Some 40/400 Quad Naum 3 vs Quad Rybka 2.3.2a games

[Spock]

A couple of new one-off projects, see here:

http://kirr.homeunix.org/chess/discussi ... f=7&t=3177


Rybka currently leads 5.5 - 4.5 (that is 9 draws, and one win for Rybka)

10 games and 9 draws....

[Ovyron]

A couple of new one-off projects, see here:

http://kirr.homeunix.org/chess/discussi ... f=7&t=3177


Rybka currently leads 5.5 - 4.5 (that is 9 draws, and one win for Rybka)

10 games and 9 draws....

Is this the expected result of a Rybka-Rybka match or is Naum the drawish engine?

[Uri Blass]

A couple of new one-off projects, see here:

http://kirr.homeunix.org/chess/discussi ... f=7&t=3177


Rybka currently leads 5.5 - 4.5 (that is 9 draws, and one win for Rybka)

10 games and 9 draws....

Is this the expected result of a Rybka-Rybka match or is Naum the drawish engine?

I suspect that it is partly a result of interface that adjudicate draws in the middle of the game for no clear reason only because both programs had 0.00 evaluation after 25.b4 Qc7 in game 1 or after 34...Nxc4 35.Qd3 in game 6 or after 42...exf2+ in game 8 or after 18...0-0 19.Bd3 in game 9


There is no justification to adjudicate games as draw only because both programs evaluate the position as 0.00


Uri

[Ovyron]

There is no justification to adjudicate games as draw only because both programs evaluate the position as 0.00

Yes, I have many games in where both engines were showing a 0.00 score for over 15 halfmoves yet these were decided games. If Ray's games were adjudicated like this then I think it should be fixed and the games replayed.

[Graham Banks]

Yes, I have many games in where both engines were showing a 0.00 score for over 15 halfmoves yet these were decided games. If Ray's games were adjudicated like this then I think it should be fixed and the games replayed.

Many people prefer to use the 'early draw' option in Chessbase GUIs or the 'etiquette' option in the Shredder GUI.
What Ray does is entirely his choice.

[Spock]

There is no justification to adjudicate games as draw only because both programs evaluate the position as 0.00


Uri

Yes - I'm very much aware of this. But almost every CCRL tester will have the chessbase GUI draw option turned on, or the Shredder GUI etiquette setting turned on. You'll find tens of thousands of early draws like this in both our 40/4 and 40/40 databases, and I suspect also in most other testing groups as well. It's something I hate about the UCI protocol, that the GUI makes can make decisions not the engines. As I understand it, the winboard protocol doesn't allow it... Yes it can be turned off, and I seriously considered turning etiquette off, but with 40/400 repeating time control you could get a single game lasting several days ! Another reason why I don't like repeating time controls, especially this long. No basis in reality. This is very much a one-off experiment, for these reasons

For my "game in 24 hrs" I have etiquette turned off. For those, the maximum length of the game is set. It's a much better option in my view.

[Spock]

I should also say - the 40/400 match will be most likely be terminated at 20 games, so I can concentrate on "game in 24 hrs" instead (with draw options turned off). I like the latter so much better