Topic Rybka Support & Discussion / Rybka Discussion / Is Rybka better than DeeperBlue (game 7 in odds match) ?
![[fr]](/mwf/v12/flags/fr.png "France")
Hello to the members,
After the impressive draw by human in Rybka(3?) vs Dzindzichashvili match, what do you think of the level of play from the best engines.
Specially, I speak about the game number 7 were in my opinion Rybka with black played like expected from 1995's engines on PC.
After reading analyses from the games Kasparov vs DeeperBlue, I'm not sure that DeeperBlue would have played so weak (from a strategic point of view, with no counter play from the computer). Never.
The game number 8 showed an very easy draw from the human too.
Despite the computer odds, i'm a bit disapointed by theses 2 games (strategic point of view) and I ask myself : "when computer's strategy is so weak, where are the - real - progress in engines's level since DeeperBlue " ?
After the impressive draw by human in Rybka(3?) vs Dzindzichashvili match, what do you think of the level of play from the best engines.
Specially, I speak about the game number 7 were in my opinion Rybka with black played like expected from 1995's engines on PC.
After reading analyses from the games Kasparov vs DeeperBlue, I'm not sure that DeeperBlue would have played so weak (from a strategic point of view, with no counter play from the computer). Never.
The game number 8 showed an very easy draw from the human too.
Despite the computer odds, i'm a bit disapointed by theses 2 games (strategic point of view) and I ask myself : "when computer's strategy is so weak, where are the - real - progress in engines's level since DeeperBlue " ?
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
I've played through the games of DB V Kasparov 1997 and there is no doubt that Rybka is stronger, on a fast PC.
Having said that I do think DB is stronger than most people will admit - I rememeber programmers claiming they'd easily past it in 1999 - 2000 when they blatantly hadn't. Personally I would put Deeper Blue at around 2800 elo, Rybka closer to 3000 elo on a quad, and Hydra...maybe 3000 - 3100 elo.
Just my opinion.
Having said that I do think DB is stronger than most people will admit - I rememeber programmers claiming they'd easily past it in 1999 - 2000 when they blatantly hadn't. Personally I would put Deeper Blue at around 2800 elo, Rybka closer to 3000 elo on a quad, and Hydra...maybe 3000 - 3100 elo.
Just my opinion.
Actually, my point is : "how can a nowdays 3000 elo engine can play like in the 7th game ?"
Of course, we will never know if DeeperBlue would have been struggle (like with a boa constrictor in game 7) playing with a such "computer like" style despite the 3200 elo.
Reading me, you can imagine I still believe in Human when I see games like 7th : "I beleive that a very strong player with the good approach (exchanges, simplifications, strategic plan - if possible and if they have read Siman :) ) may beat the monsters without odds"
Of course, we will never know if DeeperBlue would have been struggle (like with a boa constrictor in game 7) playing with a such "computer like" style despite the 3200 elo.
Reading me, you can imagine I still believe in Human when I see games like 7th : "I beleive that a very strong player with the good approach (exchanges, simplifications, strategic plan - if possible and if they have read Siman :) ) may beat the monsters without odds"
![[mx]](/mwf/v12/flags/mx.png "Mexico")
> "I beleive that a very strong player with the good approach (exchanges, simplifications, strategic plan - if possible and if they have read Siman :-) ) may beat the monsters without odds"
I believe that player hasn't been born yet.
![[gr]](/mwf/v12/flags/gr.png "Greece")
OK nowone realy knows the strength of deep blue nor rybka. personaly i am sure only that rybka score should be at 90% range against deep blue, and i think that deep blue at 97 is about equal with deep fritz8. hydra might be around equal with rybka2.1 on a quad. of course strategic planning and handicap games are a completly different story. i consider hydra the ultimate player against humans because of her playing style but the sad truth is that computers are still completly idiots as Kasparov used to say. it's like calculators are extremly fast but they don't know mathematics. in the usual positions when they have just to count they are precise else there is no end to their fullness but in an ordinary chess game is difficult to hypnotize a program that's why a human should not have any chanse.
PS: As i have told many times already i don't consider Rybka an ideal oponet against humans and i have reasons to believe that the huge step forward in engine chess the last couple of years has not to do so much against human players. see for example tactic tests the new programs are as strong as was back in '99!!
personaly i will not be suprise if i see Fritz or Shredder perform better against humans than rybka but even deep blue that goes too far.
PS: As i have told many times already i don't consider Rybka an ideal oponet against humans and i have reasons to believe that the huge step forward in engine chess the last couple of years has not to do so much against human players. see for example tactic tests the new programs are as strong as was back in '99!!
personaly i will not be suprise if i see Fritz or Shredder perform better against humans than rybka but even deep blue that goes too far.
![[us]](/mwf/v12/flags/us.png "United States")
In most tactics tests, most new programs, including even Rybka, which is not a "tactical" program (whatever that means), score far better than the older programs. It is occasionally possible to find a special test suite that tends to favor the older programs, but that is rare; recall also the one that favored Fritz 5.32 only favored it because several of the key moves were wrong--most of the tactical test suites favor programs like Rybka Winfinder 2.2, Deep Fritz 10.1, etc.
Anyway, I would guess that a conservative estimate would say that you're correct that Deep Blue 1997 would have about the same elo as Deep Fritz 8 on a quad, i.e. I think we can all be very sure that Deep Blue 1997 is not stronger than that. My own feeling is that Fritz 7 on a fast single processor computer is probably about the same elo. Recall that Kramnik had proclaimed that Fritz 7 on his 600 MHz laptop was regularly finding objectively stronger moves than Deep Blue 1997 was finding. Kramnik tends to be conservative on such claims, and so I think that is to be quite believed. On the other hand, in tactical situations, Deep Blue 1997 would presumably be stronger some of the time (though not in the longer tactical sequences that today's programs see in a few seconds, especially in endgame scenarios). Of course, even in 1997 there were programs that were positionally stronger than Deep Blue. We are seeing now that it is these types of positions that tend to be important in both human games and human vs. computer games.
Anyway, I would guess that a conservative estimate would say that you're correct that Deep Blue 1997 would have about the same elo as Deep Fritz 8 on a quad, i.e. I think we can all be very sure that Deep Blue 1997 is not stronger than that. My own feeling is that Fritz 7 on a fast single processor computer is probably about the same elo. Recall that Kramnik had proclaimed that Fritz 7 on his 600 MHz laptop was regularly finding objectively stronger moves than Deep Blue 1997 was finding. Kramnik tends to be conservative on such claims, and so I think that is to be quite believed. On the other hand, in tactical situations, Deep Blue 1997 would presumably be stronger some of the time (though not in the longer tactical sequences that today's programs see in a few seconds, especially in endgame scenarios). Of course, even in 1997 there were programs that were positionally stronger than Deep Blue. We are seeing now that it is these types of positions that tend to be important in both human games and human vs. computer games.
I consider Deep Blue very stong positionaly contrary to the general opinion but it made many tactical blunders in the games with Kasparov.
It is very strong positionally compared with most (but not all) other programs at the time, but not compared with programs after about the year 2000. And yes, there were "tactical" "blunders" that occurred (one could even make the case that the infamous 37.Be4!!? was such an example, but the reasons for that are still under huge debate), particularly allowing the forced draw that Kasparov also overlooked--but that was due to a long-term tactical blunder coming from lack of search capabilities.
![[fi]](/mwf/v12/flags/fi.png "Finland")
> Actually, my point is : "how can a nowdays 3000 elo engine can play like in the 7th game ?"
>
> Of course, we will never know if DeeperBlue would have been struggle (like with a boa constrictor in game 7) playing with a such "computer like" style despite the 3200 elo.
>
> Reading me, you can imagine I still believe in Human when I see games like 7th : "I beleive that a very strong player with the good approach (exchanges, simplifications, strategic plan - if possible and if they have read Siman :-) ) may beat the monsters without odds"
Of course it is possible, but even Kasparov hadn't showed that in practical. To get a good position out of opening is impossible because engines use books. After that opening engine has at least equal position and few pieces are already traded off so there's lots of room, so there's no change to outplay engine by slowly squeezing it into zugzwang. Ok engines has problems with king-side attacks due to horizon-effect but getting that kind of position you have to lucky. One GM said that chess will be solved in 10-20 years so that's a sign to me there's nothing more than a draw against engines for humans.
![[cr]](/mwf/v12/flags/cr.png "Costa Rica")
Of course, since the defeat of Kasparov in 1997, a lot of people are saying that they have a ''stronger'' program than Deep Blue.
I don't discuss that, in fact this can be certain, but we do not have a clear evidence that can demonstrate it clearly; the 6 games that it played in 1997 are not anything. They dismantled the machine forever. It would have been grandiose to have seen more games.
Regards,
Gaмßito.
I don't discuss that, in fact this can be certain, but we do not have a clear evidence that can demonstrate it clearly; the 6 games that it played in 1997 are not anything. They dismantled the machine forever. It would have been grandiose to have seen more games.
Regards,
Gaмßito.
This is possibly one reason why they dismantled it--to keep the "debate" up. However, I think we can look back at its games and the games played by today's programs and easily say with confidence that Deep Blue wasn't what one would call "overpowering" in today's standards. When Kasparov played more positionally, especially in the early phases of the game, it played noticeably badly. In the same general types of positions, Ktulu 8, for example, certainly seems to pick noticeably better types of moves, the types more typically expected of human grandmasters with a strong knowledge of opening theory.
turbojuice contends: "I think we can look back at its games and the games played by today's programs and easily say with confidence that Deep Blue wasn't what one would call "overpowering" in today's standards. When Kasparov played more positionally, especially in the early phases of the game, it played noticeably badly."
As noticed by whom?
Unless you are in possession of super-GM insight, I'd like to know how you can make this claim "with confidence." Contemporary strong programs disagree about best moves in complicated positions all the time. That one views the other's choices as weak - and vice-versa - merely means that they evaluate the position differently. A collection of test positions are not a reliable way of establishing a rating (which is in effect what you are doing). Only playing a large number of games in a pool of similar strength opponents is scientifically valid. For Deep Blue we don't have that.
You've "looked back" at Deep Blue's games and decided that it played badly. That's not convincing. At how much time per move, by how many software programs, on what computer, investigating how many branches? And when do you know your analysis is conclusive? That's the crux. Walking through the moves casually is insufficient. If you have exhaustive analysis that proves systematic bad play by Deep Blue, I'd really like to see it.
More to the point, how do you know? What is your method of analysis that grants such confidence?
jk
As noticed by whom?
Unless you are in possession of super-GM insight, I'd like to know how you can make this claim "with confidence." Contemporary strong programs disagree about best moves in complicated positions all the time. That one views the other's choices as weak - and vice-versa - merely means that they evaluate the position differently. A collection of test positions are not a reliable way of establishing a rating (which is in effect what you are doing). Only playing a large number of games in a pool of similar strength opponents is scientifically valid. For Deep Blue we don't have that.
You've "looked back" at Deep Blue's games and decided that it played badly. That's not convincing. At how much time per move, by how many software programs, on what computer, investigating how many branches? And when do you know your analysis is conclusive? That's the crux. Walking through the moves casually is insufficient. If you have exhaustive analysis that proves systematic bad play by Deep Blue, I'd really like to see it.
More to the point, how do you know? What is your method of analysis that grants such confidence?
jk
To a large extent, much of this is authoritative consensus. For example, when Kramnik says something like that, I agree with him, especially since it's a matter that he's studied in detail. I never said that Deep Blue is weak--I just said that it's not "overpowering". While it's true that if you try to get into some types of tactical fights with it, you'll have problems, Deep Blue was mainly an extremely fast bean counter. It was basically like having Fritz 3 on very fast hardware. There are many things that we cannot systematically study, but we make good inferences anyway. I think you are somewhat of a holdout from the late 1990's, when there was this myth that certain computer programs were unstoppable if you put them on very fast hardware (though of course Deep Blue was absolutely nothing like Hydra in comparison--in terms of the non chess chip part, Deep Blue was about the same speed as a Pentium IV 4 GHz single processor computer), and that myth lasted until about 2000, when Kramnik showed pretty well how to beat "modern" computer programs, and the methods used after that time by grandmasters were much more direct and effective than the typical slow, closed, and no plan of action play that Kasparov used in 1997 (not only was it an ineffective method, it was very much against his natural style, so he was effectively weakening himself to a 2600-level player).
2600 level player? relax my friend. Garry made his best until the final game and lost from about an equal player.
I believe though that Deep Blue was a program with deep chess knowledge and i don't see to it's games the bean counter aproach IBM confirmed. Deep Blue was at human world champion level no doubt about that.
I believe though that Deep Blue was a program with deep chess knowledge and i don't see to it's games the bean counter aproach IBM confirmed. Deep Blue was at human world champion level no doubt about that.
Well, many grandmasters, including Kasparov and Kramnik, admit that Kasparov played far below his normal level during that match, and that was true even before game 6. Nonetheless, I am not sure that Kasparov would have gotten much more than a draw against Deep Blue even if he was in top form--perhaps he would have won by 3.5-2.5 or 4.0-2.0, but not more than that. His natural playing style isn't as well suited against something like Deep Blue as many other chess players. Even Topalov might be better suited (don't forget--he is the closest any human has ever come to defeating the far stronger Hydra in a game), and certainly people like Anand, Kramnik, and Karpov at his prime would have had much better luck. Even though I don't like Kramnik, and consider him a far inferior overall player compared with the likes of people like Kasparov and Anand, I have little doubt that Kramnik in the late 1990's would have beaten Deep Blue without too much effort. The reason for this is that his natural playing style is anti-computer, and especially anti-tactical computer, and so he doesn't weaken himself by going into such a match. Of course, Kasparov adapted after 1997, and he played more like "himself" in his games against Deep Junior and Deep Fritz (in spite of one game each where he had a nasty blunder--his other games were quite good against stronger opponents than Deep Blue).
OK i don't believe in such a thing like computer play adaptation when we talk for a fair chess competition.
By the way Topalov is one of the most inaccurate chess players in our days and has tendance for complication that should not fit whith your opinion about the apropriate style.
So finaly what is your estimation about Deep Blue's elo?
By the way Topalov is one of the most inaccurate chess players in our days and has tendance for complication that should not fit whith your opinion about the apropriate style.
So finaly what is your estimation about Deep Blue's elo?
I believe it was 2700. I am open to the ideas of some other people that it might be 2800, and I know that among the people who study this stuff, that is a high estimate (that was even considered a high estimate back around 2000, when it was thought that Fritz 5.32 played at the 2700 level on a Pentium III chip).
Actualy from my own experience i estimate Fritz5 around 2650 in modern hardware and Deep Blue close to 2800. But absets can happen. Once i drew a 4 game match against Deep Junior 7 but i perform always very purly against Fritz8. There is not a good corelation with human elo.
Yeah, it's interesting how that works sometimes, and that's kind of what I mean about the importance of playing styles against computers. Deep Junior 7 on strong hardware should certainly be stronger than Fritz 8 on single processor, but the playing styles are wildly different. If someone doesn't have to change his playing style to be a good opponent for a particular program, then he can perform very well. This wasn't known well to Kasparov at the time, but Deep Blue rarely searched beyond 11 or 12 ply except with extensions--I think that many human grandmasters today, knowing this information, could take advantage of this very nicely. However, they wouldn't be able to make much use of it against modern programs that typically search to 20 ply.
![[il]](/mwf/v12/flags/il.png "Israel")
I disagree that people like karpov could be better against computers.
karpov's record against computers is clearly bad.
both kasparov and karpov played against deep thought(earlier version of deep blue).
Kasparov won convincingly when karpov could only win in a drawn endgame thanks to a blunder of deep thought that other programs do not do and probably deep blue does not do.
Uri
karpov's record against computers is clearly bad.
both kasparov and karpov played against deep thought(earlier version of deep blue).
Kasparov won convincingly when karpov could only win in a drawn endgame thanks to a blunder of deep thought that other programs do not do and probably deep blue does not do.
Uri
Well, that is still Karpov's style--he wins a lot of games in the endgame. Also, this was before the age of anti-computer thinking--when Kramnik "showed the way" around 2000, grandmasters started beating computers left and right. Before that, they were doing quite well--I remember Fritz 5.32 have a TPR of something like 2830 in a tournament in 1999, I think, against some of the world's strongest players.
I believe that Fritz5.32 never had tpr of 2830 in non rapid chess event.
I also disagree that kramnik showed the way around 2000.
The fact that kramnik won one game against Junior does not mean showing the way to other people.
kasparov has already experience of playing in anti-computer style and even in 1994 he used 1.e3 to beat Fritz3
Earlier he used anti-computer style to beat deep thought easily.
The main problem of kasparov in the match of 1997 was that he did not know his opponent and not that he played in a level that is weaker than his normal level.
It caused him to resign in a drawn position in game 2 and lose the match.
Uri
I also disagree that kramnik showed the way around 2000.
The fact that kramnik won one game against Junior does not mean showing the way to other people.
kasparov has already experience of playing in anti-computer style and even in 1994 he used 1.e3 to beat Fritz3
Earlier he used anti-computer style to beat deep thought easily.
The main problem of kasparov in the match of 1997 was that he did not know his opponent and not that he played in a level that is weaker than his normal level.
It caused him to resign in a drawn position in game 2 and lose the match.
Uri
I believe that Fritz 5.32 could have had among 2650 and 2750 elo points in non rapid chess tournaments against humans. Unfortunately these tests never happened. It has always been very difficult to see chess programs playing in tournaments against strong humans under slow time controls.
Also, Fritz 5.32 clearly had more than 2800 elo points in rapid chess tournaments against humans. It was a very strong tactician and it clearly demonstrated their tactical strength abilities in several occasions against strong opponents. I remember that, I loved the playing style that this program had.
Note that Kasparov also lose against Fritz 3 in several occasions. He also lose in a ''25 minutes game'' against Genius 3, where Genius was using a P90 and the black pieces. What great moment; that happened in 1994.
I think that not know his opponent play during the 1997 match was clearly negative for him, but I am not sure that this was the cause that make him to resign in a draw position in game 2, neither later when he also lost in game 06 after he blundered horrible with h6!?, forgetting at all of the very strong opening reply Nxe6!
Humans give always a lot of pretexts when they lose. I really do not like this. By Kasparov side, it is well known that he was never a good loser. Never.
Regards,
Gaмßito.
Also, Fritz 5.32 clearly had more than 2800 elo points in rapid chess tournaments against humans. It was a very strong tactician and it clearly demonstrated their tactical strength abilities in several occasions against strong opponents. I remember that, I loved the playing style that this program had.
Note that Kasparov also lose against Fritz 3 in several occasions. He also lose in a ''25 minutes game'' against Genius 3, where Genius was using a P90 and the black pieces. What great moment; that happened in 1994.
I think that not know his opponent play during the 1997 match was clearly negative for him, but I am not sure that this was the cause that make him to resign in a draw position in game 2, neither later when he also lost in game 06 after he blundered horrible with h6!?, forgetting at all of the very strong opening reply Nxe6!
Humans give always a lot of pretexts when they lose. I really do not like this. By Kasparov side, it is well known that he was never a good loser. Never.
Regards,
Gaмßito.
True--Fritz 5.32 didn't have TPR of 2830 in non-rapid events. However, the one in particular I'm thinking about was more like semi-rapid, if I remember correctly--the time control was something like 25'+20" or 25'+15"--not much different than the Roman match recently.
As for what Kramnik did--what he did is he showed by that playing in a fairly normal style in the opening, but getting a semi-closed position, it allowed for that sort of maneuvering to gain a nice attack against what were, at the time, modern programs. Before that game, Junior was demolishing everyone in that tournament if I recall. That was the turning point there--the others mimicked his style, and they many were just as successful. It turned a 2900 performance by Junior into a "normal" 2700 performance. Kasparov had an entirely different style--he picked moves in the opening to try to take the computer out of book early, and this worked against programs...until around the time of Deep Blue.
As for what Kramnik did--what he did is he showed by that playing in a fairly normal style in the opening, but getting a semi-closed position, it allowed for that sort of maneuvering to gain a nice attack against what were, at the time, modern programs. Before that game, Junior was demolishing everyone in that tournament if I recall. That was the turning point there--the others mimicked his style, and they many were just as successful. It turned a 2900 performance by Junior into a "normal" 2700 performance. Kasparov had an entirely different style--he picked moves in the opening to try to take the computer out of book early, and this worked against programs...until around the time of Deep Blue.
I think that humans did not need kramnik to learn how to play against computers.
They could train at home against Junior.
Junior played in human tournament earlier and it never had performance that is close to 2700 but only something close to 2500.
Junior used slower hardware but if you believe that better hardware and slightly better software is enough for more than 200 elo then you are clearly wrong.
The 2700 result of Junior is because humans were unprepared but the game against kramnik did not help them to prepare better and I believe that leko lost against Junior after the game of kramnik against Junior.
I remember that Amir ban claimed that he changed Junior after the tournament because Boris Alterman showed him that he has a strategy to beat Junior that players at the tournament did not know.
Uri
They could train at home against Junior.
Junior played in human tournament earlier and it never had performance that is close to 2700 but only something close to 2500.
Junior used slower hardware but if you believe that better hardware and slightly better software is enough for more than 200 elo then you are clearly wrong.
The 2700 result of Junior is because humans were unprepared but the game against kramnik did not help them to prepare better and I believe that leko lost against Junior after the game of kramnik against Junior.
I remember that Amir ban claimed that he changed Junior after the tournament because Boris Alterman showed him that he has a strategy to beat Junior that players at the tournament did not know.
Uri
Junior used slower hardware but if you believe that better hardware and slightly better software is enough for more than 200 elo then you are clearly wrong.
Well, consider this: going from 200 MHz to 450 MHz gave Junior 5 an increase of over 100 elo points in the SSDF rating list. That's one thing. Another is that going from Junior 5 to Junior 6 gave an increase of over 50 elo points. The rest could easily be attributed to tournament statistics.
Well, consider this: going from 200 MHz to 450 MHz gave Junior 5 an increase of over 100 elo points in the SSDF rating list. That's one thing. Another is that going from Junior 5 to Junior 6 gave an increase of over 50 elo points. The rest could easily be attributed to tournament statistics.
The 2650-2700 result of Junior in Dortmund was also due thanks to the victory without merit that Junior scored against GM Robert Huebner.
Honestly, I never understood their attitude of refuse to continue playing in a position where there was a lot of game. Huebner never likes to play against Computers and previously I remember that, he had refused to play against a program in a tournament.
He changed his mind later and was open to challenge Fritz.
The position when he resigned against Junior is this:
575815: Comp Deep Junior - Huebner,R, Dortmund SuperGM 2000
Analysis by Rybka 2.3.2a 32-bit :
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 f5 24.exf6 Qxf6 25.N2f3 Qf4 26.Nxe6 Nxe6
+/= (0.45) Depth: 5 00:00:00
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.Kg1 h6 26.h3 Nxd4 27.Nxd4
+/= (0.43) Depth: 6 00:00:00 7kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.Kg1 h6 26.h3 Nxd4 27.Nxd4
+/= (0.43) Depth: 7 00:00:00 12kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.g3 h6 26.Kg2 Rfe8
+/= (0.44) Depth: 8 00:00:01 25kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.g3 h6 26.Kg2 Rfe8 27.Nxc6 Qxc6 28.Nd4
+/= (0.45) Depth: 9 00:00:01 43kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qg5 24.N2f3 Qg6 25.Qxg6 fxg6 26.Kg1 h6 27.h4 b5 28.Re2
+/= (0.47) Depth: 10 00:00:02 78kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Ng5 g6 25.Bxf8 Nxa1 26.Bd6 Nb3 27.c4 Rb8 28.Qh3 f5 29.exf6 Qxd6 30.Qxh7+ Kf8 31.cxd5
+/= (0.36) Depth: 10 00:00:03 127kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.Nd4 Qc4 29.Kg1 a4 30.Qg5+ Kf8
+/= (0.37) Depth: 11 00:00:04 183kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.Nd4 Qc4 29.Rb1 a4 30.Qg5+ Kf8
+/= (0.37) Depth: 12 00:00:06 266kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.Qg5+ Kf8 29.Qh6+ Kg8 30.Ng5 Qd3 31.Qh5 Rf8 32.Kg1
+/= (0.38) Depth: 13 00:00:10 426kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.h3 a4 29.Qg5+ Kf8 30.Qh6+ Kg8 31.Ng5 Qd3 32.Qf6 Qf5
+/= (0.26) Depth: 14 00:00:20 885kN
20...bxc5 21.Ne4 Nxa5 22.Nfg5 Bxg5 23.Nxg5 g6 24.Ne4 Qc6 25.Nxc5 Rfb8 26.c4 Nb7 27.Qe3 Nxc5 28.Bxc5 dxc4 29.bxc4 a5 30.Red1 Rb2 31.Rd6 Qb7
+/= (0.31) Depth: 15 00:00:38 1649kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd6 Qg6 28.Qh3 f5 29.Rab1 Rab8 30.Rxb3 Rxb3 31.f4 Rb2 32.g3
+/= (0.27) Depth: 16 00:01:35 4458kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd6 Qg6 28.Qh3 f5 29.Qe3 Qf6 30.Qg3+ Qg6 31.Re3 Na6 32.Bd4 Nxd4 33.cxd4 Qxg3
= (0.23) Depth: 17 00:02:47 8171kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd6 Qg6 28.Qe3 h5 29.g3 e5 30.Red1 d4 31.Qf3 Ne7
+/= (0.26) Depth: 18 00:04:34 13315kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd2 Rb2 28.Ba3 Rb7 29.f4 Qg6 30.Qxg6+ hxg6 31.a6 Rb6 32.Bd6 Nxa6 33.Rxe6
= (0.23) Depth: 19 00:08:34 25293kN
20...bxc5 21.Ne4 Rfb8 22.Bxc5 Rxb3 23.Nfg5 Bxg5 24.Nxg5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Bd6 dxe4 28.Qc4 Qd7 29.Bxc7 Rb7 30.Bf4 f5 31.Rad1
+/= (0.38) Depth: 20 00:23:20 67248kN
(, AMD 16.03.2008)
Regards,
Gaмßito.
Honestly, I never understood their attitude of refuse to continue playing in a position where there was a lot of game. Huebner never likes to play against Computers and previously I remember that, he had refused to play against a program in a tournament.
He changed his mind later and was open to challenge Fritz.
The position when he resigned against Junior is this:
575815: Comp Deep Junior - Huebner,R, Dortmund SuperGM 2000
r4rk1/p1nqbppp/1pn1p3/P1PpP3/8/BPPQ1N2/3N1PPP/R3R2K b - - 0 1
Analysis by Rybka 2.3.2a 32-bit :
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 f5 24.exf6 Qxf6 25.N2f3 Qf4 26.Nxe6 Nxe6
+/= (0.45) Depth: 5 00:00:00
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.Kg1 h6 26.h3 Nxd4 27.Nxd4
+/= (0.43) Depth: 6 00:00:00 7kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.Kg1 h6 26.h3 Nxd4 27.Nxd4
+/= (0.43) Depth: 7 00:00:00 12kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.g3 h6 26.Kg2 Rfe8
+/= (0.44) Depth: 8 00:00:01 25kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qd7 24.N2f3 Nc6 25.g3 h6 26.Kg2 Rfe8 27.Nxc6 Qxc6 28.Nd4
+/= (0.45) Depth: 9 00:00:01 43kN
20...Nxa5 21.cxb6 axb6 22.Bxe7 Qxe7 23.Nd4 Qg5 24.N2f3 Qg6 25.Qxg6 fxg6 26.Kg1 h6 27.h4 b5 28.Re2
+/= (0.47) Depth: 10 00:00:02 78kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Ng5 g6 25.Bxf8 Nxa1 26.Bd6 Nb3 27.c4 Rb8 28.Qh3 f5 29.exf6 Qxd6 30.Qxh7+ Kf8 31.cxd5
+/= (0.36) Depth: 10 00:00:03 127kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.Nd4 Qc4 29.Kg1 a4 30.Qg5+ Kf8
+/= (0.37) Depth: 11 00:00:04 183kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.Nd4 Qc4 29.Rb1 a4 30.Qg5+ Kf8
+/= (0.37) Depth: 12 00:00:06 266kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.Qg5+ Kf8 29.Qh6+ Kg8 30.Ng5 Qd3 31.Qh5 Rf8 32.Kg1
+/= (0.38) Depth: 13 00:00:10 426kN
20...bxc5 21.Ne4 Nxa5 22.Nxc5 Bxc5 23.Bxc5 Nxb3 24.Bxf8 Nxa1 25.Bxg7 Qb5 26.Qd2 Kxg7 27.Rxa1 a5 28.h3 a4 29.Qg5+ Kf8 30.Qh6+ Kg8 31.Ng5 Qd3 32.Qf6 Qf5
+/= (0.26) Depth: 14 00:00:20 885kN
20...bxc5 21.Ne4 Nxa5 22.Nfg5 Bxg5 23.Nxg5 g6 24.Ne4 Qc6 25.Nxc5 Rfb8 26.c4 Nb7 27.Qe3 Nxc5 28.Bxc5 dxc4 29.bxc4 a5 30.Red1 Rb2 31.Rd6 Qb7
+/= (0.31) Depth: 15 00:00:38 1649kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd6 Qg6 28.Qh3 f5 29.Rab1 Rab8 30.Rxb3 Rxb3 31.f4 Rb2 32.g3
+/= (0.27) Depth: 16 00:01:35 4458kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd6 Qg6 28.Qh3 f5 29.Qe3 Qf6 30.Qg3+ Qg6 31.Re3 Na6 32.Bd4 Nxd4 33.cxd4 Qxg3
= (0.23) Depth: 17 00:02:47 8171kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd6 Qg6 28.Qe3 h5 29.g3 e5 30.Red1 d4 31.Qf3 Ne7
+/= (0.26) Depth: 18 00:04:34 13315kN
20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Ng5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Nd2 Rb2 28.Ba3 Rb7 29.f4 Qg6 30.Qxg6+ hxg6 31.a6 Rb6 32.Bd6 Nxa6 33.Rxe6
= (0.23) Depth: 19 00:08:34 25293kN
20...bxc5 21.Ne4 Rfb8 22.Bxc5 Rxb3 23.Nfg5 Bxg5 24.Nxg5 f5 25.exf6 gxf6 26.Ne4 Qg7 27.Bd6 dxe4 28.Qc4 Qd7 29.Bxc7 Rb7 30.Bf4 f5 31.Rad1
+/= (0.38) Depth: 20 00:23:20 67248kN
(, AMD 16.03.2008)
Regards,
Gaмßito.
I think that Rybka is wrong here--white is overwhelmingly winning this position--black simply cannot do anything. The evaluation should be far higher. It reminds me a little bit of a game Topalov-Anand from last year where Anand resigned in a similar type of situation; Rybka evaluated about +0.5.
I am not so sure that black cannot do anything. It seems that black has a lot of play here and it is still enough alive.
Note that Fritz 11 also doesn't worry about black position and their score is very similar to Rybka 2.3.2a. No problem at all.
Analysis by Fritz 11:
1...Nxa5 2.cxb6 axb6 3.Bxe7 Qxe7 4.Nd4 f6 5.exf6 Qxf6 6.N2f3 Qe7 7.h3
+/- (0.72) Depth: 9/24 00:00:00 117kN
1...Nxa5 2.cxb6 axb6 3.Bxe7 Qxe7 4.Nd4 f6 5.exf6 Qxf6 6.N2f3 Qg6 7.Qxg6 hxg6 8.Nxe6
+/- (0.75) Depth: 10/26 00:00:00 286kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Nxb5 Nxb5 7.Qxb5 Rfb8 8.Qd3
+/- (0.80) Depth: 11/28 00:00:01 628kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Rec1 Qd7 7.h3 Nc6 8.N2f3
+/= (0.62) Depth: 12/30 00:00:02 1355kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Rec1 Qd7 7.h3 Nc6 8.N2f3 Nxd4
+/= (0.68) Depth: 13/27 00:00:03 2117kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Rec1 Nb7 7.b4 Qg5 8.N2f3 Qg6 9.Qxg6 fxg6
+/= (0.64) Depth: 14/31 00:00:05 3675kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.h3 b4 7.Ra4 Rfb8 8.cxb4 Rxb4 9.Rc1 Rb6 10.b4
+/= (0.70) Depth: 15/30 00:00:09 6175kN
1...Nxa5 2.Kg1 bxc5 3.Bxc5 Bxc5 4.Rxa5 Qe7 5.Nd4 Bxd4 6.cxd4 a6 7.Rc5 Nb5 8.Rec1 Qh4 9.Nf3 Qe4 10.Qd2
+/= (0.64) Depth: 16/34 00:00:22 15834kN
1...Nxa5 2.b4 Nb7 3.Nb3 Nb5 4.Nfd4 Nxd4 5.cxd4 bxc5 6.bxc5 Nd8 7.Na5 Nc6 8.Nxc6 Qxc6 9.Bb4 Rfb8 10.Reb1 h6 11.Bd2
+/- (0.75) Depth: 17/38 00:00:41 31550kN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Qh3 h5 7.Qf3 Nb5 8.Re3 Rb8 9.Kg1 Rb2 10.Qf4
+/= (0.66) Depth: 17/38 00:00:51 39725kN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Qh3 h5 7.Qf3 Nb5 8.Re3 Qc7 9.Bd6 Qd7 10.h4 f5
+/= (0.53) Depth: 18/35 00:01:20 64378kN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Qh3 h5 7.Qf3 Nb5 8.Qd1 Rb2 9.c4 Nc3 10.Qc1 Rb3
+/= (0.33) Depth: 19/40 00:02:44 130mN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Ne4 Ne8 7.Qd2 h5 8.f4 Rab8 9.Ng5 Nc7 10.Kg1 Rb2
= (0.17) Depth: 20/40 00:05:45 267mN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 f5 6.exf6 gxf6 7.Ne4 Qg7 8.Nd6 Qg6 9.Qh3 Rab8 10.f4 h5 11.Qf3 Qg4
+/= (0.37) Depth: 21/41 00:09:25 437mN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 f5 6.exf6 gxf6 7.Ne4 Qg7 8.Nd6 Qg6 9.Qh3 h5 10.Rxe6 Nxe6 11.Qxe6+ Kh8 12.Rd1
+/= (0.33) Depth: 22/43 00:17:28 818mN
(, AMD 16.03.2008)
Regards,
Gaмßito.
Note that Fritz 11 also doesn't worry about black position and their score is very similar to Rybka 2.3.2a. No problem at all.
Analysis by Fritz 11:
1...Nxa5 2.cxb6 axb6 3.Bxe7 Qxe7 4.Nd4 f6 5.exf6 Qxf6 6.N2f3 Qe7 7.h3
+/- (0.72) Depth: 9/24 00:00:00 117kN
1...Nxa5 2.cxb6 axb6 3.Bxe7 Qxe7 4.Nd4 f6 5.exf6 Qxf6 6.N2f3 Qg6 7.Qxg6 hxg6 8.Nxe6
+/- (0.75) Depth: 10/26 00:00:00 286kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Nxb5 Nxb5 7.Qxb5 Rfb8 8.Qd3
+/- (0.80) Depth: 11/28 00:00:01 628kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Rec1 Qd7 7.h3 Nc6 8.N2f3
+/= (0.62) Depth: 12/30 00:00:02 1355kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Rec1 Qd7 7.h3 Nc6 8.N2f3 Nxd4
+/= (0.68) Depth: 13/27 00:00:03 2117kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.Rec1 Nb7 7.b4 Qg5 8.N2f3 Qg6 9.Qxg6 fxg6
+/= (0.64) Depth: 14/31 00:00:05 3675kN
1...Nxa5 2.Kg1 h6 3.cxb6 axb6 4.Bxe7 Qxe7 5.Nd4 b5 6.h3 b4 7.Ra4 Rfb8 8.cxb4 Rxb4 9.Rc1 Rb6 10.b4
+/= (0.70) Depth: 15/30 00:00:09 6175kN
1...Nxa5 2.Kg1 bxc5 3.Bxc5 Bxc5 4.Rxa5 Qe7 5.Nd4 Bxd4 6.cxd4 a6 7.Rc5 Nb5 8.Rec1 Qh4 9.Nf3 Qe4 10.Qd2
+/= (0.64) Depth: 16/34 00:00:22 15834kN
1...Nxa5 2.b4 Nb7 3.Nb3 Nb5 4.Nfd4 Nxd4 5.cxd4 bxc5 6.bxc5 Nd8 7.Na5 Nc6 8.Nxc6 Qxc6 9.Bb4 Rfb8 10.Reb1 h6 11.Bd2
+/- (0.75) Depth: 17/38 00:00:41 31550kN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Qh3 h5 7.Qf3 Nb5 8.Re3 Rb8 9.Kg1 Rb2 10.Qf4
+/= (0.66) Depth: 17/38 00:00:51 39725kN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Qh3 h5 7.Qf3 Nb5 8.Re3 Qc7 9.Bd6 Qd7 10.h4 f5
+/= (0.53) Depth: 18/35 00:01:20 64378kN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Qh3 h5 7.Qf3 Nb5 8.Qd1 Rb2 9.c4 Nc3 10.Qc1 Rb3
+/= (0.33) Depth: 19/40 00:02:44 130mN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 g6 6.Ne4 Ne8 7.Qd2 h5 8.f4 Rab8 9.Ng5 Nc7 10.Kg1 Rb2
= (0.17) Depth: 20/40 00:05:45 267mN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 f5 6.exf6 gxf6 7.Ne4 Qg7 8.Nd6 Qg6 9.Qh3 Rab8 10.f4 h5 11.Qf3 Qg4
+/= (0.37) Depth: 21/41 00:09:25 437mN
1...bxc5 2.Ne4 Rfb8 3.Nxc5 Bxc5 4.Bxc5 Rxb3 5.Ng5 f5 6.exf6 gxf6 7.Ne4 Qg7 8.Nd6 Qg6 9.Qh3 h5 10.Rxe6 Nxe6 11.Qxe6+ Kh8 12.Rd1
+/= (0.33) Depth: 22/43 00:17:28 818mN
(, AMD 16.03.2008)
Regards,
Gaмßito.
![[de]](/mwf/v12/flags/de.png "Germany")
yes, my rybka is better in this, after I went through some lines I got the following:
[² (+0.72)] d=14 1...Nxa5 2.b4 Nb7 3.Nb3 Nb5 4.Nfd4 Nxd4 5.cxd4 Rfc8 6.Rec1 bxc5 7.bxc5 Nd8 8.Na5 (0:00.42)
[² (+0.69)] d=13 1...Nxa5 2.b4 Nb7 3.Nb3 (0:00.07)
[² (+0.75)] d=10 1...Nxa5 2.b4 Nb7 3.Nb3 Rfb8 (0:00.04)
[² (+0.76)] d=10 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 a4 5.cxd5 Nxd5 6.bxa4 Nf4 (0:00.03)
[² (+0.75)] d=9 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 dxc4 5.Qxc4 Nd5 6.h3 a6 (0:00.01)
[² (+0.74)] d=8 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 dxc4 5.Qxc4 Nd5 6.h3 a6 (0:00.00)
[² (+0.67)] d=6 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 dxc4 5.Qxc4 Nd5 6.h3 a6 (0:00.00)
[² (+0.72)] d=14 1...Nxa5 2.b4 Nb7 3.Nb3 Nb5 4.Nfd4 Nxd4 5.cxd4 Rfc8 6.Rec1 bxc5 7.bxc5 Nd8 8.Na5 (0:00.42)
[² (+0.69)] d=13 1...Nxa5 2.b4 Nb7 3.Nb3 (0:00.07)
[² (+0.75)] d=10 1...Nxa5 2.b4 Nb7 3.Nb3 Rfb8 (0:00.04)
[² (+0.76)] d=10 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 a4 5.cxd5 Nxd5 6.bxa4 Nf4 (0:00.03)
[² (+0.75)] d=9 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 dxc4 5.Qxc4 Nd5 6.h3 a6 (0:00.01)
[² (+0.74)] d=8 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 dxc4 5.Qxc4 Nd5 6.h3 a6 (0:00.00)
[² (+0.67)] d=6 1...bxa5 2.Ne4 Rab8 3.c4 h6 4.Nd6 dxc4 5.Qxc4 Nd5 6.h3 a6 (0:00.00)
Let the module to think deeply enough. After this, post the result again.
Regards,
Gaмßito.
Regards,
Gaмßito.
it's deeper than you think when you see the plies since I went through the lines manually, they ended at +1,5 (bxc) and 0,75 (Nxa5)
how do you get +1.5 after bxc?
Assuming that the game continue bxc Ne4 you need to analyze many lines to prove +1.5 because rybka does not see even +0.7 at 4 options
depth 16:
Analysis by Rybka 2.3.2a 32-bit :
1. = (0.25): 2...Rf8-b8 3.Ne4xc5 Be7xc5 4.Ba3xc5 Rb8xb3 5.Nf3-g5 f7-f5 6.e5xf6 g7xf6 7.Ng5-e4 Qd7-g7 8.Ne4-d6 Qg7-g6 9.Qd3-h3
2. ² (0.30): 2...Nc6xa5 3.Ne4xc5 Be7xc5 4.Ba3xc5 Na5xb3 5.Bc5xf8 Nb3xa1 6.Bf8xg7 Kg8xg7 7.Re1xa1 Qd7-b5 8.Qd3-c2 Qb5-c4 9.h2-h3
3. ² (0.43): 2...c5-c4 3.b3xc4 Be7xa3 4.Ra1xa3 Qd7-e7 5.c4xd5 Qe7xa3 6.d5xc6 Nc7-d5 7.a5-a6 h7-h6 8.Nf3-d4 Ra8-b8 9.Qd3-g3
4. ² (0.64): 2...Qd7-e8 3.Ne4xc5 Ra8-b8 4.Nf3-d4 Nc6xd4 5.c3xd4 Qe8-b5 6.Qd3-d2 Rf8-c8 7.Kh1-g1 h7-h6 8.f2-f3 Be7-h4 9.Re1-c1
(, 17.03.2008)
depth 17:
New game
Analysis by Rybka 2.3.2a 32-bit :
1. = (0.21): 2...Rf8-b8 3.Ne4xc5 Be7xc5 4.Ba3xc5 Rb8xb3 5.Nf3-g5 f7-f5 6.e5xf6 g7xf6 7.Ng5-e4 Qd7-g7 8.Ne4-d6 Qg7-g6 9.Qd3-e3
2. ² (0.41): 2...c5-c4 3.b3xc4 Be7xa3 4.Ra1xa3 Qd7-e7 5.c4xd5 Qe7xa3 6.d5xc6 Nc7-d5 7.a5-a6 Ra8-b8 8.Nf3-d4 h7-h6 9.Qd3-g3
3. ² (0.50): 2...Nc6xa5 3.Nf3-g5 Be7xg5 4.Ne4xg5 f7-f5 5.Ba3xc5 Na5xb3 6.Bc5xf8 Nb3xa1 7.Bf8-d6 Na1-b3 8.c3-c4 Nb3-a5 9.Bd6xc7
4. ² (0.64): 2...Qd7-e8 3.Ne4xc5 Ra8-b8 4.Nf3-d4 Nc6xd4 5.c3xd4 Qe8-b5 6.Qd3-d2 Rf8-c8 7.Kh1-g1 h7-h6 8.f2-f3 Be7-h4 9.Re1-c1
(, 17.03.2008)
depth 18:
New game
Analysis by Rybka 2.3.2a 32-bit :
1. ² (0.28): 2...Rf8-b8 3.Ne4xc5 Be7xc5 4.Ba3xc5 Rb8xb3 5.Nf3-g5 f7-f5 6.e5xf6 g7xf6 7.Ng5-e4 Qd7-g7 8.Bc5-d6 d5xe4 9.Qd3-c4
2. ² (0.37): 2...c5-c4 3.b3xc4 Be7xa3 4.Ra1xa3 Qd7-e7 5.c4xd5 Qe7xa3 6.d5xc6 Ra8-d8 7.Ne4-d6 h7-h6 8.Nf3-d4 Rd8-b8 9.g2-g3
3. ² (0.57): 2...Nc6xa5 3.Nf3-g5 Be7xg5 4.Ne4xg5 f7-f5 5.Ba3xc5 Na5xb3 6.Bc5xf8 Nb3xa1 7.Bf8-d6 Na1-b3 8.c3-c4 Nb3-a5 9.Bd6xc7
4. ² (0.65): 2...Qd7-e8 3.Ne4xc5 Ra8-b8 4.Nf3-d4 Nc6xd4 5.c3xd4 Qe8-b5 6.Qd3-d2 Rf8-c8 7.Kh1-g1 h7-h6 8.f2-f4 Be7xc5 9.Ba3xc5
(, 17.03.2008)
Assuming that the game continue bxc Ne4 you need to analyze many lines to prove +1.5 because rybka does not see even +0.7 at 4 options
depth 16:
r4rk1/p1nqbppp/2n1p3/P1ppP3/4N3/BPPQ1N2/5PPP/R3R2K b - - 0 1
Analysis by Rybka 2.3.2a 32-bit :
1. = (0.25): 2...Rf8-b8 3.Ne4xc5 Be7xc5 4.Ba3xc5 Rb8xb3 5.Nf3-g5 f7-f5 6.e5xf6 g7xf6 7.Ng5-e4 Qd7-g7 8.Ne4-d6 Qg7-g6 9.Qd3-h3
2. ² (0.30): 2...Nc6xa5 3.Ne4xc5 Be7xc5 4.Ba3xc5 Na5xb3 5.Bc5xf8 Nb3xa1 6.Bf8xg7 Kg8xg7 7.Re1xa1 Qd7-b5 8.Qd3-c2 Qb5-c4 9.h2-h3
3. ² (0.43): 2...c5-c4 3.b3xc4 Be7xa3 4.Ra1xa3 Qd7-e7 5.c4xd5 Qe7xa3 6.d5xc6 Nc7-d5 7.a5-a6 h7-h6 8.Nf3-d4 Ra8-b8 9.Qd3-g3
4. ² (0.64): 2...Qd7-e8 3.Ne4xc5 Ra8-b8 4.Nf3-d4 Nc6xd4 5.c3xd4 Qe8-b5 6.Qd3-d2 Rf8-c8 7.Kh1-g1 h7-h6 8.f2-f3 Be7-h4 9.Re1-c1
(, 17.03.2008)
depth 17:
New game
r4rk1/p1nqbppp/2n1p3/P1ppP3/4N3/BPPQ1N2/5PPP/R3R2K b - - 0 1
Analysis by Rybka 2.3.2a 32-bit :
1. = (0.21): 2...Rf8-b8 3.Ne4xc5 Be7xc5 4.Ba3xc5 Rb8xb3 5.Nf3-g5 f7-f5 6.e5xf6 g7xf6 7.Ng5-e4 Qd7-g7 8.Ne4-d6 Qg7-g6 9.Qd3-e3
2. ² (0.41): 2...c5-c4 3.b3xc4 Be7xa3 4.Ra1xa3 Qd7-e7 5.c4xd5 Qe7xa3 6.d5xc6 Nc7-d5 7.a5-a6 Ra8-b8 8.Nf3-d4 h7-h6 9.Qd3-g3
3. ² (0.50): 2...Nc6xa5 3.Nf3-g5 Be7xg5 4.Ne4xg5 f7-f5 5.Ba3xc5 Na5xb3 6.Bc5xf8 Nb3xa1 7.Bf8-d6 Na1-b3 8.c3-c4 Nb3-a5 9.Bd6xc7
4. ² (0.64): 2...Qd7-e8 3.Ne4xc5 Ra8-b8 4.Nf3-d4 Nc6xd4 5.c3xd4 Qe8-b5 6.Qd3-d2 Rf8-c8 7.Kh1-g1 h7-h6 8.f2-f3 Be7-h4 9.Re1-c1
(, 17.03.2008)
depth 18:
New game
r4rk1/p1nqbppp/2n1p3/P1ppP3/4N3/BPPQ1N2/5PPP/R3R2K b - - 0 1
Analysis by Rybka 2.3.2a 32-bit :
1. ² (0.28): 2...Rf8-b8 3.Ne4xc5 Be7xc5 4.Ba3xc5 Rb8xb3 5.Nf3-g5 f7-f5 6.e5xf6 g7xf6 7.Ng5-e4 Qd7-g7 8.Bc5-d6 d5xe4 9.Qd3-c4
2. ² (0.37): 2...c5-c4 3.b3xc4 Be7xa3 4.Ra1xa3 Qd7-e7 5.c4xd5 Qe7xa3 6.d5xc6 Ra8-d8 7.Ne4-d6 h7-h6 8.Nf3-d4 Rd8-b8 9.g2-g3
3. ² (0.57): 2...Nc6xa5 3.Nf3-g5 Be7xg5 4.Ne4xg5 f7-f5 5.Ba3xc5 Na5xb3 6.Bc5xf8 Nb3xa1 7.Bf8-d6 Na1-b3 8.c3-c4 Nb3-a5 9.Bd6xc7
4. ² (0.65): 2...Qd7-e8 3.Ne4xc5 Ra8-b8 4.Nf3-d4 Nc6xd4 5.c3xd4 Qe8-b5 6.Qd3-d2 Rf8-c8 7.Kh1-g1 h7-h6 8.f2-f4 Be7xc5 9.Ba3xc5
(, 17.03.2008)
as I said, I went manually through some lines (5 or 6 moves further), also I used 2.3.2e13 which gives higher scores than 2.3.2a in this position.
Sorry, but I think the position is way too complicated to settle this after going '5 or 6 moves through some lines'. (And didn't Larry say that the latest versions give higher scores in general?) To give just one example, in your main line, after 1...Nxa5 2.b4 Nb7 3.Nb3 Nb5 4.Nfd4 Nxd4 5.cxd4 Rfc8 6.Rec1 bxc5 7.bxc5 (dxc looks stronger to me?!) Bg5 followed by 8...a5, there's still everything to play for. Furthermore, I am not sure why black should even allow this structure. Why not, say, 3...Rfb8 4.Nfd4 Nd8 followed by bxc? I agree that white is better anyway, but I think it's far from lost (certainly not resignable).
I think he was too pessimistic about his position and maybe miscalculated somewhere and thought it would be lost. Maybe he just had the right feeling, but I didn't expect the score to be that high, too. Of course the analysis is still far away from 100% sure. But it should be better than the "without-help" output (at least I hope :) )
I remember that Hübner gave an interview after the game where he said that he had played several games against computers before the tournament and developed the habit to resign whenever he had overlooked something. (Quite an unusual attitude, that's for sure. At least I would rarely last till move twenty if I adopted this.) And here he realized that the intended bxc would run into Ne4 because of the uncovered queen and just resigned. I am not sure he thought it was really lost.
The problem with game 7 is simple. Rybka is tuned to play normal chess, from the normal starting position. When a queenside pawn is removed, the effect on strategy is not too big, Rybka still aims to castle short normally. But when a kingside pawn is removed, Rybka is left rather baffled as to what to do. She is reluctant to castle kingside with the pawn missing, but castling queenside may just not be practical for Black. The problem is especially severe when the missing pawn is the "g" pawn, because Rybka especially hates to castle short when that pawn is missing.
We had to do the handicap this way to complete the "trilogy" (first white pawns handicap, then alternating colors, and now Black pawns handicap). For the future, I think that it is not a good idea to continue to offer all 8 pawns in turn. We want the most normal chess possible given a handicap, and in my opinion this means only the four pawns from "c" thru "f". The knights' pawns make Rybka want to play an early ...Rb8 or ...Rg8 to occupy the half-open file, which is rather strange. The edge pawns are inappropriate as the rook is already on the edge file. I think the "c" pawn is the purest handicap, while the "f" pawn is the largest one, with a large body of theory and games from the 19th century.
Perhaps the next time we play a pawn handicap match, it should be c,d,e,and f pawns with each side playing White and Black once. This would be a true pawn handicap, without the weirdness associated with the flank pawns' removal.
We had to do the handicap this way to complete the "trilogy" (first white pawns handicap, then alternating colors, and now Black pawns handicap). For the future, I think that it is not a good idea to continue to offer all 8 pawns in turn. We want the most normal chess possible given a handicap, and in my opinion this means only the four pawns from "c" thru "f". The knights' pawns make Rybka want to play an early ...Rb8 or ...Rg8 to occupy the half-open file, which is rather strange. The edge pawns are inappropriate as the rook is already on the edge file. I think the "c" pawn is the purest handicap, while the "f" pawn is the largest one, with a large body of theory and games from the 19th century.
Perhaps the next time we play a pawn handicap match, it should be c,d,e,and f pawns with each side playing White and Black once. This would be a true pawn handicap, without the weirdness associated with the flank pawns' removal.
Maybe...
But some of that weirdness will have to be overcome for Chess960. And more importantly, so what? The position exposes warts? This is a bad thing? A lowly GM can get the program given certain kinds of pawn odds? Wow.
Importantly, we all learned something, y'all learned something. That is one of the best side effects of these things. Besides all the gambling income and the rough trade associated with this kind of activity. Or is that politics? I can never remember
But some of that weirdness will have to be overcome for Chess960. And more importantly, so what? The position exposes warts? This is a bad thing? A lowly GM can get the program given certain kinds of pawn odds? Wow.
Importantly, we all learned something, y'all learned something. That is one of the best side effects of these things. Besides all the gambling income and the rough trade associated with this kind of activity. Or is that politics? I can never remember
You are right that the knowledge of this weakness with g7 missing may be useful to us, especially if we ever get serious about chess960. But if the idea of the handicap match is to measure the gap between GM and Rybka in normal chess it is not so useful. The f7 handicap, although probably larger, is much more "normal" in the sense that the program will still normally be happy to castle short if it's not too dangerous.
It sound like a good explanation that in game 7, the "g" pawn missing from the starting point has made Rybka more weaker (positional point of view) that in "normal chess".
So this match showed an surprising and unexpected weakness from the engine in the "b" or "g" pawn missing : i think it is a good news news for the Humans players as Rybka is the top level engine :)
ps : i think that the "ab-normal chess" expression when "g" pawn is missing applies only for engines (no book, less knowledge, etc...). I guess top-GM can play "normal chess" with "g" pawn missing (elaborate plan, etc...)
So this match showed an surprising and unexpected weakness from the engine in the "b" or "g" pawn missing : i think it is a good news news for the Humans players as Rybka is the top level engine :)
ps : i think that the "ab-normal chess" expression when "g" pawn is missing applies only for engines (no book, less knowledge, etc...). I guess top-GM can play "normal chess" with "g" pawn missing (elaborate plan, etc...)
> ps : i think that the "ab-normal chess" expression when "g" pawn is missing applies only for engines
Rybka performed worse without her b/g pawns than without her f pawn because a great effort and research was used for the f pawn handicap.
It's true that there was more book on the f7 handicap, but the main point is that the f7 handicap does not cause Rybka to fear kingside castling, whereas the g7 handicap does.
I am not sure that it only was because a great effort and research was used in the f pawn handicap.
Rybka and all programs clearly play less strong and never at 3000 elo in some closed positions or when those are out of book too early. They clearly presents some problems to develop their pieces correctly and also many other problems.
Many people believe that the program can play always at 3000 level or higher and that is a mistake. There is not reason to scare if we see more games like the seven game against Dzindzi. To defeat the machines in these positions is still, quite possible.
Regards,
Gaмßito.
Rybka and all programs clearly play less strong and never at 3000 elo in some closed positions or when those are out of book too early. They clearly presents some problems to develop their pieces correctly and also many other problems.
Many people believe that the program can play always at 3000 level or higher and that is a mistake. There is not reason to scare if we see more games like the seven game against Dzindzi. To defeat the machines in these positions is still, quite possible.
Regards,
Gaмßito.
![[hu]](/mwf/v12/flags/hu.png "Hungary")
That's an interesting explanation.
Iweta talked to Emil Sutovsky before the first Ehlvest match and he claimed that the biggest handicap was the g2 pawn.
Vas
Iweta talked to Emil Sutovsky before the first Ehlvest match and he claimed that the biggest handicap was the g2 pawn.
Vas
Chess is a statistical thing. If you take any 2100 player in the world, his best game is better than Kasparov's worst game.
Dzindzi did quite well against Rybka, but not mindblowingly well - he got maybe one point more than I would have expected. Rybka loses (and loses badly) all the time - just go visit the engine room on Playchess.
Vas
Dzindzi did quite well against Rybka, but not mindblowingly well - he got maybe one point more than I would have expected. Rybka loses (and loses badly) all the time - just go visit the engine room on Playchess.
Vas
Topic Rybka Support & Discussion / Rybka Discussion / Is Rybka better than DeeperBlue (game 7 in odds match) ?
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill