Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Benjamin vs. Kasparov draw odds match
- - By purplejacket (**) Date 2008-01-07 08:17
So my question is:  if this match were played with the exact same conditions (Benjamin has draw odds, plays all white in 8 games, 2 games per day, time limit of 90' plus 30"), except it was Kasparov instead of Rybka playing the black pieces, what kind of result would we have expected?  Would Kasparov have tried to create complications in the same way that Rybka did?  Or would it have been more about opening novelties.  And I use Kasparov as the "super-GM" to compare with because I think he would be the human best able to hope for a result similar to that obtained by Rybka.
Parent - - By JhorAVi (***) Date 2008-01-07 08:37
I believe Kasparov can go further by offering pawn odds against joel.
Parent - - By Permanent Brain (*****) Date 2008-01-07 09:22
Your confidence in Kasparov is ok, but by that you virtually claim that he is stronger than Rybka.

We learned that Rybka can win 7-1 against a 2574 GM, which is an Elo performance of ~2900+x (x for the fact that Rybka always had black). Therefore I see no reason to expect more from Kasparov against Benjamin, than from Rybka against Benjamin, even less with pawn odds in addition. For a normal match with alternating colors, 6-2 or 6.5-1.5 seem realistic, but always having black reduces the expectations.

This is purely speculative anyway, some will believe this, some that... :-) We will not come to a clear conclusion, or agreement about it.
Parent - By JhorAVi (***) Date 2008-01-07 15:10
Playing with a Draw condition is different.  Kasparov knows what's going on and can sense all of Benjamins intent to simplify. He knows what positions Benjamin hates and can dictate the nature of the game. And more importantly, Kasparovs Psychology which is useless against Rybka can work in full force against Humans!

No doubt Rybka is stronger than Kasparov but this special case is like Boxing where Benjamin is beaten by style and not by strength.
Parent - - By insipid (**) Date 2008-01-07 19:14
Kasparov won a pawn odds match only 2.5-1.5 against Terrence Chapman many years ago.  I highly doubt he would be able to give a GM a pawn and white and survive.  You don't really think Kasparov is stronger than Rybka, do you?   Especially now, after much inactivity.  I voted for a 7-1 score for Rybka in this match, so Joel did better than I expected, but still, the computer is an absolute monster and as long as the opening doesn't allow a dried up position (like the Pirc game did) I don't think any regular GM has a chance and I highly doubt Kasparov could do any better.

Kasparov might have an advantage in that he would be willing to take larger risks to change a drawn position into a win than Rybka did.  But considering the contempt value, he might not even have that.  Rybka, at a FIDE time control, with a proper book, would score better than Kasparov in any practical contest, in my opinion.  This match is strong evidence for that idea, to me.
Parent - - By BB (****) Date 2008-01-07 19:20 Edited 2008-01-07 19:36

> Kasparov won a pawn odds match only 2.5-1.5 against Terrence Chapman many years ago.


This is only half the story (that is, it was 2 pawns odds): see this link for more details. There was also 90/60 time odds. There was some post-match comment about 2 pawns [especially a central pawn, when Kasparov had White] being worth significantly more than twice 1 pawn, but LK would probably be the expert about that.
Parent - By insipid (**) Date 2008-01-07 19:29
Ah yes, you are correct, my mistake.  Thanks.  Kasparov did have a couple whites, however, but yes two pawns is certainly much larger a handicap than one.

Still, that was a match where I (and many others, I think) was surprised by how close it was.  I wonder how Rybka would do in the same spot against Chapman.  I think 4-0, but I could be wrong again!

This match, on the contrary, I think surprised most of the readers here (and they in theory are the LEAST likely to be surprised) by how lopsided it was.  I don't see any logical reason to believe Kasparov could outperform Rybka in this sort of situation. 
Parent - - By Uri Blass (*****) Date 2008-01-07 20:06
What about a contest in advanced chess when both kasparov and rybka are allowed to use other chess programs to help them?
I am not sure if rybka is going to score better than kasparov in this contest.

Another interesting question is if the best humans can learn from rybka to improve their rating to 2900.
Remember that in the past the top humans did not have better opponents to learn from losses against them and now they have better opponents.

The question is how many rating points the top humans can earn by training against better opponents.  

Uri
Parent - - By 8lrr8 (***) Date 2008-01-08 07:28
"The question is how many rating points the top humans can earn by training against better opponents."

i'd be shocked if kasparov's 2851 ever gets breached.  humans learn chess, and more importantly, get stronger in their chess game, in radically different ways than how programs gets stronger.  if a mega-grandmaster (i.e. 2700+ elo) spars against rybka quadcore day in and day out, and gets his assed kicked left and right, unless he learns exactly why he's losing, he may not get much stronger at all.

w/ that being said, what we are seeing as a clear trend is that the space at the top is getting crowded.  by that i mean if u go back a few yrs, there were only ~15 people in the world who were mega-grandmasters.  the current list has 24 players w/ a rating of at least 2700.  so what i think we'll see is the number of mega-grandmasters increasing, but i dont think we'll see a huge influx of players ever breaking the 2800 barrier.
Parent - - By Uri Blass (*****) Date 2008-01-08 12:16
I agree that a grandmasters cannot learn from losing if they do not learn from their mistakes but I see no reason to assume that they cannot learn from mistakes.

The main problem is that if all the top players learn from rybka and if most of their games are not against weaker players who do not learn from rybka then the progress in rating is going to be slower than the real progress that they deserve.

I wonder if it is possible to have a good program that can estimate the elo of active human players based on chess games that they play with error that is not more than 100 elo in 90% of the cases.

If it is possible to do it then it may be interesting to see what rating this program can give to players of the past including kasparov at the time that he got his 2851(january 2000).

Note that in the past I read that Fritz3 had a program that estimated the rating of chess players based on games but people who tested it found that it evaluated kasparov as 2100 or something like that when it evaluated itself as 3000 or something like that(I do not remember exact numbers).

Uri
Parent - By Permanent Brain (*****) Date 2008-01-08 12:47
I think the rating feature you mention, must have been from another program. I have Fritz 3 (the last MS DOS-Fritz) but it doesn't have that feature. I seem to roughly remember that function though, too... but I entirely forgot which program might have had it.
Parent - By 8lrr8 (***) Date 2008-01-08 14:04
"I wonder if it is possible to have a good program that can estimate the elo of active human players based on chess games that they play with error that is not more than 100 elo in 90% of the cases."

i dont see why not.  altho this elo evaluating program might need to tweaked from its original version (e.g. "rybka 3.0" ---> "rybka 3.0 elo eval version")
Parent - - By turbojuice1122 (Gold) Date 2008-01-07 12:15
I think that Kasparov would definitely have scored at least 4-4 (he would have "won" both games by draws yesterday).  I think in reality that a correct estimate would have Kasparov winning 6-2 under these conditions.
Parent - - By skulibj (*) Date 2008-01-07 13:06
Isn't Kramnik a better example? Isn't he a draw master?
Parent - By turbojuice1122 (Gold) Date 2008-01-07 14:48
Kramnik is better at keeping the games non-tactical, but he would have trouble, I think, in the games we saw yesterday--he would still have to wade his way through a host of tactical complications--Kasparov is better at this.
Parent - By 8lrr8 (***) Date 2008-01-07 14:49
doesnt leko have a higher career draw percentage?
Parent - - By lkaufman (*****) Date 2008-01-07 16:03
Kasparov and Benjamin actually did play in 1994, and despite playing Black Benjamin held the draw. I think that a match between them with the same terms as this Rybka match would be quite fair, with 4-4 being my best guess. Kasparov always relied heavily on opening theory, counting on his opponents to fight with the White pieces. But if White aimed for a draw, Kasparov would have problems. If he tried to play his favorite Najdorf, Benjamin would play 3 Bb5 check (as he often does normally), which is notoriously hard for Black to win against.
Parent - - By turbojuice1122 (Gold) Date 2008-01-07 16:26 Edited 2008-01-07 16:29
Seeing that this is draw odds with Benjamin having all eight games white, it would seem surprising if he's only able to manage four draws, when searching for a draw, against an opponent "only" 200-250 points stronger.

Also, this 3.Bb5+ and the difficulties of winning thereafter is what you were fearing about the Sicilian, I now presume?
Parent - - By lkaufman (*****) Date 2008-01-07 17:03
No, because as you saw, I chose 2...e6 rather than 2...d6.
Parent - - By turbojuice1122 (Gold) Date 2008-01-07 17:27
Yes, and I was happy to see this choice--what I'm referring to is the types of games that arise in the Bb5 variation--were you fearing that these same types of situations could arise in the ...e6 variations?  In particular, I have game 4 in mind, where Joel basically turned it (or allowed it to be turned?) into a Father-like game.  Were you coming across these types of things in your preparation, and are they similar to what you were afraid of in the Bb5 variations?
Parent - - By lkaufman (*****) Date 2008-01-07 22:05
Mostly I was concerned about the Alapin, which would have arisen in Game 4 if Joel had chose 4e5 rather than the Kopec system he used. It is considered a good way to play for a draw. But I was also concerned about various Closed Sicilians in general.
Parent - - By turbojuice1122 (Gold) Date 2008-01-07 23:25
Wow, the Alapin--I was way off in my line of thought.  I know that the Alapin is often used as a drawing method or to get only a small advantage against human players, but I wonder if it would really be able withstand a >3000 elo opponent.  I guess one could try a noticeably weaker engine like Hiarcs 11.2 as white against Rybka 2.3.2a as black and only allow the Alapin in the opening books and see what happens in a bunch of rapid games.
Parent - - By lkaufman (*****) Date 2008-01-07 23:41
I don't say that Joel could be expected to hold a draw with the Alapin, but I think his chances of doing so would have been considerably better than in the Philidor line I chose.
Parent - - By M ANSARI (*****) Date 2008-01-08 06:18
I think all of you underestimate Kasparov.  I doubt he would have any problem easily drawing every game.  But Kasparov being Kasparov would probably try to win and not only draw and thus more chances for a loss.  But really ... let's not get carried away here ... Rybka can easily be outplayed by Kasparov, Topalov or Kramnik.  Whether this is enough to beat Rybka in a match is doubtful ... but to draw ... I think it would be a piece of cake.
Parent - By lkaufman (*****) Date 2008-01-08 15:10
If Joel is a class below the top few guys, they should beat him by 6-2, the same margin as he lost this match by. This suggests that a draw and move match between Rybka and the top few guys would be a tossup-- they should be able to draw about half the games as White. This feels right to me.
Parent - By Permanent Brain (*****) Date 2008-01-08 09:07
In the 1990s, Benjamin also scored black wins against Smirin, Bareev or Svidler. In 2004, he drew a Sveshnikov Sicilian with Black against Kamsky who was rated 2717.

[Event "ch-USA"]
[Site "San Diego USA"]
[Date "2004.11.30"]
[Round "6"]
[White "Kamsky, Gata"]
[Black "Benjamin, Joel"]
[Result "1/2-1/2"]
[ECO "B33"]
[WhiteElo "2717"]
[BlackElo "2554"]
[PlyCount "102"]

1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 e5 6. Ndb5 d6 7. Bg5 a6 8.
Na3 b5 9. Nd5 Be7 10. Bxf6 Bxf6 11. c3 O-O 12. Nc2 Rb8 13. h4 Ne7 14. Nxf6+
gxf6 15. Qd2 f5 16. Qg5+ Kh8 17. Qf6+ Kg8 18. O-O-O Be6 19. Bd3 Nc6 20. Qh6 f4
21. g3 f3 22. Qg5+ Kh8 23. Qxd8 Rfxd8 24. b3 a5 25. Kb2 b4 26. c4 Kg7 27. Bf1
a4 28. Bh3 Ra8 29. Bxe6 fxe6 30. Rd3 Ra6 31. Rxf3 Rda8 32. Ra1 axb3 33. axb3
Rxa1 34. Nxa1 Nd4 35. Rd3 Kf6 36. Nc2 Nxc2 37. Rf3+ Ke7 38. Kxc2 Ra2+ 39. Kd1
Ra1+ 40. Ke2 Ra2+ 41. Kf1 h5 42. Rd3 Rc2 43. Kg2 Rc3 44. Re3 Rc2 45. Kf3 Rd2
46. Kg2 Rc2 47. Rd3 Rc3 48. Rf3 Rc1 49. Kh2 Rf1 50. Kg2 Rc1 51. Re3 Rc2 1/2-1/2
Up Topic Rybka Support & Discussion / Rybka Discussion / Benjamin vs. Kasparov draw odds match

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill