Imagine my surprise today when I received the following e-mail:
http://magma.maths.usyd.edu.au/~watkins/LOSING_CHESS/email.htmlWoohoo! Antichess has finally been solved!
people keep saying antichess has been solved but itss mainly e3
g3, nh3, na3, g4 and some others arent even close to being solved
^^I received the same email.:)
@Dragon-Lord, it is solved in the sense that the outcome is determined from the opening position with perfect play from both sides. White having the possibility of making a suboptimal opening move does not invalidate the solution.
^Yep that's several years of hard work.
@Dragon-Lord Specifically it has been "weakly solved" which is what
@Kranix describes. What you seem to be describing is a "strong" solution. Wikipedia has an article on this here:
en.wikipedia.org/wiki/Solved_gameWhat this means is that, if a computer has access to Watkin's output, it will be able to achieve a 100% win rate as white.
Unfortunately the same will be true for a cheater on lichess with access to such a computer...
^Yep but Solution-related cheating is very easy to detect. Humans can remember the lines up to a point but eventually they began to play on their own. On the other hand, cheaters have to follow the solution completely.
Hence a match with the solution up to Move 15 and sometimes even 20 can be OK in some special lines but a full 100% match including in endgame tablebase up to Move 25 is not.
Besides, human players who work hard on memorizing the lines will actually be good in other situations (they will use their knowledge and apply it to other positions!). If someone has memorized a bunch of lines after 1. e3, he or she shouldn't completely crumble when he's playing vs. 1. g3 or something.
^Yes. I know people who follow the lines till Move 10 and still lose games. There is a limit of memorization. Lichess simply needs to ask Watkins for permission to incorporate the solution into its own database. Then we can offer it here (if we don't others will do it anyway to cheat) and use it to detect cheating.
tl;dr version: You can follow the lines a bit and still lose. Humans just can not remember Watkins' 3.2GB of lines plus the endgame tablebase. If you do "remember" it you are cheating.
Oh yeah, definitely! Great idea! Lichess should offer to host Watkin's proof.
