A few weeks ago FIDE surprised with the announcement of a computer match between Fritz and Junior with a prize fund of $100,000. It's six games and will take place in Elista alongside the second round of candidates matches, starting tomorrow. Apparently the games will be broadcast live on the official candidates site as well.
IM Vasik Rajlich is the programmer of the chess program Rybka, which has dominated the comp-comp rating lists since its appearance a year and a half ago. He has issued a winner-take-all challenge to FIDE and the winner of their "Ultimate Computer Chess Challenge 2007." To top it off, he offers to spot Rybka's opponent a full point in a 24-game classical match. You can read the full challenge announcement at the Rybka site.
Since both Junior and Fritz are ChessBase products and commercially available, this is a PR event from top to bottom. And there's really nothing wrong with that, by the way. When X3D Accoona hired Irina Krush, Almira Skripchenko, and Zhu Chen to play rapid chess it wasn't with their Elo ratings in mind. (You do have to watch out for slapping "world championship" on every event, however. "Ultimate" has a good pro wrestling vibe to it.) If FIDE thinks this is good promotion – and they say it's "sponsored by FIDE and the head of the Republic of Kalmykia, FIDE President Kirsan Ilyumzhinov" – that's just another curious investment decision from Kirsan, I suppose. (And I'm always skeptical of these FIDE prize fund announcements anyway.) We'll see if it generates any mainstream buzz. If it does, is that good if they mention a comp-comp match and not the candidates matches that qualify for the world championship?
Of course for purists and afritzionados it's a little annoying not to have the top-rated engine participate in such a high-profile match. There was no qualifier other than a good working relationship between FIDE and the top chess software company, ChessBase. Again not a problem in my eyes since it's a PR match and the general public hasn't heard of any chess computer other than Deep Blue, if that. Chess programming is a business, too. As I pointed out when it was announced, anyone with a few hundred bucks could host a Junior-Fritz match at home. (Or a Rybka-Fritz-Junior tournament to 100 rounds.) It's not as if six games is going to establish supremacy. Unlike the candidates matches, there is no next step. Unless of course FIDE takes Vasik Rajlich up on his challenge!
I do like the idea of the candidates and computer match in parallel. It gives us an opportunity to examine the quality of comp-comp chess as it stands today. It will be interesting to see how obvious it is, if it is, which game of the five is between Fritz and Junior. Hint: if it's a 22-move draw it's between the humans.
I find this bizarre beyond all telling. No-one I know has the slightest interest in comp-comp games.
Two corrections:
First of all, this isn't Kirsan's entry to computer chess, as Mig well knows, as FIDE was behind the February 2003 match between Garry and Deep Junior which X3D sponsored.
Secondly, X3D was the sponsor for Kasparov-Karpov, Kasparov-Deep Junior, Kasparov-X3D Fritz, but the Almira-Zhu Chen-Irina matches and the Kasim-Fritz game were both Accoona sponsored.
(Nitpicky, but I have a horse in the fight, what can I say.)
And yeah, Vas is proving that a strong player as a programmer does make a very big difference.
Pretty silly by Rajlich to be honest. "My program is stronger! My program is stronger!" Yeah, and?
No qualification process? "Bizarre" choice of opponents? "Principles of sport and fairness"? "Democratic norms"? What does any of that have to do with anything? They simply decided, for commercial purposes, to play a match between those engines. Is that not allowed now?
I just noticed that Bill Goichberg posted an update regarding the candidates for USCF board at www.checkmate.us. Mig, any plans for articles on this?
Rybka will easily be defeated if such a match were to take place. The last time it played in the Computer Chess Championship, it lost to like Shredder and only drew Junior. It is an inferior program in every respect, with hype that is only rivaled by Hydra.
Parsnips said:
"Rybka will easily be defeated if such a match were to take place. The last time it played in the Computer Chess Championship, it lost to like Shredder and only drew Junior. It is an inferior program in every respect, with hype that is only rivaled by Hydra."
Heh, we really need this Rybka-Fritz match! If for nothing else, then just to see the afritzionados battle it out with the rybkanatics here in the comments on Mig's blog :-)
I assume that the players themselves will be allowed to walk around and have a look at how the other human games are going but not be able to view the battle of the machines...
It will be interesting to see the choice of openings of J&F and also how much time they use over moves and length of their games...
and also the draw %age though this is a short match.
I find being a fan of computer program ridiculous, but what parsnips said goes beyond ridiculous (or was it irony?). It doesn't matter what happened in one particular game in one particular event (Rybka losing to Shredder). It's well known based on results of hundreds of games on equal hardware that Rybka is about 100 Elo points superior than any chessbase program. So making statements like "inferior program in every respect" is a bit rich. Or is it a voice of chessbase?
Now for acrice's comment. Silly by Rajlich. I don't think so. If that was a chessbase promo, that one thing. But this is an official FIDE event that has the WC title. Since Rybka is indisputably the strongest program, Rajlich has every right to be irked by it's exclusion. I would be too if I were him.
Computer chess always brings this neuro-science stuff into my mind...
Just had an interesting idea about Pavlov's dog experiments regarding to chess. I don't know if this a new or an old idea. "Each time Pavlov fed the dog, he rang a bell, and at each feeding the dog's mouth watered. Then after many repetitions of the combined food bell stimulus, Pavlov rang the bell but did not feed the dog. The animal reacted to the bell alone just as it had previously reacted to the sight of food its mouth watered."
The reason humans can do so well against computers is that we actually are Pavlov's dogs. When we see something familiar on the board, it's like bell ringing and our "mouth waters". It's very reflex-like basic function that gives humans a great power and learning capacity.
You could also say that chess is a "neurotic symptom" which is systematically developed. But in the end it's no different from any other sport, like ice-hockey, football, soccer or tennis - they all require these reflex-like responses, which are mostly learned but partly improvised during the game.
Looks like you nearly got your 22-move draw: RBN.. v RBN and fairly symmetric. RYBKA as commentator (!) says +0.25
Agree with Osbender, because of the FIDE involvement. In this case there IS something wrong with the comp-comp match being just a PR event. The reason? Because it's sponsored by FIDE, and being staged during the candidates matches. Because of these two aspects it's just too easy to have two misconceptions about this match.
1) It automatically gives the match a big "computer world championship match" connotation. If it will generate mainstream buzz, the people will conclude that at the moment Junior and Fritz are apparently the strongest programs, while there are at least two stronger ones. And then that catchy word "ultimate" in the title, of course all for PR, but so untrue. 2) The second idea people will get from this match being staged in FIDE's backyard, is that it really looks like these two programs have qualified for this final match. But there was no qualification at all, and still the ICGA is involved, although they're running their own self-titled "World Computer Chess Championship" at the same time. It's all just too confusing and strange and untrue, this pr event.
Following truth exposers comments I followed the link provided and was surprised to learn that Susan Polgar married her manager Paul Truong last december but did not reveal that fact to chess Life readers as they are both standing for election to the USFC board on the same slate. Nothing wrong with a husband and wife team standing but probably a good idea to let the voters know!
Great to see FIDE assist the marketing of ChessFluff.com-products !
Kind folks, have I got my wires crossed or is the cadidates 2nd round today?? If so, why is there no noise about it??
Is it just me or Mig nowadays has started really sounding anti-Chessbase? :-P
Kannan - Is it just me or Mig nowadays has started really sounding anti-Chessbase? :-P
Nope, just stating the bleeding obvious. What a non-event this is.
I just hope that Vasik will get his match with the winner of the "Ultimate Computer Chess Challenge 2007". After everything that Vasik has achieved with his Rybka program he surely deserves a chance for a match.
I do think that Vasik has gone a bit crazy though offering odds to the winner of the Fritz-Junior match. Both of these are quality programs and will not be a walk-over for Rybka.
Is there a rapid tiebreak if J&F tie at 3 all?
If not, my prediction is for Junior to win 3.5 - 2.5 to be followed soon by the launch of Junior 11
Am I the only one confused by Fritz's participation in this "event"? Didn't Fritz's developers say over a year ago that they were not going to participate in any more computer-computer competitions and that they were trying to develop Fritz so that it could better play against humans rather than other machines?
"I do think that Vasik has gone a bit crazy though offering odds to the winner of the Fritz-Junior match."
Not crazy, just arrogant. Rybka would win, but that's not how you behave. "Hey you, I'm so good that I can beat you no matter what advantages you get. Want to see or are you scared?"
To rdh: if you don't know anyone interested in comp-comp games then you need to get out a little more. I am almost exclusively interested in comp-comp games because many of them are at a super-GM level and for every one human super-GM game that is played every year there are hundreds if not thousands of comp games available online. And I might add, advancing technology and improving engines are making the game quality of comp games climb steadily year by year.
As for Rajlich's offering of point-plus-draw odds, this isn't arrogance. It's a deliberate insult that he hopes will get their attention and generate PR buzz. Indeed, as another poster said, this is pure pro wrestling. Apparently this open letter did get ChessBase's attention and they have no intention of accepting the challenge. I don't blame them; it's bad business to lose, and right now it seems probable that they'd be trounced.
@Osbender
Laughable. First we had Hydra, which lost to 1600's and 500 dollar subnotebooks running fritz (In freestyle chess)
Now you're trying to claim that a computer which hasnt won any competitions it has entered the 'strongest' computer...
I know a bulgarian manager who could use your help to spin toiletries..
Parsnips, you're either a silly troll or just extremely clueless about computer chess. Rybka has won 7 of the last 8 major computer events and leads every computer rating list by 100-150 ELO.
Well, I just like to look at the actual results of say SSDF
Rybka 2.3.1 Arena 256MB Athlon 1200 MHz, 2962
Opponent Result
Rybka12 A1200 24½-21½
Hiar111 A1200 12½-12½
Juni101 A1200 24-16
Fruit22 A1200 27-13
Spike12 A1200 33½-6½
CT 2007 A1200 30-10
CT 15.0 A1200 17½-3½
So the highest rated program, plays the most games against it's predecessor, and ties a match with Hiarcs, beats Junior... and then dominates weak engines for the majority of it's games.
It's the Morozevich of computers! Highly rated, no results!
GO Rybka!!!!!
An interesting way of thinking parsnips, but unfortunately completely wrong.
Suppose the following would be true of Anand: plays the most games with Topalov (and beats him), ties a match with Kramnik, beats Morozevich. And then dominates weaker players (Gelfand, Kamsky, Bacrot, that kind of people) for the majority of his games.
Would you say he is highly rated with no results? In reality Anand is even weaker according to you, because he did not beat Topalov or Morozevich, and did not tie a match with Kramnik. He "only dominated weaker players".
Where is Hydra in all of this? Has it dropped out of the competition for the top programs, or is it simply not playing any more? It seems like anything could crush Michael Adams 5 1/2-1/2 should have at least been mentioned.
According the SSDF's own data, Rybka plays itself for the majority of the games, plays weak computers, and has much fewer games played overall (thus higher margin of error.. in fact the 2nd highest in the list) compared to all other engines on the site.
Not to mention the fact that in fair, unbiased, scientific competition (the WCCC) Rybka has finished only 3rd place.
It is the Morozevich of computers. Highly rated. No remarkable results.
It seems that however indisputable and clearcut facts are, there will always be some one trying to spin them. Yes, that's you parsnips, from the school of Iraqi information minister.
First, are you trying to suggest that SSDF and all other well established computer lists are biased and non-scientific? Laughable.
Second. WCCC where Rybka finished 3rd is 11 games total, one game against each opponent with no equality in hardware. Against that you have multiple rating lists with hundreds games for each engine including long face to face matches, on equal hardware. Anyone knowing anything about statistics knows that calling WCCC results "scientific" is laughable.
"According the SSDF's own data, Rybka plays itself for the majority of the games, plays weak computers, and has much fewer games played overall (thus higher margin of error.. in fact the 2nd highest in the list) compared to all other engines on the site." This is a fine mixture of outright lies (i.e. plays itself for majority of the games, plays weak engines), and weasel spin offs (yeah, Rybka 2.3 didn't play that many games in SSDF compared to other engines, but Rybka 1.2 is in the second place there and it played enough.
Moreover, SSDF is not the only well established rating list, see, for instance, CCRL: http://computerchess.org.uk/ccrl/4040/ (btw there are lovely scores there Rybka 2.2 - Deep Fritz 10 +19-0=11, Rybka 2.2 - Deep Junior 10 +23-2 = 6).
It's a scientific fact that Rybka is way stronger than Junior and Fritz.
I do think that Vasik has gone a bit crazy though offering odds to the winner of the Fritz-Junior match. Both of these are quality programs and will not be a walk-over for Rybka.
-- Posted by: Matt Helfst at June 6, 2007 07:44
Yes, it -will- be a walkover for Rybka. Go checkout the results between Rybka and Fritz/DJ at the SSDF, CEGT, etc. websites. Rybka typically beats these engines by margins of like 33-17 in 50 game matches.
Why do you think Rajilch has the confidence to offer odds of a whole game? Because he already knows what the result is going to look like. On average, he probably expects to win a 24-game match by a score like 16-8.
Moreover, SSDF is not the only well established rating list, see, for instance, CCRL: http://computerchess.org.uk/ccrl/4040/ (btw there are lovely scores there Rybka 2.2 - Deep Fritz 10 +19-0=11, Rybka 2.2 - Deep Junior 10 +23-2 = 6).
It's a scientific fact that Rybka is way stronger than Junior and Fritz.
-- Posted by: osbender at June 6, 2007 22:55
Wow, that's even worse than I had imagined. It beat DF 10 by a score of 19-0 in 30 games! That's 24.5/30 or 81.7%!!! In a 24-game match that would be 20/24 or 20-4 with 16 Rybka wins versus 8 draws. GODDAMN that's an ass-kicking.
Combined that's 42 wins for Rybka, 2(!) for DF and DJ combined(!).
How does Parsnips explain away this fact? Oh yeah, I forgot. He'll just quote that WCCC performance of the BETA(!) Rybka which finished 3rd. :-)
"If you don't know anyone interested in comp-comp games you need to get out a little more"
A little less, surely?
rdh!
If Rajlich is serious about proving something, here is something to challenge: Glaurung 2- epsilon, an open source engine.
http://www.computerchess.org.uk/ccrl/404.live/rating_list_all.html
(with less "something", of course)
Parsnips said:
"Rybka will easily be defeated if such a match were to take place. The last time it played in the Computer Chess Championship, it lost to like Shredder and only drew Junior. It is an inferior program in every respect, with hype that is only rivaled by Hydra."
It's a beta version that Deep Junior drew and Shredder won. The beta version was named Rajlich not Rybka.
I doubt they will take Vas up on his offer. Fritz would have no chance.
"Not to mention the fact that in fair, unbiased, scientific competition (the WCCC) Rybka has finished only 3rd place. It is the Morozevich of computers. Highly rated. No remarkable results."
Posted by: parsnips at June 6, 2007 18:33
Well parsnips, Rybka now stands 9/10 in WCCC 2007. With one game remaining (against Shredder, currently 7/10), Rybka can finish no worse than shared first. Do you still stand by your words? Or to eat your words, with a bowl of humility?
Better yet, tell us who you really are. Your blindness speaks of a visibly hidden agenda.
Junior's participation in that match was explained by the fact that it WAS the reigning computer chess world champion (of 2006). A few hours ago, Rybka won against Shredder in the last round of this year's WCCC in Amsterdam, and now is the new computer chess world champion.
http://www.grappa.univ-lille3.fr/icga/tournament.php?id=173
That means, with the only exception of the WCCC Turin 2006 (where it finished 2nd being 0.5 points behind), Rybka has won ALL big tournaments where it participated, and is undisputed number one on the CCRL, CEGT and SSDF rating lists.
Rybka's got it in the bag
From http://www.fide.com/news.asp?id=1392
FIDE Presidential Board in Tallinn, communique:
"President Ilyumzhinov suggested to consider the organization of another computer challenge match between Rybka and Deep Junior in Mexico City."
"Suggesting" means nothing, but still, he did mention the idea.