Even I know about Morphy and Fischer's Queen sacrifices. And I'm sure there are many dozens of others almost as well known and considered brilliant or hard to find, in the last hundred years. So i'm wondering whether the best computers have been asked their opinion of those moves. And wondering whether, if they picked a different one, it meant that the human's move wasn't so smart after all (as opposed to out thinking even the super computers).

Hello David!

The computers do indeed agree with the soundness of the beautiful sacrifices from past eras of chess in most cases. Sometimes, the sacrifice is the "only" best move. Sometimes, there may have been other options that were also clearly winning... but why not choose the most beautiful continuation? And more rarely, a sacrifice may not quite be "sound", but it still has enormous practical value since refuting the sacrifice may require perfect defense from an already emotionally rattled opponent. Good luck not panicking when Morphy himself offers you a "free piece"!

In poker terms, sometimes the sacrifice is indeed the "GTO solution" to the position. More rarely, a mad genius of the chess board such as Morphy, Tal, or Kasparov will knowingly offer an unclear or even unsound sacrifice because they understand that defending perfectly is usually much more difficult than attacking in chess (I found this out the hard way at my chess tournament last night). The latter can be thought of as an "exploitative" play. A computer may not love the move, but it may sink a human grandmaster 90+% of the time.

I'm sure they've refuted brilliancies too, finding some defensive resource a human wasn't able to find. The thing is, most of the brilliant human moves end up being semi-forced where after the fact with careful study it holds up. Otherwise it wouldn't survive decades as such a brilliant move. So, while a computer can calculate even better, many of these famous moves and combinations have already been subjected to a ton of calculation.

If you are talking about "run of the mill" brilliant moves then I'm sure computers can shred a bunch of them and not shred a bunch of them. For instance, if you take Wei Yei's somewhat recent "game of the century" or whatever it was called, he was not simply winning after sacrificing material and making quiet moves, his opponent still had to make mistakes. Similarly Ding Liren's recent masterpiece wasn't even winning until his opponent made some mistakes.

To put in other words, I'm not sure computer evaluation is super important for brilliancies. This was probably a more relevant topic ~10-15 years ago when checking over old games with strong computers was happening. At this point I don't think anyone really cares and even if they find mistakes, are generally impressed with old moves that hold up "pretty well."

Two particularly well written posts above me.

https://en.chessbase.com/post/comput...ongest-player-

One of the most well thought of books in chess history is by Bobby Fischer, called "My 60 Memorable Games". In 1995, it was re-edited with substantial amounts of computer analysis, much of which showed that Fischer had made errors or misjudgements. However, by and large, the majority showed that he simply played extremely well. (David - the story of this re-issue and the controversy that occurred would probably be of interest if you wanted to read about it, especially from the perspective of an author and publisher - http://www.chesshistory.com/winter/extra/fischer.html)

Kasparov's series "My Great Predecessors" did similarly, where he wrote 100-150 page profiles on all the previous world champions. The results were about what you'd expect - as the years went on, the computers found fewer and fewer mistakes.