December 3, 2014AISC-CODISCO 2014, revised Nov From Agent-based models to network analysis (and return): the policy-making perspective Magda Fontana Pietro Terna University of Torino retired professor of the University of Torino

December 3, 2014AISC-CODISCO 2014, revised Nov ____________________________ A few premises ____________________________

December 3, 2014AISC-CODISCO 2014, revised Nov An important perspective use of Agent-based models (ABMs) is that of being employed as tools to support decision systems in policy-making, in the complex systems framework. Such models can be usefully employed at two different levels: to help in deciding (policy-maker level) and to empower the capabilities of people in evaluating the effectiveness of policies (citizen level). As a consequence, the class of ABMs for policymaking needs to be both quite simple in its structure and highly sophisticated in its outcomes.

December 3, 2014AISC-CODISCO 2014, revised Nov The pursuing of simplicity and sophistication can be made more efficacious by applying network analysis to the emergent results. The consequences of choices and decisions and their effects on society, and on its organization, are both relevant.

December 3, 2014AISC-CODISCO 2014, revised Nov Considering together the agent-based and network techniques, we have a further important possibility. Being easier to have network data (i.e., social network data) than detailed behavioral individual information, we can try to understand the links between the dynamic changes of the networks emerging from agent-based models and the behavior of the agents. As we understand these links, we can apply them to actual networks, to make guesses about the content of the behavioral black boxes of real-world agents.

December 3, 2014AISC-CODISCO 2014, revised Nov ____________________________ How to generate emerging networks to experiment with them first step ____________________________

December 3, 2014AISC-CODISCO 2014, revised Nov recipeWorld is an agent-based model that simulates the emergence of network out of decentralized autonomous interaction. The rationale behind it is to offer a few hints to find a framework and a grammar that are flexible and straightforward enough to encompass the widest possible range of purposeful and socially meaningful individual and organizational behavior.

December 3, 2014AISC-CODISCO 2014, revised Nov Recipes are coded as strings of numbers – their components. Each number (or, if we want, each label), is related to an act, a sub-routine, of the modeled action. For instance: [ ] means: execute step 3, then execute step 1, then …

December 3, 2014AISC-CODISCO 2014, revised Nov Examples in different fields can be suggested: production, health-care scenarios, financial complex operations, opinion spreading, co-authorships, etc.

December 3, 2014AISC-CODISCO 2014, revised Nov Behind any step, we an imagine to have any arbitrary level of complicated actions.

December 3, 2014AISC-CODISCO 2014, revised Nov A simple NetLogo [ ] implementation at mples/g_productionWorld.nlogo

December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov Calculations are made using the new NW NetLogo extension

December 3, 2014AISC-CODISCO 2014, revised Nov Calculations are made using the new NW NetLogo extension

December 3, 2014AISC-CODISCO 2014, revised Nov A less simple SLAPP [ ] implementation keep the current SLAPP version and run the ‘production’ example

December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov What is making “special” this result is that, in this context, agents are activated (following their internal rules and capabilities) by the events. The network emerges as a side effect, as in the real world.

December 3, 2014AISC-CODISCO 2014, revised Nov ____________________________ How to generate emerging networks: A - actual entities, B - agents, C - emerging network second step ____________________________

e1 e3 e4 e2 A a1 a3 a4 a2 B a3 a1 a2 C … December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov ____________________________ How to generate emerging networks: now forget A and B and build C from D and try to go back to the agents (reverse engineering) third step ____________________________

e1 e3 e4 e2 A a1 a3 a4 a2 B a3 a1 a2 D ! … C December 3, 2014AISC-CODISCO 2014, revised Nov

e1 e3 e4 e2 A ? a1 a3 a4 a2 B !! a3 a1 a2 … D n1 n3 n4 n2 C December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov ____________________________ How to generate emerging networks: back from the dream to the currently possible activities ____________________________

December 3, 2014AISC-CODISCO 2014, revised Nov

December 3, 2014AISC-CODISCO 2014, revised Nov recipeWorld is currently a prototipe in NetLogo A previous implementation of the recipe idea (without the network side) already exists in Java Swarm; it is at A new version exists in SLAPP (Swarm-Like Agent Protocol in Python); SLAPP is at

December 3, 2014AISC-CODISCO 2014, revised Nov Thanks and