You get no argument from me on your point #5 (at least your points about capitalism and economics). The first problem, here, and with your point about game theory, is that (non-rational choice) sociologists love DATA; and tend to eschew “clean” (or rational) social theory. And I must note, I and most sociologists find most instrumental variables pretty UNDERwhelming (for the reasons you stated); if not outright tautological. Moreover, and I might get beaten up for this, but economists have done a better job of explaining capitalism. And mostly because capitalism cannot be seen exclusively through the lens of the social milieu.

The second problem, and you are in a good position to change this, is that sociologists assume networks, social relationships, and other ontological considerations. While economists are more interested in the emergence of such networks (or the instrumental action that gives rise to them). And if one is inclined to assume, e.g., networks and social norms, there is really no time for explaining purposive or instrumental action. Also, the game-theoretic models of the emergence of norms simply takes a utility-maximizing individual, and over-socializes her into her environment. And in so doing, neglecting the ontology of how people come to act “purposefully” or “rationally.”

Getting along demands that scholars take a hard look at rationality and ontology.

The reality is that both sides need to be more two-factor about their approach…

That is to say, the kind of behaviours you can fit into game-theoretic models aren’t really the bit of “culture” that people have trouble modeling. There’s hundreds (at least) of economists grazing wild and free on game-theoretic norms of behaviour within the firm, but none of them have provided much explanatory power for behaviours which don’t fit into a rigid economic model. That’s because once transactions are difficult to measure (i.e. the economic value is difficult to estimate) then the major error factor in your game theoretic model is mapping between the behaviour and the model.

It’s also true that there are a lot of social psychologists who spend a lot of time repeating studies of relatively simple economic behaviour issues, which is regrettable, but most of the rest of organisation studies has moved on to worry about the bits that don’t fit easily into existing game-theoretic models.

Re your third point: I don’t think this issue was discussed in depth because the kinds of sophisticated modeling you’re talking about are fairly common in organizational sociology. Most panel level studies use fixed-effects or at least random effects. It’s common to account for endogeneity with econometric models, like two-stage least squares. None of this stuff is new, nor is it seen as controversial. Ethnography and fuzzy sets modeling are both much more controversial but possibly very illuminating in helping us to understand static and dynamic differences between organizations and their contexts. That said, I heard several people, including Howard Aldrich, calling for the use of even more sophisticated modeling. By that, I think he meant moving beyond even the simple fixed-effects models to open up the black box and figure out the causes of variance within the dummies.