Fritz, Rybka, or think for yourself after all?
25 January 2007 20:14 PM CET | Last modified: 17:38 | By Merijn van Delft | Filed under: Uncategorized | Tags: Chess improvement, Computer chess, Middlegame
In the glorious past grandmaster games still had something invincible, something you would look at with respect and would not dare giving your opinion about. Grandmaster games were something very complicated, something you should not show to beginners, since it would only confuse them. Since the rise of the computer, databases and engines, that is all gone. You can check in the database up to which point it was still theory and from that point on you can judge the players’ performance with any engine. The audience at home knows exactly what is happening from move to move and top players are complaining about the decline of respect.
It is even as bad as that nobody seems to think for themselves anymore. “Fritz says +1.37�?, discussion closed. That Black may still have some nasty little threats and that the white king is not completely safe, does not matter. I found myself explaining many times that computers are not perfect either and for example often change their evaluation of a piece saccrifice after a while, but I never sound very convincing. I know this pitfall very well, but I have no idea how to explain it.
And then I fell for it myself again. While writing the report on the Dutch League match Homburg Apeldoorn – SO Rotterdam about three months ago, I quickly checked the games with Fritz and wrote about the game below: “Siebrecht and Bosch played a bit of a sloppy game – after Jeroen missed a sudden winning chance and a bit later a sudden drawing chance as well, Sebastian decided the game in his favour.�?
Now it is January, I have switched to Rybka in the meanwhile and am currently collecting some positions. “What was that win again, that Black overlooked?�?, I kept wondering, without being able to find. Then it said click and I realised that Fritz, based on several bonusses that it awards (like king safety, heavy pieces close to the opponent’s king, passed pawns, space advantage), fairly randomly evaluates the position as a win for Black without actually “seeing�? a concrete win. Rybka and other engines think that the position is about equal, which definitely looks like a more realistic starting point. Here we hit on the problem with the hero engine of the Microsoft of chess: while every human being is thinking “no idea what the position is like, looks like nothing special, I guess it is about equal�?, Fritz is telling you that one of the sides is close to winning already. It reminds us a bit of analysing in the pub: first telling everybody that White is winning and then having a look at the position. That is why an engine like Rybka is a relief: unlike really something is going on, the evaluation is always around zero. But most of all I learned my lesson: always think for yourself.
sorry for that and thanks for the thought provoking ideas!...
Ja, mij zeker bekend… moeilijke stellingen kan Fritz niet eens 4/5 zetten diep (Fritz die bij Chessbase 9 inzit) en deelt +’en en -’en uit van zet naar zet. Ik moet eerst zelf de goede weg vinden en als Fritz dan eindelijk iets echt geforceerds heeft kun je de variant weer opnieuw checken, en weer opnieuw checken. Misschien toch maar kijken voor een dedicated computer met een supersnelheid D:
Yes … Rybka has taken chess to a new level. It is programmed by an IM chessplayer … so it very accurate in its understanding of the small differences in a chess position. It is amazing to see how quickly Rybka will destroy ANY other engine. It plays like a GM playing with computer aid.
I did show though that certain examples from real life games are completely misevaluated by engines, including Rybka. Amongst the examples a gave on PlayChess, I used a correspondence game and an OTB game of my own (which is quite ironic, since the cause of “disrespect” to the GMs couldn’t even evaluate right a position between 1800 players and actually ends up blundering a whole piece in an equal position).
So I guess that supports Merijn’s last sentence, Xmas!
Yes, engines aren’t perfect… but the point is that rybka evaluates more positions “right” than fritz, so I would suggest to think first and then to use Rybka
Sometimes it’s the other way around though.
A quick scan of the position in this post with most computers should say 1. Ng4 is around =, maybe at most +=. If you make certain suggestions, meanwhile, it looks like 1. Ng4, 1. Bf3 and especially 1. e5! lead to forced wins for white. Rybka apparently gets there independently though after hours on ultra-optimistic setting.
It is also much more fun (yes, chess can be fun as well) trying to guess-the-move without using an engine while watching/analyzing GM games. It gets really annoying when kibitzers keep yelling ‘draw’ in equal positions.
I agree Frank.
‘1. Nf3 - draw!’
With heavy theoretical discussions like the Najdorf games in Wijk aan Zee I tend to enjoy looking for “the truth” using databases and engines. The interaction of computer engine and human being can be very powerful indeed. But otherwise I agree that simply watching the games and guessing the move is most fun.
wat is die kramnik een sukkel.mat in een overzien
en dan nog 500.000 KRIJGEN OOK.
en dan was het nog niet eens rybka.
GROETEN TOONTJE
Dear chess player,
Fritz-10 is a versatile programme which has beaten world champion -VLADIMIR KRAMNIK.
I do not think wheher RYBKA has beaten any chess champion of world ranking till the moment of penning this comment?????
I have always bought FRITZ & really planning to buy a copy of RYBKA for study purpose the kind of functions/coachings that it offers to an ordinary chess player like me.
By the way-any grandmaster coachings in RYBKA like fritz-10??????
rgds/prasad lbv
bij zet 34… Qa7-g1 ziet er ook zeer dreigend uit, of 34… Kh7.. save position, lijkt mijn om een kans te maken tegen wit.
Think like a beginner
Dear Prasad,
Rybka indeed has not beaten any world champignon. The reason therefore is simple, Chessbase has a monopoly in playing grandmaster versus engine. In the last engine topmatch Rybka was refuged. Yet Rybka beats fritz 10 and 11 and other top engines Alphia time after time. Analysing with fritz and rybka shows better endgame and positional understanding fore Rybka. Its weakness is kings safety. Main rule is. Don’t count blindlessley on a chess engine but use common sense.
Greetz Kiteman
Dear Kiteman,
Thanks for the enlightenment on RYBKA Engine.
In a place like India, its rare to get world’s best chess softwares like fritz/rybka etc etc.
However i managed to buy fritz-10 through a well known famous shop which sells chess products.
View RYBKA being able to match & even beat LEGEND Softwares like fritz-10…some grandmaster should volunteer playing against rybka to see the outcome,preferably VLADIMIR KRAMNIK himself????????
Txs once again for response.All the best in chess life & likewise.
best regards/prasad lbv
Dear Kiteman,
Thanks for the enlightenment on RYBKA Engine.
In a place like India, its rare to get world’s best chess softwares like fritz/rybka etc etc.
However i managed to buy fritz-10 through a well known famous shop which sells chess products.
View RYBKA being able to match & even beat LEGEND Softwares like fritz-10…some grandmaster should volunteer playing against rybka to see the outcome,preferably VLADIMIR KRAMNIK himself????????
Txs once again for response.All the best in chess life & likewise.
best regards/prasad
what a great bunch of top gms! everyone seems very happy to be playing and are all very cordial to each other (maybe cause the games are not rated!) with the exception of kramnik and topalov (who just need to kiss and make up). it seems like they are having a great time doing what they do best. great videos, great coverage!