xboard 4.3.15 draw bug

[bob]

I think the solution I originally proposed was quite logical. But it was rejected.

For the case at hand, (ICS play) nothng was changed, b.t.w. The problem is that the solution to the race condition you coocked up with the ICC site maintainer requires the claim to be sent before the move, which to most people is counter-intuitive. And that the ICS calls the draw claim a draw offer, and refuses it.

This is purely a problem of ICS protocol. You are barking up the wrong tree.

I would be willing to bet that if a reasonable proposal was submitted, Sleator and friends would be willing to add yet another "style" that would deal with draw claims and draw offers that are made in a way that is 100% compliant with FIDE rules. But it seems that no one is interested in correcting a shortfall in how it works at present. And so long as it works, I suppose one could say it is "OK". But these threads are going to continue to pop up from time to time because the winboard protocol does not follow FIDE rules, and ICC has adapted to make it work as well as possible.

[hgm]

How ICC handles the problem is in no way affected by WinBoard protocol. It is their decision to not distinguish between draw claims and draw offers, and use the same command "draw" for both.

They could have decided to use atomic claiming of draw-after-the-move, through a command "Kf8 draw" on a single line. But in stead they require a command to be sent that is indistinguishable from a draw offer, and will in fact be interpreted as a draw offer, and possibly relayed as such to the opponent, and even rejected by him, before you give the move.

All this has absolutely nothing to do with WinBoard protocol. It is how Humans must claim draws, even if they are playing through telnet.

[bob]

How ICC handles the problem is in no way affected by WinBoard protocol. It is their decision to not distinguish between draw claims and draw offers, and use the same command "draw" for both.

They could have decided to use atomic claiming of draw-after-the-move, through a command "Kf8 draw" on a single line. But in stead they require a command to be sent that is indistinguishable from a draw offer, and will in fact be interpreted as a draw offer, and possibly relayed as such to the opponent, and even rejected by him, before you give the move.

All this has absolutely nothing to do with WinBoard protocol. It is how Humans must claim draws, even if they are playing through telnet.

If you re-read my post, you will see I don't disagree with you on that point, and I pointed out that given a FIDE-compliant solution, I'd bet sleator and group would be interested in solving this.

This doesn't mean that the current way this is handled on either side is the _right_ way, it is just a way that works but also introduces a component of confusion. It is this "confusion" that can be solved. But it will take some effort and coordination. I'm not one to subscribe to the mantra "it works, so there's no need to change it" when there is an obvious issue that requires a programming kludge to handle the right way in a text window, and a different way through a GUI.

[sje]

I would be willing to bet that if a reasonable proposal was submitted, Sleator and friends would be willing to add yet another "style" that would deal with draw claims and draw offers that are made in a way that is 100% compliant with FIDE rules.

Must we be 100% FIDE compliant? In the past half century, has FIDE ever been useful for computer chess? About as useful as teats on a boar, I'd say.

As for a new style, let there be only a single modification: a threefold position repetition immediately ends the game with a draw.

[hgm]

I built the option to have XBoard make such immediate draw adjudications in local engine-engine mode.

But it might be a good idea to extend this functionality to ICS mode. By having WinBoard automatically send a claim to the ICS when such a repetition occurs. Just like you can set it to automatically call the flag, when the opponent is out of time.

[Matthias Gemuh]

Must we be 100% FIDE compliant? In the past half century, has FIDE ever been useful for computer chess?

Hi Steven,

I and a few others have been saying exactly the same thing over and over.
Bob, Miguel and a few others swear by FIDE whenever that point is mentioned.

Matthias.

[bob]

I would be willing to bet that if a reasonable proposal was submitted, Sleator and friends would be willing to add yet another "style" that would deal with draw claims and draw offers that are made in a way that is 100% compliant with FIDE rules.

Must we be 100% FIDE compliant? In the past half century, has FIDE ever been useful for computer chess? About as useful as teats on a boar, I'd say.

As for a new style, let there be only a single modification: a threefold position repetition immediately ends the game with a draw.

If you ever want to play in a human event, you have to be FIDE compliant. Cray Blitz and now Crafty have played in dozens of human events, and you have to get the draw offer, draw claim, draw acceptance, etc right.

[michiguel]

Must we be 100% FIDE compliant? In the past half century, has FIDE ever been useful for computer chess?

Hi Steven,

I and a few others have been saying exactly the same thing over and over.
Bob, Miguel and a few others swear by FIDE whenever that point is mentioned.

Matthias.

Call me a conservative, but I still think that we should keep castling and en passant

Miguel

[bob]

Must we be 100% FIDE compliant? In the past half century, has FIDE ever been useful for computer chess?

Hi Steven,

I and a few others have been saying exactly the same thing over and over.
Bob, Miguel and a few others swear by FIDE whenever that point is mentioned.

Matthias.

The problem is that FIDE rules are the rules used in human events. And I still play in one here and there. If I chose to eschew FIDE rules, I am going to lose games I should have won, miss draws that I claimed incorrectly, etc. That's the reason I mention FIDE, because I still want to play in human events in addition to the computer-only events we have.

[Matthias Gemuh]

Must we be 100% FIDE compliant? In the past half century, has FIDE ever been useful for computer chess?

Hi Steven,

I and a few others have been saying exactly the same thing over and over.
Bob, Miguel and a few others swear by FIDE whenever that point is mentioned.

Matthias.

Call me a conservative, but I still think that we should keep castling and en passant

Miguel

Your statement can mislead some readers into thinking I once suggested that castling or en passant should be dropped from computer chess.
I just want to clarify for readers that I NEVER suggested such.

Matthias.