As we left matters in the first part of this conclusion, we were wondering how much in this picture is causal and how much is simply opportunistic – or maybe some of both. As an example, Roman military technology dominated the world for 500-600 years. But initially their technology was notably inferior to that of their principal rivals, the Carthaginians – shouldn’t they then have lost? As it turned out, the Romans simply cared more about destroying Carthage than the Carthaginians cared about destroying Rome. They were able to improvise technologies nearly as good, come up with leadership nearly as good, and use their resource superiority and greater commitment to grind Carthage down. Pretty much without serious opposition, Rome then used this technology to take over the rest of the Mediterranean world during the next two hundred years, and then sat on it for two hundred after that. Eventually, the string ran out, and Rome’s lack of adaptability in the face of new barbarian tactics then proved fatal.
Jared Diamond’s Guns, Germs, and Steel proposed that the Europeans’ dominant control of his three resources after about 1600 made it virtually inevitable that they would wind up dominating the world, as long as their monopoly on these key technological advantages lasted. Europeans equipped with their technologies in the 16th century and following did in fact dominate the less technological peoples they came into contact with – although it’s not clear (a) why they were the only ones with the technologies (the Chinese could easily have been ahead of them, but weren’t, and the Japanese deliberately rejected them) or (b) why the Europeans suddenly felt it to be a good idea to go around looking afar for others to dominate when they did (they hadn’t, for at least a thousand years). So the answer isn’t just technological, but far more socio-technically complicated.
Let’s go back to the central question with which we began – that is, how to assess causality in social behavior. We’ve extensively reviewed what “science” knows about assessing causality in human micro-relations; but even carefully hedged around, these methods are approximate. And they provide only the haziest possible generalization to the macro-relations – things that happen on the level of organizations and societies – that really interest us and might make a real difference in our human condition. At those levels causality is, oddly enough, both far too easy and far too complex to determine.
Parker’s descriptions of the hideous weather conditions in the 17th century that made life generally horrible and vastly upset societies are compelling, and strongly suggest that large-scale social changes are only to be expected from such strong stimuli. But the nature of those changes isn’t necessarily predictable from the fact of change itself. For example, it’s far from clear just why the almost universal response was to start whacking on other people, rather than cooperating to alleviate conditions. Indeed, Parker’s discussion of the far more adaptive and cooperative response to equally horrible climate conditions in Tokugawa Japan is clear evidence that there were alternatives to cut-throat conflict. It’s simply that in most places imagination and/or authority failed to generate any. But what kind of model can incorporate imagination as a variable?
Recently, mathematicians have suggested that extremely complex chaos/complexity models can be decomposed into sets of simpler models. Specifically, it’s been suggested that (on a rather different scale of things), a model previously applied to the universe shortly after the Big Bang might be broken down into one based on special relativity and one based on quantum mechanics – two theories that are often considered to be incompatible (“Removing Complexity Layers from the Universe’s Creation“). This might suggest that a similar decomposition of complexity/chaos models of social behavior might be possible, assuming (a) that such models exist and (b) that the mathematics can be performed. Or (c) that one could find behavioral scientists well equipped enough mathematically to attempt the task.
There’s another problem, of course. Most “modeling” of human behavior systems isn’t really intended to provide an overall predictive framework for large-scale social change. Large-scale attempts at theory, like Talcott Parsons’ “general theory of action” or Keynes’ “general theory of employment, interest, and money”, are in most respects not predictive of individual events, but rather provide a framework within which various forces and conditions interact. The theories provide key variables and suggest testable relationships that might be explored, but really operate more at the level of an extended metaphor than “theory” in the sense that, say, physical scientists might recognize it.
The vast bulk of behavioral science modeling is ex post facto, simplistic, oriented retroactively rather than proactively, and aimed at finding “levers” – points of possible intervention – rather than real causes. Levers have value but are easily misinterpreted out of context. Most model-builders are in fact desperately afraid of predictions that might prove them wrong.