I would be interested in how it changes the model if you assume that everyone is a continually varying mix of Joker, Cooperator and Defector, with also a value

I would be interested in how it changes the model if you assume that everyone is a continually varying mix of Joker, Cooperator and Defector, with also a value for how impressionable/stubborn they are. Fixed roles and actions with a narrowly defined intents seem to be very common in these kind of thought experiments, mainly, I suspect, because it makes them easier to model. However I think it robs them of any real utility, and worse can give an illusion of insight that is entirely based on flawed assumptions.