Rybka will play Chess960 in Mainz 2007 !

[Spock]

I'm surprised Jorge hasn't noticed this already.....

http://rybkaforum.net/cgi-bin/rybkaforu ... id=1343#fp

[Tony Thomas]

I'm surprised Jorge hasn't noticed this already.....

http://rybkaforum.net/cgi-bin/rybkaforu ... id=1343#fp

You are an instigator Ray. If you hadnt told Vas that Hiarcs is currently number one on the list, he probably would have made a FRC version a lot slower.

[Spock]

Well, I think it is more a case of Rybka being able to win another World Championship to add to it's marketing credentials. In the real world though it will mean very little unfortunately, because I think chess960 is dead, or heading that way

[pichy]

Well, I think it is more a case of Rybka being able to win another World Championship to add to it's marketing credentials. In the real world though it will mean very little unfortunately, because I think chess960 is dead, or heading that way

I don't think is quite that simple, simply because Rybka plays extremely good at standard chess, doesn't necessarilly mean that it can play great at Chress960. Remember what happened to shredder the first time that it participated in Mainz, it lost to Spike 1.2 which at that moment it was rated more than 75 rating points below Shredder in standard chess

PS: Now that Anand will play Chess960 at Mainz, FRC will raise to a new level of acceptance worldwide. But like I mentioned before, being number #1 in standard chess doesn't mean that you will automatically be the best Chess960 player, and that apply to computer as well

[Spock]

I don't think is quite that simple, simply because Rybka plays extremely good at standard chess, doesn't necessarilly mean that it can play great at Chress960.

Agreed, but from what I read on the Rybka forum they will not be simply implementing the castling rules, but adding chess960 specific code as well. It will be interesting for sure !! Maybe they can make use of our ~ 10,000 game database to look for potential optimisations

[pichy]

I don't think is quite that simple, simply because Rybka plays extremely good at standard chess, doesn't necessarilly mean that it can play great at Chress960.

Agreed, but from what I read on the Rybka forum they will not be simply implementing the castling rules, but adding chess960 specific code as well. It will be interesting for sure !! Maybe they can make use of our ~ 10,000 game database to look for potential optimisations

Yes, they could look at your 10,000 game database, but the programmers of Shredder and Spike has been implementing chess960 algorithms longer and they too can look at your available 10,000 game database

[Eelco de Groot]

I thought that there wouldn't be a Chess960 computer-tournament in Mainz this year but it seems this time due to financial constraints it is decided to hold an invitational tournament, hoping to go back to the usual format that many programmers were very fond to play in, next year.

Mark Vogelgesang posted on http://f51.parsimony.net/forum204700/messages/130.htm

written by Mark Vogelgesang am 23. Mai 2007 11:33:45:

Due to financial constraints, the Chess Tigers will not be able to organise the 2007 Chess960 Computer World Championship in the format the event was held in 2006 and 2005.

Nevertheless, we would like to continue the tradition and hold a World Championship event in 2007. We are therefore planning to hold a smaller tournament with four programmers or teams, which will be invited to the event. This will allow us to continue with the tradition and keep the sponsors and the public interested, and to determine a World Champion for the third time. We will announce the details of the invitational tournament to determine the 2007 Chess960 Computer World Champion shortly.

It is our goal to return in 2008 to the format which we used successfully in 2005 and 2006.

It will nice to see a Rybka that can play Fischer Random. A tip for Vas to try: in many Chess960 openings I think it is much more about tactics than about positional play. So experimenting with Rybka's optimistic outlooks or Winfinderlike tactics wouldn't hurt. And sometimes the positions get so wild it is just better to forget about castling altogether... I believe that Tord and Glaurung are doing something like the following; giving a high bonus to having castling rights so the program won't castle too soon but at the same time will still strive to be able to castle if that can help in tactics for instance.

Wellmeant advice but at the same time I hope it will take Vas at least half a year to get this right and improve Rybka's Chess960 skills so the other programmers get a chance to catch up

Eelco

[Uri Blass]

Well, I think it is more a case of Rybka being able to win another World Championship to add to it's marketing credentials. In the real world though it will mean very little unfortunately, because I think chess960 is dead, or heading that way

I don't think is quite that simple, simply because Rybka plays extremely good at standard chess, doesn't necessarilly mean that it can play great at Chress960. Remember what happened to shredder the first time that it participated in Mainz, it lost to Spike 1.2 which at that moment it was rated more than 75 rating points below Shredder in standard chess

PS: Now that Anand will play Chess960 at Mainz, FRC will raise to a new level of acceptance worldwide. But like I mentioned before, being number #1 in standard chess doesn't mean that you will automatically be the best Chess960 player, and that apply to computer as well

I think that knowledge in FRC is overestimated.

Movei has no FRC knowledge and Movei383 has only 301 elo less than shredder10
and 252 elo less than spike1.2

For comparison in normal chess the same 383 version has 225 elo less than shredder 10 and 197 elo less than spike1.2

I believe that specific FRC knowledge give clearly less than 100 elo and the gap between rybka and second best seems to be more than 100 elo.

Uri

[pichy]

I thought that there wouldn't be a Chess960 computer-tournament in Mainz this year but it seems this time due to financial constraints it is decided to hold an invitational tournament, hoping to go back to the usual format that many programmers were very fond to play in, next year.

Mark Vogelgesang posted on http://f51.parsimony.net/forum204700/messages/130.htm

written by Mark Vogelgesang am 23. Mai 2007 11:33:45:

Due to financial constraints, the Chess Tigers will not be able to organise the 2007 Chess960 Computer World Championship in the format the event was held in 2006 and 2005.

Nevertheless, we would like to continue the tradition and hold a World Championship event in 2007. We are therefore planning to hold a smaller tournament with four programmers or teams, which will be invited to the event. This will allow us to continue with the tradition and keep the sponsors and the public interested, and to determine a World Champion for the third time. We will announce the details of the invitational tournament to determine the 2007 Chess960 Computer World Champion shortly.

It is our goal to return in 2008 to the format which we used successfully in 2005 and 2006.

It will nice to see a Rybka that can play Fischer Random. A tip for Vas to try: in many Chess960 openings I think it is much more about tactics than about positional play. So experimenting with Rybka's optimistic outlooks or Winfinderlike tactics wouldn't hurt. And sometimes the positions get so wild it is just better to forget about castling altogether... I believe that Tord and Glaurung are doing something like the following; giving a high bonus to having castling rights so the program won't castle too soon but at the same time will still strive to be able to castle if that can help in tactics for instance.

Wellmeant advice but at the same time I hope it will take Vas at least half a year to get this right and improve Rybka's Chess960 skills so the other programmers get a chance to catch up

Eelco

Lets say that the invited programs are Rybka, Shredder, Spike and Hiarcs or Glaurung. The program's winner should play the human's winner of the Chess960 championship in a a match of 3 games of 60 minutes per side.

PS: I believe that at 60 minutes or longer the human has the edge in Chess960. But at blitz of less than 30 minutes per side any of the mentioned programs would be crunching too many possible moves in such a short time, while the human needs more times to figure out a plan that could equal all those calculations done by those programs evuluating thousands of possible moves per seconds

[Spock]

I find it odd that the 4 engines invited were Shredder, Rybka, Jonny and Spike. Why no Hiarcs ? Although it hasn't played FRC in any tournaments, it is number 1 on our FRC ratings lists here:

http://www.computerchess.org.uk/ccrl/404FRC/

So I would have thought it almost certain that Hiarcs would have been invited. A pity, assuming Hiarcs could attend of course.