I don't think the arbitrary solution can apply to real-world applications. A logarithmic debit from a players preferred payoff as a computation cost might make more sense applied to the real world.
At a point the cost of calculating an opponent's preferred payoff becomes too high relative to the perception of payoff itself. By way of example, this is actually how many litigation negotiations are resolved.
I think these limits on individual computation, if I may mix disciplines, manifest as emotions or beyond that detachment. And I think there's an opportunity for modeling real world games more accurately viewed from that perspective.
At a point the cost of calculating an opponent's preferred payoff becomes too high relative to the perception of payoff itself. By way of example, this is actually how many litigation negotiations are resolved.
I think these limits on individual computation, if I may mix disciplines, manifest as emotions or beyond that detachment. And I think there's an opportunity for modeling real world games more accurately viewed from that perspective.