Cursed win at TCEC

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

mwyoung
Posts: 996
Joined: Wed May 12, 2010 8:00 pm

Re: Cursed win at TCEC

Post by mwyoung » Fri Nov 18, 2016 12:08 am

syzygy wrote:
mwyoung wrote:
syzygy wrote:
Lets make an absurd example to show the point. I play GM Carlsen a 6 game match with 6 fixed and very unsound openings. So unsound I am able to win with white every game. And so is GM Carlsen.
And that is exactly what is not happening in TCEC.
I respect your opinion.

And I will let TCEC make my point.
It is not an opinion but a fact. There have not been any 1-1 openings so far.

Of course we might yet get openings that are so advantageous for one side that 1-1 is inevitable. But as hgm explained, even if that happens there is no damage. It will be as if TCEC had one round less. There would be no actual skew.
And I showed that it does matter. What you are arguing is that wins or losses forced onto the progam do not matter. As long as the other program could get a loss from the opening as well. And yet and the point of my post. Some go crazy because a loss was forced onto Houdini 5. When the exact same thing could happen to Stockfish 8 making it fair. That's why I think this whole thread is funny. Some here have such selective moral outrage.
Professing themselves to be wise, they became fools,

syzygy
Posts: 4171
Joined: Tue Feb 28, 2012 10:56 pm

Re: Cursed win at TCEC

Post by syzygy » Fri Nov 18, 2016 1:14 am

mwyoung wrote:
syzygy wrote:
mwyoung wrote:
syzygy wrote:
Lets make an absurd example to show the point. I play GM Carlsen a 6 game match with 6 fixed and very unsound openings. So unsound I am able to win with white every game. And so is GM Carlsen.
And that is exactly what is not happening in TCEC.
I respect your opinion.

And I will let TCEC make my point.
It is not an opinion but a fact. There have not been any 1-1 openings so far.

Of course we might yet get openings that are so advantageous for one side that 1-1 is inevitable. But as hgm explained, even if that happens there is no damage. It will be as if TCEC had one round less. There would be no actual skew.
And I showed that it does matter. What you are arguing is that wins or losses forced onto the progam do not matter. As long as the other program could get a loss from the opening as well. And yet and the point of my post.
You were saying skew. There is no skew.
Some go crazy because a loss was forced onto Houdini 5.
Nothing to do with pro- or contra-H5 or SF. There is just some amazement about the fact that a bug is acknowledged to be a bug, yet is not fixed even though the fix is straightforward and painless (no replay or anything necessary).

If your point is that the bug, just like the openings, introduces no skew, then nobody will dispute that. Certainly both engines could have been on the receiving end and if it strikes again it may strike H5 and it may strike SF. But that is not at all the reason for the length of this thread.

Uri Blass
Posts: 7993
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Cursed win at TCEC

Post by Uri Blass » Fri Nov 18, 2016 2:00 am

Lyudmil Tsvetkov wrote:
gladius wrote:
Houdini wrote:
whereagles wrote:Have a look:

http://tcec.chessdom.com/archive.php?se=9&sf&ga=17

Engines showing 0.00 due to 50-move rule, but position was auto-adjudicated as an M72 TB win :D

Discuss :)
Not much discussion possible.
Both engines know that it's a draw (0.00) and play accordingly.
Suddenly the GUI decides otherwise and is clearly not following the rules of chess as implemented in the engines.
It's kinda ridiculous, but not very important.
Agreed, it should be a draw. As I posted in the TCEC chat, the adjudication should match the result if the engines had played the position out. In this game, it was a 50 move draw.
I would not agree here with Gary.
A tablebase win is a tablebase win. The position is simply won for white, so why declare it a draw? If both engines assume it is 0.0, that is only their fault they still have not implemented the much more relevant 100-move draw rule instead of the well-outdated 50-move rule. (or, what is the longest tb win without captures/promotions/pawn move?)

I am not certain what FIDE says about the 50-move/100-move draw rule, but why should engines follow FIDE? Engines are at the cutting edge of progress and progress says abovementioned position is simply a win for the stronger side. It is simply time to implement longer draw rule than 50-moves.

That should be specified in some protocol though, I agree it was not quite fair to both Houdini and SF in terms of their lack of knowledge, but a win is a win.
1)If the tournament does not use the 50 move rule in 5 piece tablebases positions then you need to mention it before the tournament and TCEC did not do it(claiming that you adjudicate games in 5 piece tablebases is not enough when people do not know that the adjudication is not based on the common tablebases that consider the 50 move rule).

2)If the tournament does not use the 50 move rule in 5 piece tablabases positions then I see no reason to use it in 6 piece tablebases positions because it means that you accept engines to be non consistent and use one type of tablebases in 5 piece positions and another type of tablebases in 6 piece positions(in case the author try to win).

3)I also do not see a reason to change rules based on having more tablebases
and maybe tomorrow an engine may be able to build some new tablebases during the game in order to find a mate in 80(when the mate is practically a draw by the 50 move rule).

4)I think that if you change the 50 move rule for some future tournament the best solution is simply not to have the 50 move rule in case that one program claims a mate for itself. The program that claims the mate will also have to show some main line of all moves to mate when there is going to be an analysis after the game to confirm the result(confirming that the main line does not have obvious blunders).

User avatar
Guenther
Posts: 2223
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Contact:

Re: Cursed win at TCEC

Post by Guenther » Fri Nov 18, 2016 6:48 am

Dirt wrote:
syzygy wrote:It is not an opinion but a fact. There have not been any 1-1 openings so far.
Rounds 17 and 18 were the same opening and each was won by white. Not unexpected, of course.
Game 25 and 26 could be counted now. That opening looks a bit too 'whitish'.

User avatar
Laskos
Posts: 7884
Joined: Wed Jul 26, 2006 8:21 pm

Re: Cursed win at TCEC

Post by Laskos » Fri Nov 18, 2016 7:01 am

Guenther wrote:
Dirt wrote:
syzygy wrote:It is not an opinion but a fact. There have not been any 1-1 openings so far.
Rounds 17 and 18 were the same opening and each was won by white. Not unexpected, of course.
Game 25 and 26 could be counted now. That opening looks a bit too 'whitish'.
Ideally, for the "resolution" of the match, it's best to have equal number of 1/2-1/2 and 1-1 openings. For TCEC conditions, where draw rate with balanced positions can reach 90%, it almost rigorously proven. So, openings as of now, are a bit too balanced compared to ideal (six 1/2-1/2, one 1-1).

PS Wait, wasn't game 25 draw?

User avatar
Guenther
Posts: 2223
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Contact:

Re: Cursed win at TCEC

Post by Guenther » Fri Nov 18, 2016 7:14 am

Guenther wrote:
Dirt wrote:
syzygy wrote:It is not an opinion but a fact. There have not been any 1-1 openings so far.
Rounds 17 and 18 were the same opening and each was won by white. Not unexpected, of course.
Game 25 and 26 could be counted now. That opening looks a bit too 'whitish'.
Ouch, I mixed the numbers - it seems opening for 27/28 could be a such one.
28 is not yet finished though...

Code: Select all

1. d4 d5 2. c4 Nc6 3. Nf3 Bg4 4. cxd5 Bxf3 5. gxf3 Qxd5 6. e3 e5 7. Nc3 Bb4 8. Bd2 Bxc3 9. bxc3 Qd6
27: 10. Qb3 +0.81
28: 10. Qb3 +0.87

Let's see if Houdini can win this?

(I want to add that the other openings so far look good and interesting
and are not too disbalanced)

User avatar
hgm
Posts: 22081
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Contact:

Re: Cursed win at TCEC

Post by hgm » Fri Nov 18, 2016 8:38 am

mwyoung wrote:As long as the other program could get a loss from the opening as well.
'Could' is not the same as 'will'. It would be a mistake to use unbalanced openings selected randomly without having the engines play both sides. Even though each of the engine could equally likely draw a losing start position. It will add additional random noise to the result.

duncan
Posts: 6494
Joined: Mon Jul 07, 2008 8:50 pm

Re: fortress_draw_rule

Post by duncan » Fri Nov 18, 2016 11:07 am

hgm wrote:If the 50-move rule makes some end-games drawn that otherwise could be won, so what? There are plenty other rules that turn end-games into draws that would have been won if the rule did not exist. Stalemate, for example. All drawn KPK positions would have been wins if stalemate did not exist, and you would have to play until King capture. If we did not have the rule that Bishops would capture only diagonally, but they could capture to any adjacent square, you could checkmate in KBK. And if Bishops could move to the other color, many draws in end-games with unlike Bishops would become wins. And if the board was 9x9, you would not even have unlike Bishops to start with. Repetition draws are annoying too. Why not simply declare the side that causes a repetition the loser? Then even K vs K could still be won with proper play.

If we are not going to change any of those rules, why would the 50-move rule be obnoxious?
(1)50 move rule was made under a misapprehension
(2)it penalises the player in the better position

User avatar
hgm
Posts: 22081
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Contact:

Re: fortress_draw_rule

Post by hgm » Fri Nov 18, 2016 11:42 am

The stalemate rule punishes the player in the better position, in KPK. That you are allowed to play on with a bare King so that KBK is a draw punishes the player in the better position...

Ralph Stoesser
Posts: 408
Joined: Sat Mar 06, 2010 8:28 am

Re: fortress_draw_rule

Post by Ralph Stoesser » Fri Nov 18, 2016 12:18 pm

To remove the stalemate rule would be a major change. Most chess studies would be worthless. The character of the game would change too much. Additionally it would result in a change of the FIDE rules, but nobody here proposes to change the FIDE rules. The proposal is to overrule the 50 move rule in case of

1) pure engine competition
2) a tablebase checkmate has been found in current position

That would be a rather mild change that seems apropriate to me. Tablebases which include the 50 move rule make sense as long as humans are involved, for analysis and for standard otb matches where humans participate. In pure engine competition, engines could easily use tablebases without this information. Imho that would make the competition slightly more interesting.

Post Reply