Why is every non-checkmate a 0.5-0.5 draw?

Hi David,

You've certainly hit a nerve, based on the number of reactions.

My question to you is: what is the problem that you are trying to solve with this solution?

Hi rterhart,

Good question!  The answer is fleshed out in my essay, but I'm happy to summarize here.

• Capablanca hated draws because he was afraid of the "draw death" of chess... where experts score draws in the majority of their games
• Fischer hated draws because the Russians used the rule to gang up on him.  They took quick draws-by-agreement when playing each other to save their energy to beat Fischer
• "Drawing openings" are abused in match play.  Once one player has a win, they can just draw-out the remaining games with safe lines and avoid a real fight.
• For fans and casual spectators, draws are the least satisfying.  They are more engaged in the match when their team loses.  A draw is just like, *yawn*, why did they even play?
• Finally consider Carlsen's comment after game 1 of the 2013 World Championship:

“about the way the game went today, in general in these lines, play develops a little more slowly, but here there was an immediate crisis and I didn’t see that any of my options were particularly promising, particularly as I missed as Vishy mentioned earlier, move 13 Qe1 and 13...Nb4 is very strong – from then on I had to pull emergency brakes, and had to go for draw.”

Does it sound like the position was equal? Should this be scored the same as king vs king? Carlsen is verbally conceding that Anand had a stronger game, but did not actually concede the game. What sense does that make? Why do our rules allow a player the “emergency brakes” option against a “very strong” move? I see this as a design flaw in our tournament scoring system. 

Altogether, the issue can be summarized by this graph:

All non-checkmate boards are conflated into one value.  Meaning there is nothing to play for when checkmate isn't an option.  But plenty of those games contain a method to force a stalemate... which is a technique we don't even bother learning, because there's no incentive to force a stalemate (unless you want to humiliate your arch-rival in a World Championship).

And most those games that cannot force a stalemate could still be decided by comparing material.  And as a final tie-breaker in cases of equal material, we can give a small partial victory to Black as compensation for the disadvantage of moving second.

My proposal gives every game a winner of some degree.

There is no tie, no outcome where White and Black walk away with the same score.  Some games are a blowout.  Others are a pitchers' duel.  But every game has a winner, and every board position has something to fight for.  Total victory is reserved for checkmates, but if I can't get checkmate, can I get a stalemate?  No stalemate?  Can I get a pawn?

I think my proposal makes the game more interesting to play and more interesting to watch.  And endgames get a new level of depth once we walk away from the crude metric of only considering "can this position force a checkmate?"  We start to appreciate positions where we can force a stalemate... or just be a pawn-up.  More depth.  More nuance.

Kind regards,
David

Hi Petrosyanovich,

My thesis is that K+B vs K isn't a 50-50 outcome.  That the player with a Bishop played a better game, even though total victory eluded them.  So they get a slight partial-victory as a reward.

I believe that it's a flaw in tournament chess to have any game terminate with an equal outcome for the reasons listed in Post #66.

Kind regards,
David

By play equally I mean that one side did something that the other side equaled. So  if one person was able to get into a stalemate, the other person was not able to prevent it. Equal. Neither side was able to win, but neither side lost. Equal. Capturing the king is not playing equally (or resigning or running out of time or disqualification). But stalemate (or any other draw) is playing equally because neither side was able to force checkmate (or force a resign, win on time, etc).

Any draw is where both sides played equally so it makes sense to award points equally. Lets say I was able to force or trick you into a stalemate even though you had an extra queen and two bishops but I had more time. I played the better game because I was able to prevent you from winning, but I still only get a half point because I didnt win  either. Neither of us accomplished the purpose of the game. Equal. Probably the only other point system that makes sense is zero points for both sides. But both sides made an effort, so a half point each makes the most sense.

There are some crazy rules in chess, like en passant and castling. But the rules about most draws make the most sense. Stalemate is probably the most sensible and logical rule in chess. And chess is a game where there is a lot of logic. Think of it like tic tac toe. It's at the end of the game and I cant move. Should I lose just because I cant move? The rules of the game say I CANT move. I cant make up a new rule that says there is another square to mark an X in. The rules prevent me from moving, and the other person didn't have 3 in a row, so it's a stalemate. A tie.

Awarding points based on which pieces are left on the board doesn't make any sense, and this is one good reason why. A good sacrifice might not win the game, but it part of how the game is played and sometimes won. I think there is no better reason to keep the current  system than what you just pointed out. 

Well, I suspect that a relatively low proportion of drawn games could become stalemates. In fact, if we stopped at stalemate, with 0.6 or 0.8, I suspect the draw problem in chess would be hardly affected.

As a great player, Lasker's intuition is indeed to be trusted. What I think, though, is that my goals may be different from his.

One reason I chose a low value for stalemate was to reduce the controversy my suggestion would face. By making the reward for stalemate a small fraction of that for checkmate, I can defend the proposition that my proposed change wouldn't turn chess upside-down, it wouldn't be a big change that would destroy existing knowledge about the game.

Endgame theory as it is, showing how to achieve checkmate, and how to avoid mistakenly turning a checkmate into a stalemate, would retain its validity and importance. Forcing stalemate would be an add-on to the existing body of endgame theory rather than a replacement.

I've started another thread with my own idea in its complete form.

I allow perpetual check as a victory, but worth only 1/10 as much as stalemate - with a 0.51 - 0.49 split of the point for a game. So, again, while I reward the dominance of the board that allows perpetual check to be inflicted, since the reward is so much smaller than that for checkmate - and even stalemate - this is, in terms of points value, still a way to draw instead of a way to win. So instead of turning chess upside-down, I simply add an opportunity for effective play within the large proportion of games that at present result in draws without disturbing the normal struggle for checkmate.

And at the next tier, I really depart from Lasker, who used 0.8 - 0.2 for stalemate and 0.6 - 0.4 for bare king.

While Korean chess has a way of breaking ties based on material superiority, I took comments about an idea I had toying with doing that for chess seriously. Sacrifices are an important part of chess, and sometimes they're speculative, so rewarding material superiority is dangerous, and can go against the spirit of chess.

So at the bottom level, instead of awarding .501 - .499 for bare king, I modified the win condition so that additional play aside from the achievement of material superiority is required. The scheme I suggest is this: when a player bares the opponent's King, on the next move, he has the option of replacing one of his pieces by a Pawn, if he feels he can't win without doing that. If he does so, he can only win by checkmating, and the win scores .501 - .499.

So it is not enough to bare the opponent's King, one has to maneuver to do so on the right square to get a winning K+P versus K ending out of it.

That, though, may be a bad idea, making that kind of victory almost random.

As I understand it, one major reason that competitions like the World Chess Championship involve multiple games, is because winning one game does not guarantee superiority in the game of chess. Winning the series, on the other hand, demonstrates some indication of true results. Even if one player plays slightly better and has more material at the end, unless they can demonstrate true superiority (checkmate) in one game, you cannot truly use it as an indication of who is a better chess player, and thus it shouldn't be used as such. 

"My thesis is that K+B vs K isn't a 50-50 outcome.  That the player with a Bishop played a better game,"

But not a better game of chess! Who cares if you have a king and bishop while the other side has a king? Why does this indicate superior play. You were unable to mate your opponent so why should we consider your play superior in any way? If chess was a game where the object was to have one extra pawn, even if you can never queen it, it would have died out a couple of centuries ago. What is so great about being a pawn up, but unable to convert it?

Watching a great player brilliantly overcome his difficulties and save a half-point is exciting. Watching some oaf clumsily push his extra pawn down the board until he stalemates his opponent, or stumbles into an unwinnable ending is dull 

He still failed to win.  He had zero chance of winning so eventually he ended with stalemate.  I fail to see why that is humiliating for the player getting stalemated.  He's just moving pieces around knowing he can't lose.

Stalemate can have an effect of endgame theory, but I don't see it in a bad way.  If someone can't achieve checkmate then they don't deserve any kind of 'win'.  It certainly doesn't prove any 'superiority' when everyone knows it is just a draw.  

There are plenty of inferior spin-offs of chess here people get rewarded for taking pieces or delivering checks or whatever else it is that don't matter in real chess other than as a route to checkmate.

That's a good point. Chess is defined as a game with certain rules. So the players are trying to achieve certain goals; if they achieve something else, how is that relevant to chess?

Today, a new record was set: a World Championship chess match began with nine draws in a row. This must have a negative effect on how the public views chess as having interest. And the result will be more problems in obtaining funding for chess events.

Something has to change. I believe that going to ever-shorter time controls will do violence to one of the essential elements of chess. Chess is a sport, an art, and a science. All three elements contribute to its unique appeal. So in attempting to address the draw issue, we need to avoid doing serious damage to any of these three elements.

So this is why I also have proposed to award points for stalemate and other things less than checkmate. I have avoided rewarding material superiority, because that indeed is dangerous. To avoid this distorting the game, I have made sure that all the rewards for lesser victories are much smaller than the one for checkmate.

The chance of a draw is reduced, helping chess as a sport.

Existing endgame theory gains a supplement, but is not overthrown or replaced, avoiding damage to chess as a science.

Material superiority is not over-emphasized, and so players are encouraged to sacrifice even for the sake of stalemate or perpetual check, thus protecting chess as an art.

This is very interesting. As was said earlier though, now sacrificing pieces for a draw won't work out so well.

Chess works fine. Draws are the natural result of evenly matched players playing well. I can assure you that there is not an epidemic of draws at the club level! 

"but doesn't it seem silly that every other outcome is scored as 0.5-0.5 as if both players played an equal game? "

No!  The purpose of a chess game, aside from all the pleasure it brings, is to determine who won, not to measure who played better. In fact, the loser will often have played much better chess than the winner--except for that one blunder! Listen to the post postmortems at any tournament and you will hear a hundred variations of that theme.  In American football, if the score end up the same, even after an overtime period, it's a tie. We don't award a partial victory to the team that had the most first downs, or gained the most yards.  Because chess is a very fair game, where White has only a tiny initial advantage, there will be draws.

Agree with DeidreSkye. But let me add that it's not an accident that it's hard to win a chess game. That's by design. A win is an accomplishment. It's frustrating, of course, but it's one of he reasons chess has survived for centuries, and is played by millions, while many other games have faded away

I don't have much knowledge in chess, but I can see a problem/concern with this proposal, I don't think it will work well on swiss type tournament, and every tournament now should be like football league when each match should be played twice (one as black and one as white). Personally think 3-1-0 , or even 4-1-0 would work better for the antidraw (this doesn't work on WCC format though, since it doesn't matter).

And I assume those scoring is adjustable, depends on experience

But you just showed that your algorithm is completely bunk. Why should a player who had a material advantage but have his advantage blunted deserve 0.7? Clearly the player with LESS material should get more points, by this logic! Here's the exact same thing, with a few words changed. How is it any different?:

"If White creates a material gaining attack, but it turns out that Black can blunt it with a 3-fold repetition, then White would get 0.3 and Black would get 0.7.  Why should White get 0.5 for screwing up that attack?"

Exact same situation, reversed onto your proposed criteria.

As for the depressing point, it means that there's no way to get a positive outcome if you enter an endgame down a bit of material in many cases. No stalemate tricks, no Q vs. Q+P draw etc. You're consigned to an endgame where you're fighting for 0.2 points. In addition, you remove a huge amount of nuance from the endgame, in that ending up in a rook's pawn+K vs. K endgame is fine, rather than a draw.

The goal of the game is to checkmate the king. If you succeed, you get 1 point, if you fail, you get 0. If both sides fail, you get 0.5. There's no precedent in the rules for involving material. Essentially, having more material doesn't mean you played better, it means you played insufficiently well to win the game. Why are you trying to reward that?

It doesn't really matter, the first point's ironclad.

How does this make sense? It doesn't... you have no legal move to play, since the king can't move into check.