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

I really am curious about the Nash equilibrium solution. I assume that as a commenter has mentioned, for the guesser it involves returning random numbers near the binary search. But I’m curious if for the picker it involves a uniform or non uniform initial distribution?? I’m sure someone on HN knows/can explain?


I've taken a stab at analyzing the game for 3, 4, and 5 numbers here: https://quuxplusone.github.io/blog/2024/09/04/the-game-is-fl...

There's obviously a huge gap between the 5-number game and the 100-number game; it's possible that the best mixed strategies settle down as the number of choices gets larger, or (for all I know) it's possible that the best mixed strategies get crazier and crazier. I'd love a ping if anyone does any real exploration of the 6-, 7-, etc.-number games.


I don't see how Ballmer randomly guessing a number and candidate doing binary search would NOT be Nash Equilibrium.

For anything else, the candidate has a strategy of "guessing a trick number", and Ballmer has a strategy of "not choosing a trick number".

The candidate can't force Ballmer to try to be tricky.


"Candidate always starts by guessing 50" can't possibly be part of a Nash equilibrium, since then Ballmer would never ever choose 50 as his secret number. (And indeed, he says as much in the linked video.) And if he'll never choose 50, then it's obviously silly for the candidate to waste a guess on it... and so on. Nash equilibria are usually (always?) located at mixed strategies.




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

Search: