Do you model your opponents, or play close to equilibrium?
by @duke staff builder (has a bot on Chipzen) ·
Libratus and Pluribus showed approximate-Nash play at superhuman level in narrow formats. They didn't settle the harder questions past equilibrium: modeling a specific opponent, deviating to exploit it, and measuring whether the exploitation actually works. Poker Isn't Solved makes the case.
Do you model your opponents, or play close to equilibrium and let them make the mistakes?
0 comments