does of value.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
13
Defining how the client stratifies their
relationship to demand
supply demand
identity realization
WHAT WHO FOR WHOM
HOW WHY Cartesian
Cut
Heisenberg Cut
Endo-Exo Cut
formal cause
material cause efficient cause
final cause
domain of relevance
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
14
Stratification and Asymmetries
WHAT
WHO FOR WHOM
HOW
WHY
1st asymmetry: the technology does not
define the product.
Economies of Scale
2nd asymmetry: the business model does not
define the customer's solution.
Economies of Scope
3rd asymmetry: the customer’s demand does not
define the experience that the customer wants
Economies of Alignment
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
15
PAN (Projective ANalysis)
Tools for Sociotechnical Systems Engineering
Visual PAN: elicits graphical, relational representation of client’s
enterprise model.
Stratifier: transforms into stratified Boolean matrix
Analyser: computes projections of stratified matrix
Cohesion: traces paths through stratified matrix
Landscaper: generates 3D histograms of Boolean matrices
considered as simplicial complexes
Feeds results back to client:
classification, location and magnitude of risks
quantifying value and competitive advantage of flexibility, agility, etc.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
16
Projective Analysis (PAN) Workshops
Black Team:
What are the contexts-of-use and
customer situations that are generating
the demands that you are targeting, and
how will you synchronize the composite
capabilities needed to satisfy them?
Blue Team:
What are the critical technologies that
you have to be able to master and/or
source in creating your products
(constituent performances)?
White Team:
What are the
key constituent performances
that you need in order to construct the
output performances
that go into your market channels?
Red Team:
How are you going to have to customise
and orchestrate outputs
to generate the composite capabilities
you need to synchronize with the
customer situations you are targeting?
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
17
Visual PAN a Visio-based graphical editor
for recording the results of PAN workshops
Layer: Structure- Trace Hierarchy Synchronisation Demand
function
Meta-
ontology
and
its
Stencil
know-
how
capability
process
event
outcome
design
digital
system
digital
process
trace
unit order
digital
order
problem
demand
situation
customer
situation
driver
Clients' models are elicited as binary relations, of restricted signature,
between instances of these types identified by the clients.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
18
Example: A Diabetics Clinic*
*B. Cohen and P. Boxer, Why Critical Systems Need Help to Evolve, IEEE Computer May 2010
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
19
Detail of Diabetics
Clinic
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
20
6-layer stratification
critical technologies
Technology
output performances
Construction Key constituent
performances
Engineering
Customised
outputs
Composite
Capabilities
Customisation
Orchestration
Customer
situations
Contexts-of-use
Synchronization
Visual PAN outputs a text file containing
every pair of typed objects in the graph.
The Stratifier transforms this file into a stratified Boolean matrix.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
21
Strategic Analysis
An r-type proposition A c-type proposition
The Analyser examines this relational structure and reports the occurrence of
patterns symptomatic of the types of proposition that the supplier makes to its clients
in response to their expressions of demand.
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
22
A K-type proposition
A P-type proposition
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
23
Stakeholder and the Value Deficit
The supplier defines the
customer situation in a way
that serves its interests,
and has the control to
subordinate the demands
of the customer situation
to that definition of
demand (in this case for a
K-type proposition)
The supplier as stakeholder
in the way they have defined
their market, and in the way
they expect to satisfy it
The customer as stakeholder
in the way they have expressed
their demand, and in the way
they expect it to be satisfied
The value deficit
is the gap between the
customer’s expectations
and what is actually
supplied.
The more variability and
change there is in the
nature of demands, the
greater the scope for
value deficits to appear
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
24
The Stratified Matrix for the Diabetics Clinic
0
1
2 3
6
5
7
4
0,1 WHAT
2,3 HOW
4,5 FOR WHOM
6,7 WHY
alignment
stakeholders
Value deficit
Alignment
complexity
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
25
Landscapes Each sub-matrix may be considered as a Simplicial Complex* and represented as
a 3-dimensional histogram that graphically illustrates the risks to which the system-of-system is
exposed by inadequate communication among its components.
Y: The number
of data
elements
synchronized
by the platform
X: The separate data platforms within the healthcare system
Z: The number of
other platforms with
this level of
synchronization
Orthotic
Clinic data
Patient
Administration
data
GP
data
These 'holes' threaten
the K-type proposition
'othotist-with-patient'.
*R H Atkin, Q-Analysis, A Hard Language for
the Soft Sciences, Futures, Vol 10, Issue 6, 1978
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
26
-10123456789
1011121314151617181920
q
1
12
23
unitord
er\
consultant
unitord
er\
patie
nt
unitord
er\
gp_bookin
g
tracevent\
waitin
g_ro
om
unitord
er\
prim
ary
_care
_gro
up
unitord
er\
directo
rate
unitord
er\
ort
hotics_dept
tracevent\
gp_dia
gnosis
tracevent\
c_dia
gnosis
tracevent\
c_w
aitin
g_lis
t
tracevent\
ort
hotics_w
aitin
g_lis
t
tracevent\
needs_adju
stin
g
tracevent\
o_re
ferr
al
tracevent\
readym
ade_fo
otw
ear
tracevent\
bespoke_fo
otw
ear
unitord
er\
acute
_tr
ust
tracevent\
repair_and_adapta
tio
n
tracevent\
surg
ical_
outc
om
e
tracevent\
podia
try_outc
om
e
tracevent\
physio
_outc
om
e
tracevent\
physio
_w
aitin
g_lis
t
tracevent\
podia
try_w
aitin
g_lis
t
tracevent\
request_
for_
sta
bili
satio
n
unitord
er\
com
ponents
_biz
tracevent\
com
ponents
unitord
er\
uk_agent
tracevent\
i_sto
ck
tracevent\
m_
sto
ck
unitord
er\
sto
ck_pro
duct_
supplie
r
unitord
er\
bespoke_m
fr
tracevent\
b_sto
ck
tracevent\
o_goods_in
tracevent\
b_goods_in
tracevent\
import
s
tracevent\
adju
ste
d_sto
ck
unitord
er\
podia
try_dept
unitord
er\
physio
thera
py_dept
tracevent\
direct_
sale
n
simplexes
slice2-3_transProcess Alignment Complexity
X: The component outputs of the healthcare system
Y: The
number of
processes
that must be
aligned to
generate
each output
Z: The
number of
other
outputs
sharing this
level of
alignment
The orthopaedic consultant, the
patient and the general practitioner
Diagnostic
processes
Orthotic
treatment
process Orthosis
manufacture
BCS FACS and Central London Branch Feb 16 ©2012 Prof. B. Cohen, City U
27
Further Work Software Engineering: Transform current prototype software
from VB (!) to Open Source under e.g. Eclipse.
Graph Theory: Formalise and implement the graph-theoretic
operators involved in the composition of graphs from different
subjects, given the ontological mappings on which they agree.
Algebra: Select an algebraic structure that adequately represents
the composition and transformation of 'triply articulated'* models.
* P. Boxer and B. Cohen, Modelling the Enterprise and its Actors as Triply-Articulated Anticipatory Systems, ICCS
2004