RSIF as alternative to armageddon

[Ozymandias]

If we are to eliminate draws, which is the theme of this thread, then draws either have to be awarded based on some aspect of the position (or game history) or else just arbitrarily awarded to one color.

I took from the title, that the elimination of draws was a way of solving ties, without having to resort to armageddon games.

From your four point proposal, I was trying to brainstorm ideas that would help alleviate the draw death at the highest level, without actually eliminating the possibility of a draw. In Freestyle and correspondence chess, we need to lower the draw rate, so that strength actually reflects on the scoreboard. I was just picking your brain to see what impact different ways of avoiding the 50-move rule, might have on results.

[lkaufman]

If we are to eliminate draws, which is the theme of this thread, then draws either have to be awarded based on some aspect of the position (or game history) or else just arbitrarily awarded to one color.

I took from the title, that the elimination of draws was a way of solving ties, without having to resort to armageddon games.

From your four point proposal, I was trying to brainstorm ideas that would help alleviate the draw death at the highest level, without actually eliminating the possibility of a draw. In Freestyle and correspondence chess, we need to lower the draw rate, so that strength actually reflects on the scoreboard. I was just picking your brain to see what impact different ways of avoiding the 50-move rule, might have on results.

I think that anything that gets the draw percentage down to 50% or less in top engine or correspondence play would be satisfactory, we don't necessarily have to eliminate all draws. But for sporting reasons, it is often necessary to have a winner.

[Ovyron]

Has awarding a win for "style" been considered? Like, on a draw the side with least material gets the win? There's gambits and sacrifices in some positions that may be unsound, or useless because they lead to a draw, but they look so cool. And in engine v engine matches, something very stylistic is when the opponent hangs a piece, and the engine doesn't care and never captures it, so we'd have some of that, too (a player only capturing if they think the material advantage is enough to force a win.)

I think games of this variant would show some fireworks.

[lkaufman]

Has awarding a win for "style" been considered? Like, on a draw the side with least material gets the win? There's gambits and sacrifices in some positions that may be unsound, or useless because they lead to a draw, but they look so cool. And in engine v engine matches, something very stylistic is when the opponent hangs a piece, and the engine doesn't care and never captures it, so we'd have some of that, too (a player only capturing if they think the material advantage is enough to force a win.)

I think games of this variant would show some fireworks.

A general rule to that effect would cause a huge INCREASE in draws, because no one would dare even try to win a game by winning a pawn unless they were confident it would be enough to win. Usually the side who is a pawn up has played better than the opponent, but often the side who is down managed to get just enough compensation or simplification to hold the draw. Of course awarding points for "style" in general would require an arbiter to make such judgments, and much of the appeal of chess is that the result depends only on the players, not on some third party's subjective opinion. As long as draws are awarded to the side that appears to have played better, and as long as the rules are clear and reasonably simple, I would be ok with such draw-breaking rules. If everything is totally even at the end, like K vs K, then Black has played better and deserves the win.

[Uri Blass]

I suggest the following variant for computer chess(of course we need some interface and engines to support it)

draw is a win for black but there is a limited time when white can take back moves and black cannot take back moves so practically we have the same time limit that we have today and add to it limited time that white can take back moves.


For example time control may be 1 minute per game+1 second per move when white is allowed to take back moves in the first hour.
There is no limit on the number of moves white take back and white can decide even at move 20 to go back to move 1 and try a different move.

Taking back moves also mean taking back the times so the strategy to play fast in order to get a big time advantage on the clock after you take back moves is not going to work(playing fast in order to try more options and maybe get time advantage without takebacks is of course possible).

Note that if black has advantage in time control of x minutes per game+y seconds per move+z hours for take back you can increase z to make the game fair and if white has advantage you can reduce z to make the game fair.

[Joerg Oster]

If we are to eliminate draws, which is the theme of this thread, then draws either have to be awarded based on some aspect of the position (or game history) or else just arbitrarily awarded to one color.

I took from the title, that the elimination of draws was a way of solving ties, without having to resort to armageddon games.

From your four point proposal, I was trying to brainstorm ideas that would help alleviate the draw death at the highest level, without actually eliminating the possibility of a draw. In Freestyle and correspondence chess, we need to lower the draw rate, so that strength actually reflects on the scoreboard. I was just picking your brain to see what impact different ways of avoiding the 50-move rule, might have on results.

I think that anything that gets the draw percentage down to 50% or less in top engine or correspondence play would be satisfactory, we don't necessarily have to eliminate all draws. But for sporting reasons, it is often necessary to have a winner.

Well, what do you think of only changing the scoring of 3-fold repetitions then?

A repetition is always a loss for the side that goes for the 3-fold repetition,
except if this side is down in material by at least a knight (or bishop?) or three pawns!
This is to not penalize an attacking side with some sacs which turns out to only end with a repetition at best.

This rule should not be too difficult to implement in both, GUIs and engines,
as an alternative to the standard 3-fold rule, of course!

[lkaufman]

If we are to eliminate draws, which is the theme of this thread, then draws either have to be awarded based on some aspect of the position (or game history) or else just arbitrarily awarded to one color.

I took from the title, that the elimination of draws was a way of solving ties, without having to resort to armageddon games.

From your four point proposal, I was trying to brainstorm ideas that would help alleviate the draw death at the highest level, without actually eliminating the possibility of a draw. In Freestyle and correspondence chess, we need to lower the draw rate, so that strength actually reflects on the scoreboard. I was just picking your brain to see what impact different ways of avoiding the 50-move rule, might have on results.

I think that anything that gets the draw percentage down to 50% or less in top engine or correspondence play would be satisfactory, we don't necessarily have to eliminate all draws. But for sporting reasons, it is often necessary to have a winner.

Well, what do you think of only changing the scoring of 3-fold repetitions then?

A repetition is always a loss for the side that goes for the 3-fold repetition,
except if this side is down in material by at least a knight (or bishop?) or three pawns!
This is to not penalize an attacking side with some sacs which turns out to only end with a repetition at best.

This rule should not be too difficult to implement in both, GUIs and engines,
as an alternative to the standard 3-fold rule, of course!

Since the point values for the pieces are not part of the rules, and are not universally agreed upon, any rule that refers to them is problematic. I don't see why a player should be able to sacrifice pieces knowing that he can make perpetual check and then look for ways to win, there's nothing brilliant about that. The most satisfactory repetition rule to me is that you cannot repeat the position unless you are in check. So it is always up to the side giving perpetual check to vary. Perhaps even a better rule would be that no piece can return to the square it vacated on the previous turn unless either the intended move or either of the previous moves by either side was a capture or pawn move, or if you are in check. That way no game history need be considered except the last move pair. It has the further advantage that if combined with making stalemate a loss for the side with no move, and eliminating insufficient material draws, I believe it makes all king vs king endings a win for the side with opposition, although I haven't proven that. King and minor vs king would always win. With these rules there would only be fifty move rule draws to deal with. Awarding them all to Black would probably be too much, but who knows?

[Ovyron]

As long as draws are awarded to the side that appears to have played better

I think it was revealed that in several games, Tal actually played worse and won because their opponents weren't finding the refuting lines against his unsound moves, so the idea is to create a scoring system that rewards such play objectively (so a player going for it would get the win even if the game ends in draw.)

But, yeah, any such system wouldn't be simple.

Another idea is to let black have draw wins, but let some condition in the game let white change the player's colors. Say, the first repetition of position makes black white and vice versa, or a pawn promotion does it, or a check with some piece does it, or an exchange of some piece does it, or n moves without pawn move or capture does it, or absence of a kind of piece on the board does it (etc.) and then another condition where the black player that is now white can do something else to become black again, so if a draw happens they win.

Chess has become "mate-centric", but keeping the opponent from achieving that is easy enough that we have the draw problem, and perhaps an idea that has other objectives in the game so that you can achieve those and draw the game and get the win (so, besides drawing the other side wants to stop you from achieving those) could solve it.

[Erdmundr]

A proposal I made in the TCEC chat is to take the earlier idea that a king reaching the opposite rank wins and connect it to the following draw scoring proposal.

When the game does not end in mate, or by the king reaching the opposite side, then each player receives 1/8 of a point for every rank their king has traveled to. In addition, a king moving to a rank which it hasn't previously reached resets the 50 move rule.

On digital chess boards, a draw is often shown by placing the white king on e4 and the black king on e5. If that position were legal, it would be a literal example of 1/2-1/2 under these rules.

[lkaufman]

A proposal I made in the TCEC chat is to take the earlier idea that a king reaching the opposite rank wins and connect it to the following draw scoring proposal.

When the game does not end in mate, or by the king reaching the opposite side, then each player receives 1/8 of a point for every rank their king has traveled to. In addition, a king moving to a rank which it hasn't previously reached resets the 50 move rule.

On digital chess boards, a draw is often shown by placing the white king on e4 and the black king on e5. If that position were legal, it would be a literal example of 1/2-1/2 under these rules.

So to make that a two-result version, the rule would be that in case of a draw by normal rules the side who had reached the more advanced rank with his king gets the win, and if that is also a tie Black wins. I don't think that resetting the fifty move rule for king advancement would be necessary in this version, although it is an option. I suspect that the game is a win for White with perfect play, but of course it could be a win for Black.