In chess, the 50-move rule, introduced in the 1500s, is designed to prevent a player from senselessly prolonging a game where he has a small advantage but no real hope of winning. It says that if 50 moves pass without a capture or pawn move, either player may insist on a draw. For some time it was thought that all winnable positions could be won or simplified within 50 moves, so that the 50-move rule would not change the value of any position given proper play. Then in the 20th century some positions were discovered that can be eventually won, but not in 50 moves. The rule was briefly modified to allow more moves when such a position arose, but eventually this was considered too complicated and the traditional rule returned in 1992 (thanks to Wikipedia for the history.)
I propose a modification to the rule: If Bob wants a draw under the 50-move rule but Ann thinks she can win, she may insist on an extension of 50 moves. There is a cost, however: If Ann fails to win the game, she scores a loss rather than a draw.
The idea is to avoid an outcome I think most chess players would find contrary to the spirit of the game: Ann knows she can win in a few more moves, but the rule kicks in and it’s a draw. When the extension is invoked (which would surely be quite rare), we know it’s not a waste of time — someone will win the game. I don’t know what the wisest length is for the extension, but 50 seems reasonable. It’s doubtful any human being would have the confidence to gamble on the extension if victory isn’t coming within 50 moves, anyway. Of course, you could let the extension length depend on the position, but this would again result in an unwieldy rulebook, which was the objection to the variable-move rule which was briefly in effect.
By the way, the world record, established by computer search in 2008, for moves necessary to win a position with optimal play is an amazing 517!
6 comments
May 18, 2010 at 12:27 pm
michael webster
Isn’t there some research which shows that computers are very bad at playing chess when the draw rule is expanded?
An odd result, to be sure.
Have I got that research wrong?
May 18, 2010 at 11:44 pm
Peter Dorman
I understand the motivation, but this is not a good rule. Many such endgames are extremely difficult to win — the win may be there in theory, but the chance of even a very strong player finding it can be well under 50%. Consider, to take one example, king and two bishops vs king and knight (no pawns). This is a theoretical win for the bishops, but try your luck at it. (I’ve played this endgame in a tournament.) In practice, imposing a draw penalty (draw = loss) will discourage most players from going beyond 50 moves. If that’s what you want, just set a hard 50 move limit. If you want flexibility, look elsewhere. Maybe computer chess programs can be used for this purpose: have the engine calculate the shortest winning sequence (if a win exists), and then double this number of moves.
May 19, 2010 at 5:00 pm
Jim S
Jonathan’s rule does not hold a gun to Ann’s head: if she does not think she has a strong chance of finding her theoretical win in another 50 moves, she simply declines her option, and her opponent seizes a draw, which is as the current rule would work.
This is quite a reasonable rule to consider. It hurts two types of people: (1) those who get themselves into theoretically lost but computationally intractable positions, and (2) those who get themselves into theoretically won positions, but who are too self-unaware to know whether they know how to win it!
Joking aside, I can imagine someone in a tournament needing a desperation win, and suffering little downside from exercising the option, prolonging agony for another 50 moves. This could be considered a minor cost. Perhaps its relative frequency (compared to Jonathan’s scenarios) would be enough to squash the rule, however.
May 19, 2010 at 12:38 am
Jonathan Weinstein
Fair enough, maybe it would be so rare that anyone would want to invoke the rule that it would be simply irrelevant. I was mostly thinking that if in 50 moves someone happens to reach a position that they can win in just a few more moves, they could invoke the extension. It would seem odd to get in 50 moves to where you can see the light at the end of the tunnel, and then have to stop. But maybe that just hardly ever happens, I don’t have the chess experience to know. The rule does satisfy “first, do no harm” (to the game.)
June 8, 2010 at 5:39 pm
craig kensek
I like this. Forget those ties. I can’t say I have sympathy for the self unaware people. This would sharpen (hopefully) their instincts as to how to go for the kill.
November 25, 2011 at 11:15 pm
David
Is this a real problem in chess? Unlike your previous post (about basketball foul trouble), I think situations where a player could have won but for the 50 move rule making it a draw are very rare in practice. There is a reason no one even thought such situations existed before the computer age. I’m wondering what the purpose of your rule change is. Is it to handle the very rare situation where it theoretically requires more than 50 moves? Or is it to handle a situation where someone plays one of the more difficult endgames (eg King-Knight-Bishop v King) less efficiently than optimal but could win given a bit more time?
In the latter situation I think your proposal makes sense. In the former situation, the question then becomes one of how you set the extended limit. There is, I will admit, a certain inelegance to the fact that a situation (however rare) may be a theoretical win that cannot be completed in fewer than 50 moves. But if you extend it just by another 50 moves, you haven’t remedied the problem because, as you note, situations requiring up to 517 moves have been found. The inelegance remains. You cannot simply allow Ann, in your scenario, to set her own limit, because she has nothing to lose by requesting thousands of moves. But you also cannot be sure that even 517 moves is the theoretical maximum (I think it is the highest known but it is unproven whether there are even longer endgames yet to be found).
I do like your proposal, though, in the case where someone has a clear path to checkmate after 50 moves but just needs a few more moves to complete the win. It seems a fairer way to resolve such situations. Given the extreme rarity with which I’d expect this option to be exercised, though, I’d allow more than 50 moves for the extension: maybe 100 or 200 moves. If this leads to a marathon game once in a blue moon, then so be it.