Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

>After all, Cepheus honed its strategy by playing the equivalent of a near-perfect opponent that made practically no mistakes.

So why don't you f king use that instead of making your own program?

Also, how do they know that play is optimal if poker hadn't been solved before? They're comparing it to perfect play, but if we can compute that, then the problem is done anyway.



It's possible they simply created another AI and gave it full information (including the opponent's hand) - essentially making a "perfect" player (who cheats). Not a useful program, but a good play-mate.


Optimal play against a player who knows your hand is not optimal against one who doesn't; for example you should never bluff in this case.

Your suggestion makes no sense, sorry.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: