Talk:Anti-computer tactics

This article was nominated for deletion on 29 January 2010. The result of the discussion was keep.

Rybka vs Nakamura

It's hard to believe that Rybka chose to search only 2 ply on the 181th move Kb3. What's the difference between 2 ply and 3 ply? Wangzirui (talk) 12:21, 10 November 2010 (UTC)[reply]

Sorry for the late reply. The difference is one ply. 2001:9E8:4626:1800:A9F9:BE83:24A6:F82A (talk) 15:05, 23 September 2023 (UTC)[reply]

Useful

http://www.xs4all.nl/~timkr/chess2/honor.htm SunCreator (talk) 12:55, 30 January 2010 (UTC)[reply]

Do anti-computer tactics still work?

Given the recent improvements in both brute-force tree searching and the addition of deep learning techniques, are anti-computer tactics still effective against powerful, well-programmed computer opponents? The lack of high-level wins against chess computers, and the recent wins by AlphaGo against apparent attempts to use anti-computer tactics, suggests not. Is this something that has been discussed somewhere we can cite? -- The Anome (talk) 12:22, 15 October 2016 (UTC)[reply]

Yes there are anti-computer tactics being used successfully in present days (2020-2021).

Andrew Tang has defeated Stockfish 11 (before NNUE) in 14 games. He takes advantage of engines poor evaluation of the clock time it should use at each move. Andrew Tang uses a very closed and defensive position were he just makes prophylactic moves. Trying to penetrate, the engine spends all of it's clock time. Then, when the engine is almost running out of time, the human can open up and attack. To avoid being flagged, the engine just plays random moves, making it easy for the human to checkmate it. Andrew Tang used super fast games Ultrabullet games from Lichess, were each opponent has just 15 seconds on their clock for the entire match.

And Jonathan Schrantz has found a different tactic which is to win positional advantage that the engine can't see. He has beat Stockfish 11 and even Stockfish 14 NNUE using Urusov Gambit. Stockfish is greedy. It takes up material while not evaluating correctly the positional drawback, which has long term consequences. Further on the game, when the engine detects it's disadvantage, but then it's too late.

The full description with game references has been posted in Human-Computer page, but perhaps the best place to post it is here in Anti-Computer page. Please anyone, feel free to review, re-edit and re-post it:

https://wiki.riteme.site/w/index.php?title=Human%E2%80%93computer_chess_matches&oldid=1078973247#Stockfish_defeated_in_quick_online_matches_(2020-2021) Murilo Perrone (talk) 04:03, 25 March 2022 (UTC)[reply]