bramcohen: Eliminating draws from Chess

Bram Cohen (

bramcohen) wrote,
@ 2005-11-04 10:58:00

Eliminating draws from Chess
Here's an idea for eliminating draws from chess: add a 'tiebreak' rule, which is that games which are draws under the regular rules are instead won by the last player to have made a capture.

That would completely eliminate draws, and in non-drawish situations leave play completely unchanged.

(Post a new comment)

	improvements mongoltrophies 2005-11-04 07:12 pm UTC (link)
	That would make chess more fun to play. Fischer Random Chess makes the game more interesting, too. (Reply to this) (Thread)

	Re: improvements bramcohen 2005-11-06 06:44 pm UTC (link)
	I like random start openings as well, but they sure do change the character of the game. (Reply to this) (Parent)

	feallanastrum 2005-11-04 11:00 pm UTC (link)
	That's a really great idea. I'm going to start playing that way. (Reply to this)

and_ham
2005-11-04 11:12 pm UTC (link)

Excellent idea.

In an old book at my grandmother's house, a chess variant suggested that one player play with a full set of chess pieces that follow regular rules. The other player plays with a queen that can move like a queen or a knight. The object of the Queen/Knight is to capture the opponent's King, which doesn't require the player to call check before hand. The player with the full set attempts to capture the Queen/Knight. The only flaw I've found in this variant is that the player with the full set can slowly convert all of her pawns to queens. I added a rule that limits the number of additional queens or rooks to 3 (to prevent the player from covering every open row with these pieces). I've found this is an EXCELLENT variant for teaching a new player how to play chess defensively and think several moves ahead. You must defend your pieces and think critically to defeat the Queen/Pawn.

(Reply to this) (Thread)

	and_ham 2005-11-04 11:13 pm UTC (link)
	last word should be Queen/Knight, not Queen/Pawn. (Reply to this) (Parent)

bramcohen
2005-11-06 06:41 pm UTC (link)

I've played that variant. Even without the ability to queen pawns, the regular chess pieces side has an extremely strong advantage, and I suspect there's a sequence of moves which that side can play blindly and always be guaranteed a win, because it eventually controls every square on the board without ever leaving a piece unprotected.

It's a very good exercise in learning not to go nuts with the queen though.

(Reply to this) (Parent)(Thread)

	Amazon chess clumma 2005-11-14 07:08 am UTC (link)
	I've never heard of this variant, but it reminds me of Amazon chess, in which both colors get Amazons instead of queens. They move like a rook and a knight instead of like a rook and a bishop (as queens do). (Reply to this) (Parent)

Stops exchanges

brad_templeton
2005-11-05 12:04 am UTC (link)

A common tactic in Chess is an exchange or sacrifice. You offer up a piece, which is taken, and you get to take a piece in return, and you hope to get better position. With this rule, after the early game, players will not want to risk being the first taker in an exchange, and they won't want to make a pure sacrifice unless they know it will bring a win. Normally though you happily do sacrifices to queen a pawn or gain other valuable positions.

Worse, players will search for forced draws whenever they are the last player to have taken a piece. Normally only players who feel they have little chance of winning will search for forced draws. With this rule the dominant player would start searching for them. There are lots of forced draws (making the king dance with a rook or queen shows up often) available to the dominant player, and often to both players.

If somebody starts an obvious exchange with me, I can't surprise them by not taking the exchange without a lot more risk.

This really changes the game more than you suggest.

To make it work you would have to consider improvements, such as the flag is reset if a pawn is queened, and the last capture has to be a long time ago, but even that's not enough.

Not saying that no-draw chess could not be an interesting game, but it would not have the same endgame or even midgame as chess.

(Reply to this) (Thread)

Re: Stops exchanges

bramcohen
2005-11-06 07:21 pm UTC (link)

The effect of this rule on exchanges is to encourage players to offer exchanges, since the player which offers gets the last capture. That has the effect of increasing player aggressiveness.

Changing the rule to 'capture a piece or promoting a pawn' is entirely reasonable, although I doubt it would change very many situations. I'm pretty sure that changing it to 'capture a piece or push a pawn' would make the game far too dug in.

It is true that this rule could encourage a player to 'play for a draw', but then it could also encourage players to do things like exchange a piece for three pawns if they can still hang in afterwards, and to engage in dangerous tactics like leaving opponent pawns hanging for a long time until simplification is complete. Whether the balance of new tactics is to slow the game down or make it more aggressive is unclear, and can only be resolved by actual play-testing.

Undoubtedly this would change the game a lot at high levels, simply because a large fraction of all grandmaster games are draws and this rule would change all of them, but it does do a decent job of keeping the game recognizable as chess, which other no-draw chess variants don't.

(Reply to this) (Parent)

	root_fu 2005-11-05 01:39 am UTC (link)
	Worse case scenario: You kick ass from opening move to closing game. At the end you have 5 pieces left. Your opponent has a king. While attempting to capture the remaining king one of your pawns is sacrificed. You lose--your opponent captured the last piece... (Reply to this) (Thread)

	bramcohen 2005-11-06 06:46 pm UTC (link)
	That would be highly unlikely. At high levels of play spotting even a pawn at the beginning of the game leaves one completely lost. (Reply to this) (Parent)

	discountbrain 2005-11-05 02:02 am UTC (link)
	this would upset the natural balance of chess. The white player has the numerical advantage because he gets first move. (Reply to this) (Thread)

	bramcohen 2005-11-06 06:47 pm UTC (link)
	I'm not sure what you're saying - this tiebreak rule doesn't alter the numerical advantage at all, except that a fair number of draws may get decided in white's favor. (Reply to this) (Parent)

	darius 2005-11-05 04:48 am UTC (link)
	Isn't it technically possible to get a draw without any captures? I vaguely remember a repeated-position rule for drawing. (Reply to this) (Thread)

	bramcohen 2005-11-06 06:48 pm UTC (link)
	Technically yes, but in practice it never happens, and one could somewhat arbitrarily decide such games in favor of black. (Reply to this) (Parent)(Thread)

	clumma 2005-11-14 07:10 am UTC (link)
	Draw by repetition happens all the time! (Reply to this) (Parent)(Thread)

	bramcohen 2005-11-18 05:16 pm UTC (link)
	I meant, in practice you never get a draw by repitition before any captures have happened. (Reply to this) (Parent)

misterajc
2005-11-05 02:40 pm UTC (link)

Draws would still be possible, provided they happened before any pieces had been taken. A draw can happen by mutual consent. Bobby Fischer once complained that in tournament play the Russian Grand Masters were agreeing easy draws against each other and then playing each game to the bitter end with him in an effort to tire him out. Under those sort of circumstances a game between two Russians might consist of white advancing the king's pawn and offering a draw, and black accepting.

Even if you do not allow a draw to be mutually agreed, if both players were cooperating, it would be possible to arrange a draw by repeated moves, or by the 50 move rule, before any pieces have been taken.

(Reply to this) (Thread)

	nvop 2005-11-06 07:39 am UTC (link)
	The 'draw' rules could be changed a bit as well: in an intentionally forced draw (repeating the same position, 50 moves without taking a piece), the player making the 'draw-making' move would lose (since he can make a different move, not causing a draw) (Reply to this) (Parent)(Thread)

	bramcohen 2005-11-06 07:12 pm UTC (link)
	Better to just give it to whoever got the last capture. The idea is to give the win to the player most in control at the end, not to penalize the one who's being a party pooper. (Reply to this) (Parent)

	bramcohen 2005-11-06 07:11 pm UTC (link)
	The 'last to capture wins' applies to repeated move draws as well. Games which involved no captures could be decided in favor of black. Draw by mutual agreement is simply banned. (Reply to this) (Parent)(Thread)

clumma
2005-11-14 07:23 am UTC (link)

Draws are a natural part of chess. They fall out of many endgame positions. And it seems likely that draw is the outcome of the game if both sides play optimally. Your suggestion has many problems, including that the last capture may have happened long before the drawish position arose, and that there may not be any legal or logical captures available. It also ignores the possibility of an otherwise losing player making a brilliant stalemate (fails to force winning players to prevent this).

Draw by mutual agreement is a different story, and banning it is possibly a good idea. This was tried for the first time in a high-profile tournament this year, with good results...

http://www.mtelmasters.com

Coincidentally, there has been a discussion on the 'problem of draws' recently on ChessBase. See for example...

http://chessbase.com/newsdetail.asp?newsid=2719
http://chessbase.com/newsdetail.asp?newsid=2729

-Carl

(Reply to this) (Parent)(Thread)

bramcohen
2005-11-18 08:00 am UTC (link)

I read that first article previously and was inspired by it to come up with this technique. I think that deciding games based on extremely short time control tiebreaks is about as reasonable and sporting as using a coin toss.

There are two criteria at work here (1) chess should be difficult for computers, and (2) the better player should consistently win. Relying on short time controls is terrible for both of those (although one could reasonably argue that it's a bit late to be worrying about man vs. machine at chess). My last capture rule is something which allows for long-term planning, both good for humans against computers and good at giving the advantage to the player who can plan farther ahead.

(Reply to this) (Parent)(Thread)

clumma
2005-11-18 04:16 pm UTC (link)

I agree that the blitz tiebreak is a bad idea. I think the Sofia rule banning draw by agreement is better, and should be enacted. Another suggestion that I've seen is to make draws worth 2/3 points for black (or perhaps a value derived from statistics such that the total tournament performance of the colors is equal).

Agree also that chess should be difficult for computers. Yes it's a little late, but one way to help would be to deny computers opening or endgame databases. Or allow humans access to them. (Interesting that in the 'anything goes' tournament held by ChessBase, strong humans playing with the aid of computers easily beat lone computers and humans.)

As for your assertion that the better player should consistently win, how much better should they have to be? Of course, "better" is in fact defined only by "they win". So if they don't consistently win, they're not that much better. All you can really say is that you feel there are too many ways to draw in chess, which is an opinion you're entitled to of course.

As for your last capture suggestion... why not send it in to ChessBase and see what other players think?

-Carl

(Reply to this) (Parent)(Thread)

bramcohen
2005-11-18 05:35 pm UTC (link)

Draws are in general a real problem, if for no other reason than they suck from a spectator standpoint. They may be a part of the game as it stands today, but I see that as a problem with the game.

Denying computers opening and endgame books would be completely artificial, and in the case of openings, it isn't clear that opening books are consistently better than what computers figure out over the board. Anything goes is quite an interesting format. Notably the person who won the last anything goes format isn't even a competitive player, but seems to be able to augment a computer very well.

The better player should consistently win thing isn't a comment on the number of draws. In Hex or Go you frequently see small round-robin tournaments which form a strict ordering of all the players from best to worst, with no out of order results, while chess tournaments tend to be much more fuzzed out.

How does one send suggestions in to Chessbase?

(Reply to this) (Parent)(Thread)

clumma
2005-11-18 07:58 pm UTC (link)

Reducing draws in chess sounds good to me, but I won't play the resulting variant unless it's as deep as the existing game. Chess is fun because ply destroy information about the result, but some leaks through in the form of patterns (and of course, material balance). My hunch is that the last capture rule makes a strong change in the result that doesn't leak through in a meaningful way. I've already mentioned long endgames -- why should the last capture, made perhaps a dozen moves earlier, matter? The existing game has discontinuities when converting to an endgame, but things like pawn structure and king position (which are hard to achieve) leak through.

In fact, now that I think about it, your rule may do the opposite of encouraging aggressiveness, since the last capture is usually a recapture made in response to capture. Maybe you want to reward the player who *initiated* the last capture, but how do you identify this (what if there are in-between moves)?

I don't know anything about Shogi, but I understand captured pieces can re-enter the game and that this makes it more decisive. And this abstract makes it sound at least as deep as chess:
http://portal.acm.org/citation.cfm?id=512148.512155

That notable person, ZachS, is keeping me up at night. There were hints that he used a *statistical* approach, which is something I've been thinking about for years... IIRC ChessBase said they'd try to get more details, but nothing so far. I tried to mail him but haven't heard back.

'Better player wins' is a comment on the number of draws. If more Hex and Go games were drawn there wouldn't be such nice rankings in the tournaments! But it's possible that even without draws the players may be of equivalent strength. The spread of ratings has always been pretty narrow at the top of chess, and that this remains true when you include computers is very interesting... -Carl

(Reply to this) (Parent)

	clumma 2005-11-18 09:32 pm UTC (link)
	Oh, there's a "contact" link on the left at chessbase.com. http://chessbase.com/contact/index.asp The "news service" is what you want, though it threw an error yesterday when I tried to use it. -C. (Reply to this) (Parent)

I haven't played thinking that way

brad_templeton
2005-11-06 08:16 pm UTC (link)

But my impression remains that the dominant player is routinely getting opportunities to force a draw, especially once the other player's king becomes modestly exposed.

These opportunities don't change the game because the dominant player is not interested in a draw under classic rules. The dominant player does have to make sure she doesn't get in a position where the weaker player can force a draw in classic chess, and indeed sometimes they make a mistake and the game becomes a draw when they could have won.

All it takes is an exposed king and a queen typically, or open ranks and a rook to make the king dance in check.

My concern was that you felt this rule change only affected draws, but in fact it strongly changes all sorts of non-draw games. Though admittedly if it just makes the dominant player win faster that's not as much of a change.

Of course with exchanges you have two types. You have somebody offer a piece up for exchange, often in a place so powerful that the other player has strong incentives to initiate the exchange, or quite commonly you have pieces which sit for long periods able to do a capture for exchange, and suddenly it makes strategic sense to do it. Under these rules, it will not make sense to do it nearly as often because it gives the other player the last-to-capture status.

(Reply to this) (Thread)

Re: I haven't played thinking that way

bramcohen
2005-11-11 08:25 pm UTC (link)

But my impression remains that the dominant player is routinely getting opportunities to force a draw, especially once the other player's king becomes modestly exposed.

There's a frequent claim that at the grandmaster level white can draw almost automatically, but in match situations where a player only needs a draw as white, they bungle it enough to make that claim fairly suspicious. It would be interesting to have a grandmaster tournament where draws were awarded as wins to white and see how well they can really pull it off.

Forced draws by repetition aren't all that common because the king is a fairly powerful piece and generally capable of warding them off on his own.

My rule does encourage players to leave hanging opponent pawns hanging for extended periods, although that's already a common practice because the capture uses a tempo.

(Reply to this) (Parent)

	slayemin 2005-11-07 07:16 am UTC (link)
	I wouldn't like that game. I've been in positions where the best I could hope for is a draw, and managed to wiggle my way into it. It's probably more challenging to draw from a losing position then to lose. (Reply to this) (Thread)

	bramcohen 2005-11-11 08:26 pm UTC (link)
	In Hex there are times when you spend the entire game warding off an opponent's attack. The odd thing about the rules of hex is that if you successfully ward off an attack you eventually win. Psychologically it doesn't really feel like a win though, it feels like avoiding a loss. (Reply to this) (Parent)(Thread)

slayemin
2005-11-12 03:12 pm UTC (link)

chess is a bit different. The winner is the person who can stay one or two moves ahead of their opponent and correctly figure out the cost/benefit of making each move. In an equal match, the game will be neck and neck where even losing the difference of the lowest pawn could spell doom to either player. When I accidentally find myself playing the losing end with no possibility for victory, I try to stalemate the game rather then concede defeat. Psychologically, from the losers side it feels like a great victory lets them breath a sigh of relief when they manage to stalemate...

(Reply to this) (Parent)

Username:		Create an Account Forgot your login or password? Login w/ OpenID
Password:
	Remember Me
	English • Español • Deutsch • Русский…

About

Help

Get Involved

Legal

Store

LJ Labs