Complex Social Systems: a guided tour to concepts and methods

Transcription

1 Complex Social Systems: a guided tour to concepts and methods Overview Presentation Martin Hilbert (Dr.; Ph.D.) MartinHilbert[at]gmail.com

2 Today s questions I. What are the characteristics of complex systems? II. What are some of the tools to analyze complex systems?

3 Definitions of Complexity Seth Lloyd. Measures of complexity: a nonexhaustive list. IEEE Control Systems. Aug;21(4): definitions of complexity: Three questions that researchers frequently ask to quantify the complexity of the thing: 1. How hard is it to describe? [entropy, minimum description length, etc.] 2. How hard is it to create? [computational complexity, cost, etc.] 3. What is its degree of organization? [tree subgraph diversity, mutual information, etc.]

4 Historical context of Complexity Weaver W. Science and Complexity. American Scientist. 1948;36: Problems of simplicity: science before 1900 was largely concerned with two-variable problems temperature & pressure; population & time; production & trade, etc. 2. Problems of disorganized complexity: subsequent to 1900 scientists developed powerful techniques of probability theory and of statistical mechanics each of the many variables has a behavior which is individually erratic. billiard balls & air molecules; normal distributions; etc. 3. Problems of organized complexity: dealing simultaneously with a sizable number of factors which are interrelated into an organic whole cannot be handled with the statistical techniques so effective in describing average behavior Science must, over the next 50 years, learn to deal with these problems of organized complexity

5

6 Characteristics of Complexity Connected Interdependent Diverse Adaptive Path-dependent Emergent

7 Characteristics of Complexity Connected Interdependent Diverse Adaptive Path-dependent Emergent

8

9

10 Source: Christakis and Fowler, 2007 Social Networks

11 Different network structures lead to different diffusion curves % of U.S. households General tendency: increased social connectedness leads to increased speed of diffusion

12

13 Innovation & Industry Social = network 9/11 Terrorist Network Friendship by race Food Network Source: Powell et.al. (2010);

14

15

16 Characteristics of Complexity Connected: social network analysis Traditional database of attributes Gender Location Income Educat. Jorge M Urban 700 Tertiary Maria F Urban 500 Second. Juan M Rural 300 Primary Magda F Rural Network database of links Jorge Maria Juan Magda Jorge Self --- Maria Self --- Juan Self --- Magda --- Self Jorge Maria Jorge Maria Juan Magda Juan Magda

17 Summary of characteristics and (selected) methodological tools of Complexity Connected: social network analysis Interdependent: dancing landscapes Diverse Adaptive Path-dependent Emergent

18 Scientific management Mount Fuji (Osorno, Villarica) landscapes (from Taylorism to Fordism) Amount of coal lifted in 8 hours Interdependency between coal and shovel size 21 pounds Size of shovel

19 Performance / fitness Rugged Landscape: engineering of internal trade-offs Variable costs Fixed costs

20 Wave-like nature of punctuated equilibria (discovering new peaks) Evolution of Aircraft Technology Perfromance airspeed propeller mph airspeed jet mps 5% growth % growth 2% growth Continuous technological progress Disruptive technological progress 50 Disruptive technological progress Source: Devendra Sahal, Patterns of Technological Innovation, 1981

21 Dancing/ coupled/ tunably rugged Landscapes: interdependencies The NK model N: number of components K: degree of interdependency

22 Dancing business strategy / public policy landscapes Interdependencies among internal choices (trade-offs) makes landscape rugged Interdependencies with external agents makes landscape dance!

23 Exploration and Exploitation & dancing landscapes Fitness individual agents balance the necessity to explore and exploit, producing complexity as a result (S.E. Page, 2009)

24 Summary of characteristics and (selected) methodological tools of Complexity Connected: social network analysis Interdependent: dancing landscapes Diverse: entropy and differences Adaptive Path-dependent Emergent

25 Claude E. Shannon (1948) A Mathematical Theory of Communication, Bell System Technical Journal, Vol. 27, pp , From Entropy

26 to Variance Ø = 4.2 Ø = 2.5 Ø = 1.8 Ø = 1.8 Ø = 1.5 Ø = 1.2

27

28 Summary of characteristics and (selected) methodological tools of Complexity Connected: social network analysis Interdependent: dancing landscapes Diverse: entropy and differences Adaptive: information processing Path-dependent Emergent

29

30

31 Summary of characteristics and (selected) methodological tools of Complexity Connected: social network analysis Interdependent: dancing landscapes Diverse: entropy and differences Adaptive: information processing Path-dependent: dynamics & chaos Emergent

32

33 Chaos => change over time depends on initial conditions! Henri Poincaré ( ) 0.1 =>> 0.01 =>> =>> If we knew exactly the laws of nature and the situation of the universe at the initial moment, we could predict exactly the situation of that same universe at a succeeding moment. But even if it were the case that the natural laws had no longer any secret for us, we could still only know the initial situation approximately. it may happen that small differences in the initial conditions produce very great ones in the final phenomena. A small error in the former will produce an enormous error in the latter. Prediction becomes impossible, and we have the fortuitous phenomenon. (Science and Method, 1903)

34 Summary of characteristics and (selected) methodological tools of Complexity Connected: social network analysis Interdependent: dancing landscapes Diverse: entropy and differences Adaptive: information processing Path-dependent: dynamics & chaos Emergent: agent-based models

35

36 Schelling s segregation model Empirical data from Los Angeles, Milwaukee, Cincinnati, Omaha, & Kansas City show that the Schelling description of preferences is broadly correct but that the empirical curves are less regular than those posited by Schelling Clark (1991). Residential Preferences and Neighborhood Racial Segregation: A Test of the Schelling Segregation Model. Demography, 28(1), p agents: 51 % =>? 49 % =>? 33% =>? 26 % & 25 % =>? 75 % & 76 % =>? ; Total is different than sum of parts (total racists lead to mixed society ) Phase transitions Dependence on initial conditions Invariant Distribution

37 Alternative Approaches I cannot too strongly urge you to get the dimes and pennies and do it yourself there is nothing like tracing it through for yourself and seeing the thing work itself out. In an hour you can do it several times and experiment with different rules of behavior, sizes and shapes of boards, and (if you turn some of the coins heads and some tails) subgroups of dimes and pennies (p. 150) Schelling, T. C. (1971). Dynamic models of segregation. The Journal of Mathematical Sociology, 1(2),

38 Sugarscape Pareto sugar/income distribution emerges! Initial endowments: 1:2? 1:200? 199:200? Why? Physical landscape Rule: move to next open spot with most sugar and consume it Random genetic endowments: Cognitive ability (vision) Metabolism? Maximum age?? Initial location (random)??? Li, J. and Wilensky, U. (2009). NetLogo Sugarscape 1 Immediate Growback model. Center for Connected Learning and Computer-Based Modeling, Northwestern University. Wilensky, U. (1999). NetLogo. Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL.

39 just one of a couple of dozen of Sugarscapes Fine-tuned capability diversity Foresight and planning Seasons Migration patterns Off-springs and inheritance Sugar & Spice Trade and price Lending and interest rates Pollution of resource extraction Disease Cultural belonging Iain Weaver (2009) The Sugarscape;

40 Macro can be surprising Complex Challenges Micro can be surprising Schelling (1969). Models of Segregation. Am. Econ. Review, 59(2), p.488 Economists are familiar with systems that lead to aggregate results that the individual neither intends nor needs to be aware of, the results sometimes having no recognizable counterpart at the level of the individual. savings decisions cause depression or inflation The interplay of individual choices is a complex system with collective results that bear no close relation to the individual intent. Epstein & Axtell (1996). Growing Artificial Societies. Bradford; pp upon first exposure to these familiar social, or macroscopic, structures some people say, Yes, that looks familiar. But, I ve seen it before. What s the surprise? The surprise consists precisely in the emergence of familiar macrostructures from the bottom up from simple local rules that outwardly appear quite remote from the social, or collective, phenomena they generate. In short, it is not the emergent macroscopic object per se that is surprising, but the generative sufficiency of the simple local rules. Descriptive / bottom-up / micro => macro: Given (change in) individual rules, what is the global behavior? Interventionist / top-down / macro => micro: Given (desired) global behavior, what should be the individual rules?

41 more realistic / complex models of society: Source: Computational Modeling at Olin College.

42 Summary of characteristics and (selected) analytical tools of Complex Systems Connected: social network analysis Interdependent: dancing landscapes Diverse: entropy and differences Adaptive: information processing Path-dependent: dynamics & chaos Emergent: agent-based models