Jump to content

Talk:First-move advantage in chess

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Featured articleFirst-move advantage in chess is a featured article; it (or a previous version of it) has been identified as one of the best articles produced by the Wikipedia community. Even so, if you can update or improve it, please do so.
Main Page trophyThis article appeared on Wikipedia's Main Page as Today's featured article on March 21, 2009.
Article milestones
DateProcessResult
May 3, 2008WikiProject A-class reviewApproved
June 16, 2008Good article nomineeListed
June 27, 2008Featured article candidatePromoted
Current status: Featured article


are we going to talk 1st move advantage in chess960?

[edit]

eg https://chess.stackexchange.com/questions/37673/why-dont-these-statistics-disprove-whites-supposed-larger-practical-advanta Thewriter006 (talk) 07:16, 29 January 2022 (UTC)[reply]

No. This article is not concerned with chess variants. MaxBrowne2 (talk) 12:06, 29 January 2022 (UTC)[reply]
(1/2) says who? and also, calling fischer's chess a 'variant'. lol. ok. i suppose we should treat 9LX the same as crazyhouse even if 9LX already has recognition from FIDE. Thewriter006 (talk) 09:13, 31 January 2022 (UTC)[reply]
The idea that chess960 is anything other than a lightly-played modern variant is very strange. --JBL (talk) 13:26, 31 January 2022 (UTC)[reply]
(1/2) really then why was magnus carlsen 'deeply ashamed' https://www.theguardian.com/sport/2019/nov/08/chess-magnus-carlsen-deeply-ashamed-wesley-so-defeat-random-world-final-oslo and https://lichess.org/forum/team-fischer-random-chess-center/world-fischer-random-champion-should-automatically-be-part-of-the-candidates and https://www.youtube.com/watch?v=PTJ100arAbw ? Thewriter006 (talk) 13:49, 31 January 2022 (UTC)[reply]
It is not clear to me what you think this has to do with anything. --JBL (talk) 16:21, 31 January 2022 (UTC)[reply]
i don't think magnus would be 'deeply ashamed' if 9LX were 'a lightly-played modern variant'. additionally, check out 2021 chessdotcom move of the year. it was from wesley so's 9LX game against MVL https://www.chess.com/news/view/2021-chesscom-awards-winners#move there wasn't like a separate category for 9LX moves or anything. Thewriter006 (talk) 17:53, 1 February 2022 (UTC)[reply]
If Jun Mizutani and Mima Ito lost the World Series of Beer Pong, it would not elevate the game to some higher status, no matter how ashamed they were of it. Chess is a game that has been played competitively for centuries, whose rules are known by millions of people, for which major events receive coverage in mainstream media worldwide, and which serves as an idiomatic reference-point in phrases like "eleven-dimensional chess". In comparison, Chess960 is a lightly-played variant that hardly anyone has heard of and which lacks any cultural significance. It is my impression that you think this state of affairs is unfortunate (and maybe you are right), but your feelings (and Magnus Carlsen's opinion about his loss) doenot change the reality. --JBL (talk) 18:29, 1 February 2022 (UTC)[reply]
So Magnus would've been 'deeply ashamed' of losing to Wesley in say crazyhouse? bughouse? chessboxing? golf? tennis? Thewriter006 (talk) 05:42, 14 August 2022 (UTC)[reply]
'Chess960 is a lightly-played variant that hardly anyone has heard of and which lacks any cultural significance.' --> Then why is Magnus quitting WCC but playing WFRCC? Thewriter006 (talk) 02:45, 3 October 2022 (UTC)[reply]


While that’s true at present, if reliable sources exist for a discussion of first-move advantage in other chess variants, I think this is one possible location for a short summary of such (though maybe the main discussion should be in the article on chess960). —JBL (talk) 12:49, 29 January 2022 (UTC)[reply]
(2/2) why don't we just say eg statistics in computer chess and computer 9LX show an X increase in white win from Xa to Xb but a Y decrease in white win relative to black win from Ya to Yb? Thewriter006 (talk) 09:13, 31 January 2022 (UTC)[reply]
Using what reliable sources to support that? Above you posted a stackexchange thread, which is user-generated content. Analysis of statistics with no secondary coverage would fall afoul of some mixture of WP:OR and WP:DUE. --JBL (talk) 13:26, 31 January 2022 (UTC)[reply]
(2/2) true. but some part of some of the stackexchange posts is based on an actual data. hypothetically if i could get a better source? Thewriter006 (talk) 13:50, 31 January 2022 (UTC)[reply]
I personally feel that, if there are two high-quality sources, it could be appropriate to add a short section to this article. (It is possible that other editors will disagree, of course.) I see that the article Fischer random chess already has a brief discussion (with two ok-but-not-great sources), and it might be natural to expand or improve the discussion there first, assuming better sources exist. --JBL (talk) 16:21, 31 January 2022 (UTC)[reply]
'brief discussion' --> on white's advantage? which part please? the 'theory' ? https://en.wikipedia.org/wiki/Fischer_random_chess#Theory Thewriter006 (talk) 16:24, 1 February 2022 (UTC)[reply]
Yes, it is the subject of two or three sentences of the section on theory, alongside the discussion of opening strategy. --JBL (talk) 18:29, 1 February 2022 (UTC)[reply]
THANK YOU Thewriter006 (talk) 08:28, 8 March 2022 (UTC)[reply]

For those interested in the situation in shogi: Yoshiharu Habu said (though I haven't found a good RS to cite) 打ち歩詰めのルールがなければ、将棋は先手が有利 (if drop-pawn-checkmate were allowed, the first player would win). OTOH, ru.wp mentions a translated shogi proverb: Где есть утифудзумэ, найдётся и цумэ (when there is drop-pawn-checkmate, there is a forced mate too). This is of some interest, considering that Crazyhouse differs from shogi in that nifu and uchifuzume are allowed in Crazyhouse. (That nifu is allowed shouldn't be a problem, since doubled pawns are normal in chess, unlike in shogi when they wouldn't be possible moving from the initial position without drops; but maybe the latter does play a role.) Double sharp (talk) 13:47, 26 September 2023 (UTC)[reply]

This article needs more emphasis on science

[edit]

The article is littered with anecdotes, quotes and wild claims, and barely mentions computer chess. AlphaZero is a particularly interesting case in point, because it by design is not "biased" by human preconceptions. However, the only mention is that of a match between A0 and Stockfish. Indeed, self-play statistics would be more appropriate.

In addition to a section about definitely solving chess (as either a forced white win, a forced black win or a mutually forced draw, assuming best play) and another one on perspectives from computer chess, it seems fair to emphasize how time controls, intelligence and psychology may affect the balance in games between humans. Unfounded opinions should be deemphasized. Elias (talk) 13:00, 26 January 2023 (UTC)[reply]

Wikipedia articles are based on what reliable sources say about a topic. If you have access to good-quality sources that discuss first-move advantage in chess through the lens of computer chess, I do not think anyone would object to the addition of content based on them. --JBL (talk) 18:18, 26 January 2023 (UTC)[reply]
Check out page 7 in this paper *preprint* from DeepMind, coauthored with GM Kramnik: https://arxiv.org/pdf/2009.04374.pdf
"Figure 2. AlphaZero self-play game outcomes under different time controls. As moves are determined in a deterministic fashion given the

same conditions, diversity was enforced by sampling the first 20 plies in each game proportional to their MCTS visit counts[*]. Across all variations the percentage of drawn games increases with longer thinking times. This seems to suggest that the starting position might be theoretically drawn in these chess variants, like in Classical chess[**], and that some of the variants are simply harder to play, involving more calculation and richer patterns."

* This is analogous to sampling solid openings from an opening book. I have reservations to the choice of 20 plies here, though. It seems unnecessarily deep considering the purpose of generating unique games. A shallower opening sampling could have yielded very different results.
** Chess is not proven to be theoretically drawn with perfect play. I assume that they were trying to say that the growing proportion of draws with thinking time happens with classical chess too, not only with the proposed variants.
The final paper should be available somewhere, I guess...Elias (talk) 15:09, 15 February 2023 (UTC)[reply]
It has already been proven that when Stockfish plays itself, the game ends in a draw. The only time this isn't so is if a refuted opening or an imbalanced opening is played, in which case the stockfish that has the advantage will win. Acecat (talk) 12:05, 13 September 2023 (UTC)[reply]
This does not constitute proof that chess is drawn from the starting position. Stockfish plays well, but not perfectly. Elias (talk) 07:32, 22 September 2023 (UTC)[reply]
We can talk about empirical evidence, which strongly suggests that the game is a draw with perfect play, but "proof" is impossible since we can't make tablebases containing more positions than there are atoms in the universe. MaxBrowne2 (talk) 08:37, 22 September 2023 (UTC)[reply]
I suppose the only way we will ever know if chess is solved is when there is a tablebase of moves up to 32 pieces, the max on a chessboard at once. Acecat (talk) 15:17, 21 November 2023 (UTC)[reply]
No doubt engines will continue to evolve into closer and closer approximations of perfection, but they'll never quite get there due to physical limitations on computing power. SF is already close to perfect for all practical purposes, but mathematics doesn't deal in practicalities. MaxBrowne2 (talk) 20:31, 21 November 2023 (UTC)[reply]
+1 Double sharp (talk) 09:17, 23 December 2023 (UTC)[reply]

PS: GM Vladimir Kramnik has not shown himself as a champion of science lately, in the case of his cheating accusations against GM Hikaru Nakamura. I think his participation in the mentioned DeepMind study makes it weaker, not stronger. Elias (talk) 14:26, 18 December 2023 (UTC)[reply]

"Some Openings have been used as drawing weapons" Berlin Defence???

[edit]

Marshall Attack is a very sharp but reputable gambit, while the Berlin is a solid opening famous for being solid and the "Berlin Wall" Variation. Jishiboka1 (talk) 05:47, 12 March 2023 (UTC)[reply]

There are forced draws in the Marshall, and among super GMs it's mostly seen as that (among correspondence chess players, even more so). That's the entire point of mentioning it here: it's a perfect case of a sharp opening line getting analysed out to a draw and from then on becoming a drawing weapon because everyone memorised 30 moves of engine analysis, same story as the Poisoned Pawn Najdorf with 10.f5 as mentioned immediately afterward. And it's also why everybody avoids the Marshall at top level, either with 6.d3 as we're seeing this World Championship, or with 8.a4 or 8.h3 as we saw in the previous one. (Though Larry Kaufman has commented that 8.h3 is going out of vogue because 8.h3 Bb7 9.d3 d5! sneaks the Marshall idea through anyway.)
From what I've heard, the same problem afflicts the Najdorf, Sveshnikov, and Grünfeld at top level. (Nepo stopped playing the Najdorf and Grünfeld for WCC matches and switched instead to the Petroff and ...e6 Indians; not sure if this is the reason, but it seems plausible to me.) It's a different problem from what happened to the other openings that mostly disappeared (like the QID), which mostly did so because the most recent engines proved that actually space is really important and that giving it up means you really have to suffer to get the draw. Of course I say "mostly", since Ding recently brought the French out as a surprise weapon. :) Double sharp (talk) 00:13, 29 April 2023 (UTC)[reply]
(That said, the issue with the Najdorf is that you need a lot of preparation to survive and get the coveted equal middlegame! There are very many dangerous ways for White to play for an advantage.) Double sharp (talk) 09:54, 24 September 2023 (UTC)[reply]

Grammar

[edit]

A consensus IS general. That’s the meaning. Remove the word “general” 2001:569:733B:8600:BC7F:B571:81AF:6B5C (talk) 18:09, 29 August 2023 (UTC)[reply]

Fair enough. Done. Double sharp (talk) 08:24, 12 September 2023 (UTC)[reply]

Statement about proportion of drawn games in lead

[edit]

The lead states, without reference, that As the standard of play rises, all the way to top engine level, the number of decisive games approaches zero, and the proportion of White wins among those decisive games approaches 100%.. Looking in the body of the article, it does not seem that decisive high-level engine games approach anything near zero, nor is white's win percentage nearing totality (it is 55%) . Chess Engines Grand Tournament games appear are reportedly 41% draws, which is higher than human play statistics but not enough to support this statement. --LukeSurl t c 13:54, 20 November 2023 (UTC)[reply]

If the top engines are left to play their own openings, the result is close to 100% drawn. Inferior openings are deliberately used in engine tournaments to increase the chance of a decisive result. MaxBrowne2 (talk) 21:34, 20 November 2023 (UTC)[reply]
I can’t see such a statement being supported by the existing references in the article. LukeSurl t c 02:45, 21 November 2023 (UTC)[reply]
Anyone familiar with chess engine competitions knows this to be true. Draw death is even more of a factor in correspondence chess, e.g. check out the cross table from the current World Correspondence Chess Championship. These are essentially engine games with long time controls, making a draw a practical certainty. MaxBrowne2 (talk) 02:53, 21 November 2023 (UTC)[reply]
I don't doubt you, it makes intuitive sense. But, WP:V, there are no citations to support such a strong statement in the article at present, and the actual data referenced in the table (from 2009) show only 41.3% draw rate and only a 5% white edge in decisive games (I assume these are games from the standard opening position, as otherwise they would not be comparable to the other data in this table).
I'm looking around and I'm finding it difficult to find substantative, recent, statistics for engines playing from the standard opening positions. LukeSurl t c 11:52, 21 November 2023 (UTC)[reply]
@LukeSurl: It is stated explicitly by Kaufman in his 2021 book Chess Board Options, p. 206: As the level soars past 3000, the percentage of decisive games drops ever closer to zero, while the percentage of White wins among those decisive games approaches 100. This is why engine vs. engine tournaments and rating lists have to use arbitrarily chosen openings (playing each side once) to keep things interesting. This quote is already in the first section of the article, immediately above the results from CEGT. Hence I have restored the statement. (It's difficult to find statistics precisely because the result is so boring that no one feels the need to try it, but Kaufman was one of the Komodo developers till this year, so he should know. As MaxBrowne2 has noted, correspondence chess is probably what you actually want to look at here, since there engine use is allowed.) Double sharp (talk) 08:19, 3 December 2023 (UTC)[reply]
@Double sharp: You are required to provide a citation IN THE ARTICLE, not on the talk page. You're an experienced editor; be better. Quale (talk) 09:24, 23 December 2023 (UTC)[reply]
@Quale: I did do that, in the section "Winning percentages". Per MOS:LEAD it is quite normal for citations to appear in the body but not the lede, when the lede is repeating stuff already cited in the body. But okay, since people have already questioned the verifiability, I've added the same ref to the lede. Double sharp (talk) 09:51, 23 December 2023 (UTC)[reply]
@Double sharp: I apologize, I need to take my own advice. You are correct that that the lead does not normally need cites because it should summarize key points in the article body and the cites properly belong there. Somehow I missed the relevant part of the article body. That's rather inexplicable since it is not difficult to find. Quale (talk) 09:30, 27 December 2023 (UTC)[reply]

other Kaufman solutions that unfortunately can't go in the article because they're not in his 2021 book

[edit]

From a 2022 post by him on the TalkChess forum: when a game is drawn, immediately play another game with reversed colours, but don't reset the clocks. This is what's done in shogi, which is why I put it on the talk page (there's at least a non-zero chance that someone will try it in a chess event, since it's already done for a game in the chaturanga family). Kaufman suggests an improvement where the colour order is WBBWBWBW... instead of WBWBWBWB... (i.e. for the 3rd game and only the 3rd game, colours don't reverse).

Another TalkChess post by him suggests Chess324 (with two games played with reversed colours from the starting position). He calls it "the only variant that I know of which solves both the preparation and draw problems of high-level chess with no rule changes from normal chess, only the start positions". Double sharp (talk) 15:05, 11 January 2024 (UTC)[reply]