Rybka's Immortal Game (Part 2)
Today an article by a guest author: Jeroen Noomen, writer of the opening book of the computer world champion, Rybka.
Some time ago ChessVibes published the "immortal game" of the chess computer program Rybka, currently the strongest engine in the world. Although this game was quite beautiful, I think we have a new candidate that fits this description better. It was played in the WCCC (world computer chess championship) 2007 in Amsterdam. Rybka beat the dutch program Diep in an amazing game, full of spectacular tactics and difficult-to-find moves. After 29 moves the game was over, something that is very rare in computer games these days. Especially for ChessVibes I have analysed the new Rybka immortal game and a few other Rybka games from the WCCC.Since Rybka entered the chess computer scene in december 2005, its dominance in the chess computer world is amazing. Rybka won all its official tournaments, with the exception of the WCCC 2006 in Turin. Let's take a look at the tournaments that have been played after this worldchampionship in Turin:
Jeroen Noomen (l.) receives the trophy from arbiter and ICGA member Jaap van den Herik. As the worldtitle was still missing on Rybka's agenda, it was clear Vasik Rajlich and I (the book author) were very eager to clinch it in the WCCC 2007 Amsterdam tournament. For Vasik this meant to have the strongest possible Rybka version at the start. For me as bookexpert a period of 2 months of preparing began. To have a good book is not really easy, I spent many evenings and weekends improving the book. Not only the strongest moves have to be played, but Rybka must understand them and also improvements have to be found. What is very natural to a strong player doesn't necessarily means this is best for a computer program. Sometimes continuations that are dismissed by GM's are better for computer programs. To find a good balance between these aspects is the most difficult for a book expert.For the WCCC I had prepared the Najdorf Poisoned Pawn in a very thorough way. As soon as the WCCC pairings were known, I was sure to use this line in the last round against former worldchampion Shredder. I was not disappointed, as Shredder went for it, as I expected. Unfortunately the game was not really interesting: Shredder left its book far too quickly, after which I knew the game was going to be an easy win for Rybka. A pity I couldn't demonstrate the harsh refutation of Shirov-Ftacnik, so that line still remains a secret. Apart from 2 draws against Zappa (in which Rybka could have won) and Loop (in which Rybka was slightly worse) all other 9 games were won in solid style. What striked me most, was that all games were pretty much decided before move 30. The evaluation of Rybka played the decisive factor in these games. The question is: is there any program that is able to challenge Rybka's playing strength in the coming years? I think this is going to be tough. Rybka's programmers have the IM status, while other programmers are much weaker chess players. This is a very important factor. Vasik Rajlich has proven that there is a lot to gain in the evaluation of positions, not in improving tactical strength. Im my opinion the evaluation of other topprograms is often very weak. Therefore you get scores like Rybka-Deep Fritz 10 40.5-11.5 or Rybka-Deep Junior 10 43.5-6.5 like in the CCRL ratinglist. For us as chessplayers this is a good development. Since Rybka the strength of computer programs has been given a boost. Other programmers try to catch up, while Vasik Rajlich keeps on improving Rybka. We only need to wait and see to get even better analysis tools available. Because Rybka is not an opponent for human chess, but a very strong and helpful analysis tool that can bring chess to a higher level.Jeroen Noomen
Some time ago ChessVibes published the "immortal game" of the chess computer program Rybka, currently the strongest engine in the world. Although this game was quite beautiful, I think we have a new candidate that fits this description better. It was played in the WCCC (world computer chess championship) 2007 in Amsterdam. Rybka beat the dutch program Diep in an amazing game, full of spectacular tactics and difficult-to-find moves. After 29 moves the game was over, something that is very rare in computer games these days. Especially for ChessVibes I have analysed the new Rybka immortal game and a few other Rybka games from the WCCC.Since Rybka entered the chess computer scene in december 2005, its dominance in the chess computer world is amazing. Rybka won all its official tournaments, with the exception of the WCCC 2006 in Turin. Let's take a look at the tournaments that have been played after this worldchampionship in Turin:Dutch Open Leiden 2006: 9/9 (1st place) Paderborn 2006: 6.5/7 (1st place) CCT9 2007: 6/7 (1st place) ICT Leiden 2007: 7,5/9 (1st place) WCCC Amsterdam 2007: 10/11 (1st place)Thus in these 5 tournaments Rybka achieved 39 points from 43 games, a score over 90%. She (yes, Rybka is female) didn't lose a single game and only shedded 8 draws.
Apart from 2 draws against Zappa (in which Rybka could have won) and Loop (in which Rybka was slightly worse) all other 9 games were won in solid style. What striked me most, was that all games were pretty much decided before move 30. The evaluation of Rybka played the decisive factor in these games. The question is: is there any program that is able to challenge Rybka's playing strength in the coming years? I think this is going to be tough. Rybka's programmers have the IM status, while other programmers are much weaker chess players. This is a very important factor. Vasik Rajlich has proven that there is a lot to gain in the evaluation of positions, not in improving tactical strength. Im my opinion the evaluation of other topprograms is often very weak. Therefore you get scores like Rybka-Deep Fritz 10 40.5-11.5 or Rybka-Deep Junior 10 43.5-6.5 like in the CCRL ratinglist. For us as chessplayers this is a good development. Since Rybka the strength of computer programs has been given a boost. Other programmers try to catch up, while Vasik Rajlich keeps on improving Rybka. We only need to wait and see to get even better analysis tools available. Because Rybka is not an opponent for human chess, but a very strong and helpful analysis tool that can bring chess to a higher level.Jeroen Noomen
