Path Dependence and the Problem of Counterfactuals in Social Systems

I am intrigued by the title - tell me more - john mahoney

---

I may have covered much of this in the Geopolitics link, but I'll try and elaborate below.

Path dependencies are cases where certain outcomes in the state space are conditional, and unless some precursor set of events occurs, they cannot be achieved. In economic systems, these have gone hand-in-hand with the study of increasing returns, and the phenomenon of "lock-in" of technological standards or other institutions. Of interest to me, is whether or not outcomes in social systems are structurally determined, or contingent on micro-level dynamics and choices.

There has been a lot of work on this in economics, looking at how prices emerge and stabilize. It turns out that even in cases where markets clear, the final price of a good will can vary based on structure of buyers and sellers. Imagine a population buyers and sellers all of whom are slightly different in terms of their preferences. Buyers all want to purchase strawberries, but each is willing to pay a slightly different maximum price. Sellers all have strawberries to sell, but each is willing to sell at a different minimum price. It turns out that the final price the system actually achieves is contingent on who buys and sells with who. Thus, some final average prices can only be achieved based on a specific, or subset, of histories.

Path dependence is important because it means that the history of a system is part of the system's future trajectory (although it need not determine it). Again, we all use the QWERTY keyboard because it is an artifact of the mechanical typewriter. A system with a different history, say no typewriters, and we don't have QWERTY for use on digital computers (the associated phenomenon of increasing returns and lock-in seeks to explain why we still have QWERTY despite that other technologies may be better for digital computer interfaces).

A closely related phenomenon to path dependence is the problem of counterfactuals. Counterfactuals are problems where one tries to figure out the effects of changing a fact. For example, could the Cold War could have occurred had nuclear weapons not been invented or demonstrated? Because history provides us with a sample size of 1, we have no information about this alternative world. Counterfactuals are especially important to social science, because they help us think through what really matters, but because we have no true repeatability or laboratory we are stuck with mental or computational simulation as the tool of choice.

I have not been able to get through all of the text on the subject, but essentially we alter the facts of a situation in two ways. Consider these two statements:

A) If Lee Harvey Oswald didn't kill Kennedy someone else did. B) If Lee Harvey Oswald didn't kill Kennedy someone else would have.

I think we largely know how to deal with A through traditional research tools. The problem is statement B. How do we investigate it, and what kinds of properties or outcomes in social systems should we regard as inevitable, and what are contingent?

I'm interested in both path dependence and counterfactuals from methodological perspectives and for philosophical/epistemological reasons.