Andy Stirling SPRU – science and technology policy research presentation to workshop on …...

Post on 31-Mar-2015

216 views 1 download

Tags:

Transcript of Andy Stirling SPRU – science and technology policy research presentation to workshop on …...

Andy Stirling SPRU – science and technology policy research

presentation to workshop on ‘… Management of Variation and Diversity’,

Forum Chriesbach, Zurich, 15-8 April 2007

1: Diversity – much discussed but surprisingly little analysed

2: The nature of diversity and some different approaches

Pathways, Politics and Diversity

in the Governance of Technological Transitions

3: A new general conceptual framework and heuristic

explanatory / normative heuristic – destabilising / keystone concept(epistemic ‘niche’ not ‘landscape’)

4: Some implications: towards ‘Transition Portfolios’?

Disciplinary Interests in Techno-Institutional Diversity

History and philosophy of science (Merton, Kuhn, Nowottny) interactions in

diverse scientific disciplines enhance rigour and creativity

Research and innovation strategies (Rosenberg, Rothwell, Landau) diverse

portfolios offer flexibility and learning across programmes

Governance of science and technology (Gibbons, Walker, Jasanoff) diversity of engagement

promotes trust, democracy and social robustness

Regulation of technological risk and sustainability (Norgaard, NRC, EEA) diverse knowledges

/ perspectives foster adaptive, precautionary policy

Strategic technology policy (eg: energy security) (Brooks, Folke, EC, IEA) diversity of

technological options provides resilience to shock and surprise

Mainstream market economics (Ricardo, Markowitz, Hayek) diversity of firms resists

oligopolistic concentration, enhances competition

Evolutionary economics (Grabher, Dosi, Metcalfe, Arthur) diversity of actors,

functions and contexts generates ‘fitness’ in innovation

PROVENANCETEMPORALITY

transientdisruption(transient disruption)

enduringpressure

(enduring pressure)

internal(to system / frames)

external (to system / frames)

STABILITY

DURABILITY

RESILIENCE

ROBUSTNESS

Diversity and Sustainable Technologies

Dynamic Properties of Sustainable Infrastructures (Adrian)

PROVENANCETEMPORALITY

transientdisruption

enduringpressure

internal external

STABILITY

DURABILITY

RESILIENCE

ROBUSTNESS

Diversity and Sustainable Technologies

diversity is ubiquitous in general strategies

towards sustainable technological systems

Dynamic Properties of Sustainable Infrastructures

eg: major element in energy security policy

Diversity in Energy Technology Transitions

Like all technological systems, diversity is key emergent property of energy

systems, even if not deliberately planned

reflect context diversity allows sensitivity to heterogeneous context

Austrian hydro, Norwegian gas, Swedish biomass

point arose strongly – especially in

Raimund’s and Eva’s presentations

Diversity in Energy Technology Transitions

Socio-technical diversity also relevant to deliberate transition-building

foster innovation diverse configurations and contexts stimulate creativity

interactions between microgeneration technologies

Like all technological systems, diversity is key emergent property of energy

reflect context diversity allows sensitivity to heterogeneous context

Diversity in Energy Technology Transitions

foster innovation diverse configurations and contexts stimulate creativity

mitigate ‘lock-in’ diverse portfolios resist pressures to concentrate

support for different PV concepts and trajectories

Like all technological systems, diversity is key emergent property of energy

reflect context diversity allows sensitivity to heterogeneous context

Socio-technical diversity also relevant to deliberate transition-building

Diversity in Energy Technology Transitions

foster innovation diverse configurations and contexts stimulate creativity

mitigate ‘lock-in’ diverse portfolios resist pressures to concentrate

hedge ignorance ‘eggs in different baskets’ anticipate surprise

energy security or unforeseen environment issues

Like all technological systems, diversity is key emergent property of energy

reflect context diversity allows sensitivity to heterogeneous context

Socio-technical diversity also relevant to deliberate transition-building

Diversity in Energy Technology Transitions

foster innovation diverse configurations and contexts stimulate creativity

mitigate ‘lock-in’ diverse portfolios resist undue pressures to concentrate

hedge ignorance ‘eggs in different baskets’ anticipate surprise

accommodate dissent diversity helps reconcile plural values and interests

polarisation behind nuclear, renewable, clean coal…

both about orientation and mode of transition path

Like all technological systems, diversity is key emergent property of energy

reflect context diversity allows sensitivity to heterogeneous context

Socio-technical diversity also relevant to deliberate transition-building

– What is diversity? Variety? Difference? Concentration?

– Which things to diversify?

Towards a Systematic Approach

Staffan: technologies

Raimund: communities

Fred: actors, attributes

Kornelia: 3 types of variety in TIS

Eva: expectations, levels, local ‘solutions’

Daniel: regime plasticity – supporting sustainability

Uli: transformative capacity / institutional adaptability

– How to make diversity symmetrically operational for all salient aspects?

address Jochen’s point on levels of aggregation and system delineation

Some Key Questions

– What is diversity?

– Which things to diversify?

– How should we articulate diversity with other portfolio properties?

Towards a Systematic Approach

Philip: trade-offs

Marko: 7 functions

Staffan: 5 (6) functions

Kornelia, Frans: portfolio interactions

Adrian: normative frameworks and pathway stability

Raimund: scale economies, sustainability performance, business criteria

Some Key Questions

– What is diversity?

Ambiguities inhibit practical policy attention

– ‘apple pie’ rhetoric is vulnerable to special pleading

– Which things to diversify?

– How should we articulate diversity with other portfolio properties?

– high profile, but circumscribed and surprisingly neglected

Towards a Systematic Approach

Key point: diversity is not a ‘free lunch’

– foregone benefits, standardisation, scale, transaction costs,

diminished accountability, reduced stability of transition paths

Highlights need for systematic framework for analysing ‘diversity’ …

Some Key Questions

increasing diversity

What is Diversity?

capital investments / research programmes / development strategiessocio-technical trajectories / strategic niches / transition portfolios

comprising mix of elementseg: coal, oil, gas, nuclear, wind

increasing diversity

balanceevenness in contributionseg: nuclear – Japan vs France

varietynumber of options in mixeg: Norway vs USA

disparitydegree of differences

eg: renewables vs fossil

What is Diversity?

VARIETY = the number of options in a portfolio

conventional framing in economics and policy

convenient proxy in absence of complex analysis, BUT

- partitioning of ‘techno-institutional options’?

eg: ‘biofuels’ – or biodiesel, bioethanol, biogas?

eg: Saviotti,, Metcalfe, Llerena, Kaufmann – much discussion at this workshop

Conventional Approaches to Technological Diversity

- when to start counting?

eg: one PV array? ten thousand? 0.1% of system?

- what about varying degrees of niche representation in transition portfolio?

eg: “90% / 5% / 5%” or “33% / 33% / 33%”

- what about the degree to which options are different from each other?

eg: biodiesel, bioethanol, biogas, CaTe PV, CIS PV, a-Si PV?

BALANCE = ‘evenness’ of option contributions

indices from ecology / information theory

readily applicable and comprehensive in scope

- does address problem of when to start counting (partitioning)

eg: one PV array? ten thousand? 0.1% of system?

eg: Stirling, 1994; DTI, 1995; 2006; Grubb, 2004; use Shannon (– i pi.ln pi )

Conventional Approaches to Technological Diversity

- does address varying degrees of representation in system

eg: “90% / 5% / 5%” or “33% / 33% / 33%” BUT:

- still raise questions over ‘when is it one option and when two’?

eg: ‘biofuels’ – or biodiesel, bioethanol, biogas?

- treat options as if they are all equally different from each other

eg: biodiesel, bioethanol, biogas, CaTe PV, CIS PV, a-Si PV?

DISPARITY = degree to which options are different

portfolio theory, taxonomy, cladistics

Portfolio methods are powerful for financial planning in firms, BUT:

- address differences entirely in terms of past experience

eg: MVPA (Markowitz, Lucas, Awerbuch); Lancaster, Weitzman

Conventional Approaches to Technological Diversity

- highly circumscribed in scope (eg: fuel prices)

- highly restrrictive assumptions (eg: normal probability distributions)

- assumes single objective characterisation of difference

- neglects variety

- neglects balance

need to address and explore variety, balance and disparity

ie: - number of options in the portfolio

- proportional representation of options in the portfolio

- degree to which options are different from each other

Towards a Complete, Integrated Diversity Concept

need to avoid sensitivities to arbitrary assumptions

eg: - when to start counting? when one option and when two?

- assumption that difference reduces to single parameter

- assumption uncertainty can be treated probabilistically

- assumption that past predicts future

- assumption that all data is normally distributed need symmetry on divergent contexts, views and dimensions

eg: - include wider economic, environmental and social criteria

- be flexible towards different values and priorities

For any given perspective on the appraisal of a given transition context

option differences can be seen as dimensions in ‘disparity space’

Constructing a General Diversity Heuristic

number of dimensions represent different aspects of option disparity

Constructing a General Diversity Heuristic

For any given perspective on the appraisal of a given transition context

option differences can be seen as dimensions in a ‘disparity space’

number of dimensions represent different aspects of option disparity

positions of options in space determined by any salient disparity attributes

eg: functions (Staffan / Marko)

features (Eva / Daniel)

capacities (Uli)

attributes (Fred)

In principle, this framework can address any perspective

on salient features of institutions, technologies, functions, networks or effects

Constructing a General Diversity Heuristic

For any given perspective on the appraisal of a given transition context

option differences can be seen as dimensions in a ‘disparity space’

Distances between pairs of options represent their mutual disparity (da,b )

a

b

c

From Disparity to Diversity

For any given perspective on the appraisal of a given transition context

option differences can be seen as dimensions in a ‘disparity space’

Distances between pairs of options represent their mutual disparity (da,b )

Disparity of a portfolio of options is given as a function of these pairwise distances

a

b

c

da,b db,c

and…

Variety and balance can be captured by weighting this by the product of the proportional importance in the system of each option in the pair (pi.pj) = .pi.pj ij (ij) dij

From Disparity to Diversity

For any given perspective on the appraisal of a given transition context

option differences can be seen as dimensions in ‘disparity space’

1 Scaling of variety: where variety = 1, = 0

Formal Conditions for a General ‘Diversity Heuristic’

2 Monotonicity of variety: for equal B / D; rises monotonically with V

3 Monotonicity of balance: for given V / D; rises monotonically with B

4 Monotonicity of disparity: for given V / B; rises monotonically with D

5 Scaling of disparity: where aggregate difference = 0; = 0

6 Open Accommodation: is symmetric to any perspective on disparity

8 Parsimony of Form: has few components and simple structure

9 Explicit Aggregation: allows explicit weightings on V, B and D

7 Robust to Partitioning: is insensitive to aggregation on taxonomy

10 Ready Articulation: can be incorporated in portfolio performance

Can be shown robustly to fulfill first eight quality criteria:

= ij dij.pi .pj

Some Properties of the New Diversity Heuristic

Can be shown robustly to fulfill first eight quality criteria:

Some Properties of the New Diversity Heuristic

´ = ij (dij) .(pi .pj)

9 Allows explicit aggregations of variety, balance, disparity

Can be shown robustly to fulfill first eight quality criteria:

Some Properties of the New Diversity Heuristic

´ = ij (dij) .(pi .pj)

9 Allows explicit aggregations of variety, balance, disparity

0 0

0 1

1 0

1 1

´

ij dij0

ij pi.pj

ij dij

ij dij.pi.pj

diversity property

variety

balance

disparity

diversity

Can be shown robustly to fulfill first eight quality criteria:

Some Properties of the New Diversity Heuristic

10 Additional single term allows heuristic attention to interactions

(eg: Kornelia, Frans)

= ij dij.iij.pi .pj

9 Allows explicit aggregations of variety, balance, disparity

Some Properties of the New Diversity Heuristic

75% CCGT 25% wind

PORTFOLIO A

Can be shown robustly to fulfill first eight quality criteria:

10 Additional single term allows heuristic attention to

interactions

= ij dij.iij.pi .pj

9 Allows explicit aggregations of variety, balance, disparity

Some Properties of the New Diversity Heuristic

75% CCGT 25% wind

PORTFOLIO A

+

+

Can be shown robustly to fulfill first eight quality criteria:

10 Additional single term allows heuristic attention to

interactions

= ij dij.iij.pi .pj

9 Allows explicit aggregations of variety, balance, disparity

Some Properties of the New Diversity Heuristic

75% CCGT 25% wind

PORTFOLIO A

+

+

75% nuclear 25% wind

PORTFOLIO B

value [ { A } ] > value [ { B } ]

Can be shown robustly to fulfill first eight quality criteria:

10 Additional single term allows heuristic attention to

interactions

= ij dij.iij.pi .pj

9 Allows explicit aggregations of variety, balance, disparity

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio relationships, system effects

ANALYSE PRIORITIESdiversity / performance

Mapping Diversity – Performance Relationships

– economicsfunctions ,

capacitiessustainability metrics

– Fred’s “facilitation of reflexive action”

– attributes dimensionstypesfunctions,

capacities– research programmes

innovation systems transition paths

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio relationships, system effects

ANALYSE PRIORITIESdiversity / performance

Mapping Diversity – Performance Relationships

Fred:

“embrace cognitive diversity”

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio relationships, system effects

ANALYSE PRIORITIESdiversity / performance

Mapping Diversity – Performance Relationships

DELIBERATE OVER

FINDINGS

divergent assumptions

and perspectives

Look for Pareto dominance in all possible portfolios

under each particular perspective (eg: for three options)

diversity

aggregate portfolio performance

(economic, functions, sustainability)

dominant portfolios lie on this boundary

Mapping Links between Diversity and Performance

Results of a Schematic Energy Example

Identifies ‘efficient frontier’ for each perspective

portfolio

contribution

not optimisation – each perspective treated separately

max max

performance diversity

but a ‘heuristic’ – allows exploration of divergent perspectives

Reflexivity on Diversity in Analysis and Policy

Per

spec

tive

X

Per

spec

tive

Y

Per

spec

tive

Z

HEURISTIC FRAMEWORK FOR ANALYSIS / CATALYST FOR REFLEXIVE DELIBERATION

Implications: towards ‘Transition Portfolios’?

• diversity is crucial in building & understanding technological transitions

- fosters innovation of a kind essential to radical infrastructure change- mitigates ‘autonomy’, ‘momentum’, ‘lock-in’ and ‘entrapment’- hedges ignorance, so helps promote resilience and precaution- accommodates divergent social interests and values

• diversity is not a ‘free lunch’, but analysis remains strangely neglected

therefore vulnerable to incoherent, rhetorical and expedient arguments

• ‘variety’, ‘balance’, ‘disparity’ are necessary properties of diversity

crucial issue is how to articulate these properties and strike a balance

between diversity and other aspects of science and technology portfolios

• heuristic framework for explanatory analysis and normative engagement

allows open, reflexive deliberation – engaging different perspectives• aids move from managerial view: unitary niches, single transition paths

towards reflexive, robust, accountable politics of transition portfolios

ANNEXES

functional application

Social, cultural, economic, historic and geographic studies converge

Diversity and Innovation

easily shown to fulfill first eight quality criteria:

= ij dij.pi .pj

Some Properties of the New Diversity Heuristic

1: if variety = 1; = 0

2: rises monotonically with variety

3: rises monotonically with balance

4: rises monotonically with disparity

5: if disparity = 0; = 0

6: accommodates any perspective on disparity

8: is simple and parsimonious

= ij dij.pi .pj

in particular: takes full account of disparity (criterion 9)

Easily shown to fulfill first eight quality criteria

Some Properties of the New Diversity Heuristic

coal gas wind coal gas wind

PORTFOLIO A PORTFOLIO B

70% 5% 25% 70% 25% 5%

{ A } > { B }

in particular: takes full account of disparity (criterion 9)

= ij dij.pi .pj

fulfils basic quality criteria outlined earlier

Some Properties of the New Diversity Heuristic

in particular: takes full account of disparity (criterion 9)

= ij dij.pi .pj

fulfils basic quality criteria outlined earlier

Some Properties of the New Diversity Heuristic

allows exploration of different weights on variety, balance, disparity

´ = ij (dij) .(pi .pj)

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

Vaggregate = i ri.pi

where:

Vaggregate = value of aggregate

performance of individualoptions

ri = value of

performance of ith option

pi =proportional representation of ith option

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

{i x i matrix}yields set of (i - 1)2 / 2

disparity distances (dij)

where:

i = number of options in portfolio

dij = disparity distance

between options i and j

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

{i x i matrix}yields set of (i - 1)2 / 2

Interaction multipliers (iij)

where:

i = number of options in portfolio

iij =interaction multiplier for options i and j

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

ANALYSE TRADE-OFFS

diversity / performance

Mapping Diversity-Performance Relationships

total value of value of aggregate value due to system

portfolio = performance of + interactions and

performance individual options portfolio diversity

Vportfolio = i ri.pi + . ij dij.iij.pi .pj

where: Vportfolio = total value of portfolio performance

ri = value of performance of ith option

pi = proportional reliance on ith option

= marginal value of portfolio diversity

dij = disparity distance between options i and j

iij = interaction multiplier for options i and j

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

APPRAISE PERFORMANCEtechnology options / policy criteria

CHARACTERISE DISPARITYinstitutions / functions / technologies

DEFINE INTERACTIONSportfolio effects

ANALYSE TRADE-OFFSdiversity / performance

Mapping Diversity-Performance Relationships

DELIBERATE OVER

FINDINGS

divergent assumptions

and perspectives

functional application technological system

Social, cultural, economic, historic and geographic studies converge

Diversity and Innovation

functional application technological system

Social, cultural, economic, historic and geographic studies converge

Diversity and Innovation

functional application technological system institutional context

Social, cultural, economic, historic and geographic studies converge

Diversity and Innovation

functional application technological system institutional context

Social, cultural, economic, historic and geographic studies converge

Diversity and Innovation

functional application technological system institutional context

Social, cultural, economic, historic and geographic studies converge

Diversity and Innovation

densely connected network

homogeneous connectivity

impedes innovation

functional application technological system institutional context

after Grabher and Stark (1997)

Diversity and Innovation

densely connected network compartmentalised networks

homogeneous connectivity

impedes innovation

unconnected subsystems

impedes innovation

functional application technological system institutional context

after Grabher and Stark (1997)

Diversity and Innovation

densely connected network loosely coupled networks

homogeneous connectivity

impedes innovation

functional application technological system institutional context

heterogeneous connectivity

fosters innovation

after Grabher and Stark (1997)

compartmentalised networks

unconnected subsystems

impedes innovation

Diversity and Innovation

loosely coupled networks

heterogeneous connectivity

fosters innovation

technological and institutional diversity

helps foster ‘loosely coupled’ social networks

which promote more robust innovation

Diversity and Innovation

space of technologicalpossibilities

time

Social studies, philosophy, history and economics paint common picture

Diversity and Lock-in

space of technologicalpossibilities

time

Social studies, philosophy, history and economics paint common picture

Diversity and Lock-in

Technologies can present different equally-viable paths

space of technologicalpossibilities

time

Social studies, philosophy, history and economics paint common picture

Diversity and Lock-in

space of technologicalpossibilities

time

Technologies can present different equally-viable paths

but, in practice, reduced by momentum, autonomy, entrapment and lock-in

Social studies, philosophy, history and economics paint common picture

Diversity and Lock-in

space of technologicalpossibilities

time

Technologies can present different equally-viable paths

diversity mitigates lock-in and so enhances deliberate reflection and learning

Social studies, philosophy, history and economics paint common picture

for society: raises key questions over power, agency and choice

Diversity and Lock-in

knowledge aboutoutcomes

knowledge about

not problematic

problematic

outcomesnot problematic problematic

RISK

UNCERTAINTY

AMBIGUITY

IGNORANCE

Diversity, Ambiguity and Ignorance

engineering failureknown epidemicstransport safety

‘human element’global climate changeunassessed chemicals

“apples and oranges”landscape / emissions / safety

definition of GM ‘harm’

surprises like: BSE,CFCs

endocrine disruption

probabilities

Risk assessment limits are practical, methodological and theoretical Arrow’s

Nobel Prize shows ‘sound scientific’ policy is an oxymoron !

not problematic

problematic

not problematic problematic

RISK

UNCERTAINTY

AMBIGUITY

IGNORANCE

INCERTITUDE

knowledge aboutprobabilities

knowledge aboutoutcomes

Diversity, Ambiguity and Ignorance

knowledge aboutprobabilities

not problematic

problematic

knowledge aboutoutcomes

not problematic problematic

RISK

UNCERTAINTY IGNORANCE

risk assessmentdecision analysiscost-benefit analysis

‘rules of thumb’sensitivity analysisscenario analysis

deliberation, negotiationcitizen participation

‘mapping’ approachesplural/conditional outcomes

horizon scanningresearch and monitoring

‘social learning’flexibility, resilience, robustness

AMBIGUITY

Diversity addresses fundamental limits to analysis and deliberation accommodates ambiguities and hedges against ignorance

Diversity, Ambiguity and Ignorance

SECURITY OF SUPPLY IS MORE THAN DIVERSITY

SECURITY DIVERSITY

- control

- interdependence

- self reliance

- planning

- efficiency

- capacity, stocks

- resilience, flexibility

- fuels

- technologies

- producer regions

- industrial interests

- supply and trade

- infrastructures

- hedge ignorance

- foster innovation

- promote competition

- accommodate values

- mitigate ‘lock-in’

DIVERSITY IS MORE THAN SECURITY OF SUPPLY

Diversity in Energy Policy

Sustainability

maintain system functions over

long termsociety, economy,

environment

stability

againstendogenous disruptions

resilience

againstexogenous

shocks

robustness

under externalshifts

The Properties of Sustainability

durability

under internal change

Diversity in this Workshop

Title key theme: ”managing variation”

Jochen ”important to maintain and enhance variety”

Staffan ”imperative to pursue in parallel varied technology clusters”

Fred ”need to focus on variety creation and selection”

Kornelia ”need to address three types of variety”

Harald ”diversity relates to institutional incoherence and adaptability”

Raimund ”diversity as a means to selection of superior products”

But many questions over meanings and contexts?

DISPARITY = degree to which options are different

complex taxonomic indices

focus directly on concepts of difference BUT:

- assumes single objective characterisation of difference

eg: Weitzmann Function maxiS { DW(S \ i) + dW(i, S \ i) } (also Lancaster)

Conventional Approaches to Technological Diversity

- assumes universally ‘rooted directed tree’ (utltrametric distances)

- neglects variety

- neglects balance