Fabian Adelt Johannes Weyer SKIN 3 Workshop, Budapest, May 2014

Transcription

1 Fabian Adelt Johannes Weyer SKIN 3 Workshop, Budapest, May 2014

2 Content 1. The issue 2. Basic principles 3. Purpose and concept of SimCo 4. Simulation framework in detail 5. Results from previous work (SUMO-S) 2

3 The issue: Innovation policy Interventions into complex socio-technical systems Goals: stabilization: avoid breakdown, improve performance change an existing system Problem: where / how / when to intervene? Innovation policy needs knowledge about the functioning of complex socio-technical systems effects of interventions types of actors population 3

4 Design principles Model of sociological explanation (Hartmut Esser) James Coleman s bathtube 4

5 Design principles Model of sociological explanation (Hartmut Esser) James Coleman s bathtube logic of situation logic of aggregation logic of selectio n 5

6 Design principles Model of sociological explanation (Hartmut Esser) James Coleman s bathtube Two levels (and interdependencies between them) system dynamics (macro level) emergent phenomenon actors choices (micro level) decision making, based on subjective expectations (range of view, bounded rationality ) different types of actors: rational egoist, norm follower, emotional man etc. Impact of innovation policy and interventions on macro behaviour? 6

7 Agent Based Modeling and Simulation (ABMS) Agent State Transitions Rules Actions Behaviour Environment Information Structure Time Quelle: Van Dam et al. (2013:58) 7

8 Agent Based Modeling and Simulation (ABMS) Agent State Transitions Rules Actions Behaviour A micro-founded ABMS can implement a sociological explanation! Environment Information Structure Time Quelle: Van Dam et al. (2013:58) 8

9 general model of networked infrastructure system energy, transportation, information nodes, edges, agents, technologies micro-macro-interaction dynamic behaviour in time innovation policy / governance instruments of intervention management, control, regulation... 9

10 Agents tasks move in the network pursue individually rated goals visit (selected) nodes use resources and technologies stay within limits 10

11 Decision making based on sociological theory of action subjective expected utility actions: (individually) rated goals: probabilities: indicate how far action i serves goal j depend on state of the network and the actor s position possibly influenced by interventions Choose the action that fits your goals best! 11

12 Decision making based on sociological theory of action Example Type A reach green nodes Rated goals behave ecofriendly act moneysaving move fast SEU Actions direct path via gray node Probabilities?

13 Decision making based on sociological theory of action Example Rated goals Type A reach green nodes behave ecofriendly act moneysaving move fast SEU Actions direct path via gray node

14 Decision making based on sociological theory of action Example Rated goals Type B reach green nodes behave ecofriendly act moneysaving move fast SEU Actions direct path via gray node

15 Decision making based on sociological theory of action Example Type B reach green nodes Rated goals behave ecofriendly act moneysaving move fast Empirical data micro-foundation SEU Actions direct path via gray node

16 Nodes (intermediate) targets for agents lifetime costs/benefits (for agents, nodes and external ) multi-dimensional features memory / payoffs status rights-of-use (technology) limits (short and long term) 16

17 Edges directed paths between nodes lifetime costs/benefits (for agents, edge and external ) multi-dimensional features memory / payoffs rights-of-use (Technology) limits (short and long term) 17

18 Technologies Needed to pass nodes and edges lifetime purchase price costs/benefits-factor multi-dimensional absolute costs/benefits depend on node/edge and technology 18

19 Aggregating single actions Complex systems huge number of actors Actors influencing their environment and each other Emergent macro behaviour macro result may differ from a (individually or politically) desired behaviour Control is an intentional intervention into a social or sociotechnical system aiming at changing it in an intentional way. (Adelt et al. 2014) by changing boundary conditions parameters like rights-of-use, costs, limits without knowing the result in advance 19

20 Innovation policy / How to change the system new technologies or change comparative costs of existing technologies (by tax changes, subsidies etc.) change the topology of the net create incentives electric vehicles vs. SUVs add technologies change their parameters build new streets, or streets usable for some technologies only bike-parking facilities, charging stations for electric vehicles change node parameters 20

21 To which extent is the system affected (or even changed) by interventions? Depends on the type of intervention and the strategies of the different types of actors. 21

22 Simulation of Urban Mobility (SUMO-S) simulation of a traffic guidance system modes of governance strong control soft control self-coordination Adelt/Weyer/Fink, 2014: Controlling complex systems. In: Ergonomics. Adelt/Weyer/Fink, 2014: Steuerung komplexer Systeme. In: Soziale Welt. 22

23 System stability (low numbers are better) mode of governance Mix0 (only obedient) mix of drivers Mix1 (realistic) Mix2 (more obedient) Mix3 (more singleminded) mean values of rows (Mix1 to Mix3) strong control 5,858 8,011 9,340 9,308 8,886 soft control 6,301 10,187 5,820 9,591 8,533 selfcoordination mean value of columns 8,500 10,064 10,181 10,603 10,283 6,886 9,421 8,447 9,834 23

24 Conclusion: SimCo abstract representation of infrastructure systems ABMS with a sociological micro-foundation actor types, empirical foundation dynamic, emergent behaviour examine the effects of interventions in order to gain more knowledge about the functioning of complex systems depends on actors reaction measure micro and macro performance 24