The link is to a chessbase article speaking about a group of researchers who claim that that king's gambit is busted. While this was an April Fools prank, is this technically possible? With modern software(Stockfish 5, open-source at that), new software, and the fact that it has worked before, has there been any similar research? Have openings such as the Fried Liver Attack been disproved by a chess engine evaluation?

A small clarification. Not disproved by a GM who has some small support by an engine, but by a full on Rybka(Or other chess engine) analysis, like in the article. I am also interested in projects that are still running, or even have just started.

  • 4
    That Chessbase article is April Fool's joke, KG is definitely not busted. As for other openings, I doubt that any similar project occurred by now. Even if they actually tried it, they would fail in my opinion. – AlwaysLearningNewStuff Aug 21 '14 at 15:54
  • I honestly didn't know that. Should I let the question in case it gets an answer or should I delete it? – MikhailTal Aug 21 '14 at 16:10
  • 1
    @MikhailTal Chessbase has a post here about the KG April Fools' Day joke if you want to learn more: en.chessbase.com/post/the-chebase-april-fools-revisited It's very hard to declare an opening “busted” three half-moves into a game (1. e4 e5 2. f4), even with an engine, because the number of variations is still incredibly high at that point; finding a forced win in a short number of additional moves is unlikely. – Nick Aug 21 '14 at 16:59
  • What about forced variations, with lots of checks? Sych as the Fried Liver, Cochrane Gambit, etc, or a main line dragon with a full Yugoslav formation? – MikhailTal Aug 21 '14 at 17:08
  • 1
    Let it stay, it can be edited to exclude the article mentioning or something like that. Just edit it to be in the same context as the original. I honestly didn't know that. -> I fell for that too, and I wasn't the only one... – AlwaysLearningNewStuff Aug 21 '14 at 17:08

There are no complete openings refuted by computer analysis from scratch as far as I know. There has been computer assistance in the analysis of more than a few gambits, here you can look for more details. However, in modern opening theory, computer evaluations (running on powerful software) are given a very high regard by the top players, and this has caused some openings to fall in use, as being thought to not provide any advantage. One of these opening is the Italian game:

[Title "Italian game - drawing line"]
[FEN ""]
1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. c3 Nf6 5. d4 exd4 6. cxd4 Bb4 7. Bd2 Nxe4 {This was considered to be a greedy line that lead nowhere for black, since white gets active play and centralizes his pieces effortlessly. Nowadays, this has become a drawing line for black.} 8. Bxb4 Nxb4 9. Bxf7 Kxf7 10. Qb3 d5 11. Ne5 Ke6! {This was the discovery of the computer.} (11...Ke8 12.Qxb4 {This was the old line, with a very comfortable game for white.}) 12. Qxb4 Qf8! {And now the black queen attacks both the white queen and the f2 pawn, forcing the exchange.} 13. Qxf8 Rxf8 {The position is absolutely equal.}

The problem is the huge number of variations the computer has to calculate. This number of different chess positions is considered to be around 10^50, and these positions can be achieved via many different move orders. Thus, the current algorithms that involve MinMax and similar weight-based methods can hardly cover a fraction of those variants.

An enlightening example is to see how the endgame tablebases are generated. They start from the mated position and go backwards, they do not use the tree-search that characterizes computer engines because it has too many branches for an efficient computation. This is a reason why a computer sometimes gives advantages in theoretically drawn positions: he does not see the end of the variation.

IMHO, it doesn't really make sense to have a project to solve an opening (at least not operating in polynomial time) due to the enormous number of possible lines and the time that takes to analyze them all.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.