Cartesian Frames seems like the sort of frame in which it would be interesting to represent decision problems. E.g., you might be able to get a formalization of vertical decision problems as opposed to just horizontal ones (Prisoner’s Dilemma), there are probably nice abstractions around Agency here, and if you manage to incorporate probability measures you can start to represent EUMs broadly (and find a categorical semantics for oracle-accessing-agents?)
Epistemic note: in the process of reading [Cartesian Frames](https://arxiv.org/abs/2109.10996)
Cartesian Frames seems like the sort of frame in which it would be interesting to represent decision problems. E.g., you might be able to get a formalization of vertical decision problems as opposed to just horizontal ones (Prisoner’s Dilemma), there are probably nice abstractions around Agency here, and if you manage to incorporate probability measures you can start to represent EUMs broadly (and find a categorical semantics for oracle-accessing-agents?)
More specifically, it feels like you might be able to make progress on [The Ubiquitous Converse Lawvere Problem](https://www.alignmentforum.org/s/5WF3wmwvxX9TEbFXf/p/5bd75cc58225bf06703753b9) because something-something Nash equilibria in category theory implies some Kakutani ⇔ Lawvere correspondence?