Actions

Difference between revisions of "Smith abs"

From Santa Fe Institute Events Wiki

 
(oops)
 
Line 12: Line 12:
 
''Good social science as good physics?''
 
''Good social science as good physics?''
  
The social sciences – the so-called "softer" sciences – have never matched the precision, and especially the predictive power, of the "harder" physical sciences. This can be explained, perhaps, by the tremendous psychological complexity of individual people relative to the simple regularities of the matter (atoms, molecules, etc.) making up physical systems. The existence of minds, individual free will and our unparalleled capacity for reason seem to beset the social sciences with particular difficulties. Consequently, as many social scientists and philosophers have argued, the social and physical sciences seem to be essentially different, and, perhaps, must always remain so.  
+
In attempting to benefit from lessons learned in the quantitative natural sciences, social science all too often borrows specific models, rather than the more fundamental style of reasoning by which those models were originally discovered and validated in their home domains. It could be argued, however, that in physics the validation of models is a much less important motivation for the quantitative treatment of strong data regularities than the discovery, navigation, and validation of _abstractions_ which carry quantitative consequences. In this talk I will argue -- with a few working examples -- that the methods of physics can be usefully applied to social phenomena in a way that transcends any models that arise from them.
  
Nevertheless, research over the past two or three decades suggests that this conclusion is certainly premature and quite probably mistaken; that social science can indeed be done along the lines of physical science, and that something akin to a "social physics" may be possible. Success requires several components. First, social physics requries a sound, empirically-rooted picture of the typical behaviors of individual people, grounded in biology. Second, it must exploit analytical and computational tools to explore the kinds of collective patterns and phenomena likely to emerge when many people interact with one another. Third, and perhaps most importantly, it must test theoretical insight (from agent-based models, etc) against empirical reality at the collective level with hard data.  
+
The distinction between behavior and institutions, and the interactions between the two, offer some easy starting points where physicists can both contribute with familiar methods, and begin to enrich the frameworks of abstraction within which social dynamics are conceptualized. Abstracted theories of pure institutional process, unfortunately known as "zero-intelligence models", can not only enrich individual-based frameworks such as neoclassical micro-economic theory; they can provide better-focused questions about individual behavior as well. Strong abstractions such as dimensional analysis, which turn out to be fundamental to the interpretation of quantitative systems, can take on new dimensions where they are found to apply in the social domain.  I will show how dimensional analysis and zero-intelligence modeling of financial markets have enabled us to identify quite strong data regularities that are not tied to behavior, and thereby to isolate other strong regularities which can be targets of either focused behavioral science, or perhaps of richer abstractions of information flow mediated by market institutions.
   
 
I will review recent empirical work in social psychology, experimental economics and evolutionary psychology which suggests that while people are indeed sometimes complex, their behavior as individuals is often surprisingly simple. This suggests that the real root of complexity in the social world is often not individual human complexity, but collective complexity that emerges spontaneously out of the interactions of many people. I will also review a number of computational efforts to model and understand the nature of these emergent collective patterns. All of this work, which grounds the study of social reality in the biological and computational sciences, promises a new and much more powerful social science – perhaps even deserving the name "social physics" -- in the near future. (However, from experience, it seems to me the term "social physics" often leads to serious misunderstandings and is probably best avoided.)
 

Latest revision as of 22:16, 2 January 2008

Workshop Navigation


Is There a Physics of Society? January 10-12, 2008, Santa Fe NM

Organizers: Michelle Girvan (University of Maryland) and Aaron Clauset (Santa Fe Institute)

Thursday, January 10, 2008

9:50 - 10:30 D. Eric Smith (homepage)

Good social science as good physics?

In attempting to benefit from lessons learned in the quantitative natural sciences, social science all too often borrows specific models, rather than the more fundamental style of reasoning by which those models were originally discovered and validated in their home domains. It could be argued, however, that in physics the validation of models is a much less important motivation for the quantitative treatment of strong data regularities than the discovery, navigation, and validation of _abstractions_ which carry quantitative consequences. In this talk I will argue -- with a few working examples -- that the methods of physics can be usefully applied to social phenomena in a way that transcends any models that arise from them.

The distinction between behavior and institutions, and the interactions between the two, offer some easy starting points where physicists can both contribute with familiar methods, and begin to enrich the frameworks of abstraction within which social dynamics are conceptualized. Abstracted theories of pure institutional process, unfortunately known as "zero-intelligence models", can not only enrich individual-based frameworks such as neoclassical micro-economic theory; they can provide better-focused questions about individual behavior as well. Strong abstractions such as dimensional analysis, which turn out to be fundamental to the interpretation of quantitative systems, can take on new dimensions where they are found to apply in the social domain. I will show how dimensional analysis and zero-intelligence modeling of financial markets have enabled us to identify quite strong data regularities that are not tied to behavior, and thereby to isolate other strong regularities which can be targets of either focused behavioral science, or perhaps of richer abstractions of information flow mediated by market institutions.