Re:The neural network of the Stockfish chess engine
by magicalhippo on Wednesday January 13, 2021 @11:45AM
(#25764495)
I think I got a big hung up on the following from Wikipedia:
It says that if the game cannot end in a draw, then one of the two players must have a winning strategy (i.e. force a win).
It made me think it didn't apply in situations where the other player could force a draw.
Reading the linked article[1] though made it more clear what it's about, and I agree with what you said.
[1]: http://www.math.harvard.edu/~elkies/FS23j.03/zermelo.pdf
Maybe it's not clear to human players or current engines, but that does not mean it doesn't exist, and indeed the theorem mentioned above states that at least one player has an optimal strategy (either forcing a draw or winning). Obviously, that does not necessarily mean that we'll ever be able to design an engine that can compute this strategy in a reasonable time.
I just think that OP meant something other than the usual definition of "solved", maybe they meant that engines are plateauing, but as others have pointed out, there is also little evidence for that for now.