I think we want a few AI difficulty ratings, an easy one that will intentionally make some mistakes so a learning player can still beat them most of the time, and then at least a hard one, but ideally a medium then a hard one. We can use the simulation's AI as a baseline for the hard AI.
We also want to when simulating be able to have AIs face each other, rather than re-implementing that AI in the simulation code. In order to prove out that the game is truly strategic, I would like it to be extremely likely (like >90%) chance that a hard AI would beat an easy AI.
I think we want a few AI difficulty ratings, an easy one that will intentionally make some mistakes so a learning player can still beat them most of the time, and then at least a hard one, but ideally a medium then a hard one. We can use the simulation's AI as a baseline for the hard AI.
We also want to when simulating be able to have AIs face each other, rather than re-implementing that AI in the simulation code. In order to prove out that the game is truly strategic, I would like it to be extremely likely (like >90%) chance that a hard AI would beat an easy AI.