Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / GM Joel Benjamin vs. Rybka: Game 2
1 2 Previous Next  
- - By Felix Kling (Gold) [de] Date 2008-01-04 01:13
[White "Benjamin"]
[Black "Rybka"]
[WhiteElo "2575"]
[BlackElo "3100"]
[Result "1/2-1/2"]
[GameID "171"]
[UniqID "759923"]
[WhiteClock "0:13:16"]
[BlackClock "0:24:17"]
[Stamp "1257"]
[LastMoves "67.h3 Kh6"]

1.e4 {This is game 2 of draw odds match. Rybka must win to get a point. First game was finished in Rybka favour.} d6 2.d4 Nf6 3.Nc3 g6 4.Nf3 Bg7 5.Be2 O-O 6.O-O a6 7.Re1 e6 8.e5 dxe5 9.Nxe5 Nfd7 10.Bf4 { N} Nxe5 11.Bxe5 Nc6 12.Bxg7 Kxg7 13.d5 exd5 14.Nxd5 Be6 15.Bf3 Bxd5 16.Bxd5 Qf6 17.c3 Rad8 18.Qf3 Na5 19.Rad1 c6 20.Be4 Qxf3 21.Bxf3 Nc4 22.Rxd8 Rxd8 23.Re7 Rb8 24.Re2 Kf6 25.Kf1 Rd8 26.Ke1 g5 27.Rc2 a5 28.Be2 Nb6 29.Rd2 Rxd2 30.Kxd2 Nd5 31.Bg4 Ke5 32.Bc8 b5 33.Bd7 Kd6 34.Bg4 Ne7 35.Be2 f6 36.g3 f5 37.Ke3 Nd5+ 38.Kd2 Nb6 39.Kd3 Ke5 40.Bf3 c5 41.Bc6 c4+ 42.Kd2 b4 43.Be8 Nd5 44.Bf7 bxc3+ 45.bxc3 f4 46.gxf4+ gxf4 47.Bxd5 Kxd5 48.f3 a4 49.a3 Kc6 50.Ke2 Kd6 51.Kf2 Ke5 52.Kg2 Kf6 53.Kf2 Kf7 54.Kg1 Kg6 55.Kf2 Kf6 56.Kg1 Kg5 57.Kf2 Kh4 58.Kg2 h6 59.Kf2 Kh5 60.Kg2 Kg5 61.Kf2 Kg6 62.Kg2 Kh5 63.Kh3 Kg5 64.Kg2 Kf5 65.Kh3 h5 66.Kh4 Kg6 67.h3 Kh6 { 1/2-1/2 (White wins)} 1/2-1/2

The game ended in a draw which means that the current standing is 1-1.
Parent - - By Uly (Gold) [mx] Date 2008-01-04 01:26
When did this game went wrong?

If 500 ratings points of difference aren't enough to win, we have a problem.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-04 02:02
But do we have a problem if 500 rating points aren't enough to win when the stronger side plays the Pirc?! with black?
Parent - - By lkaufman (*****) Date 2008-01-04 03:46
I think we went wrong in the opening book. I was a bit worried about the line he chose, but I couldn't find a better option quickly (after move 6) and with hundreds of possible openings to prepare for, I didn't have more time to try.
Parent - - By FirebrandX (**) [us] Date 2008-01-04 04:03
The Pirc just sucks for comp play. I tried it once myself in an engine-engine game and the draw was unavoidable and stagnant. Have Rybka play the Sicilian for Christ's sake!
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-04 04:48
It may be that even with anti-human settings, Rybka would have a problem if the human (to wit: Benjamin) played 2.c4 to lock things up nicely.
Parent - - By BB (****) [gb] Date 2008-01-04 15:03

> It may be that even with anti-human settings, Rybka would have a problem if the human (to wit: Benjamin) played 2.c4 to lock things up nicely.


Seems that Joel wasn't listening - he played 2 Nf3 instead...
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-04 15:20
So I see...and I predict that he will lose this game...(it's at move 10 now)
Parent - - By FirebrandX (**) [us] Date 2008-01-04 17:19
Sweet! He had Rybka play it! Looks like Rybka has Joel wrapped up too.

1. e4 c5 2. Nf3 e6 3. Nc3 a6 4. d4 cxd4 5. Nxd4 Qc7 6. g3 Bb4 7. Bd2 Nf6 8. Bg2 Nc6 9. Nxc6 dxc6 10. O-O O-O 11. Nb5 Qe7 12. Bxb4 Qxb4 13. Nd6 Qxb2 14. Rb1 Qxa2 15. e5 Nd5 16. Rb3 b5 17. Qc1 b4 18. Bxd5 exd5 19. Rxb4 Bh3 20. Re1 Rab8 21. Rh4 Be6 22. Qd2 c5 23. Qd3 h6 24. Nf5 Rb1 25. Rxb1 Qxb1+ 26. Kg2 d4  (white is lost)
Parent - - By JohnL (***) Date 2008-01-04 17:37
Now we're talking, e6 sicilian, great choice :-)

But how could Benjamin go into that pawn sac variation?? Theory?
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-04 18:30
It didn't necessarily need to be a pawn sac at the beginning of that variation, as far as I recall in the game, but that's how it wound up, especially after Rb1.
Parent - - By lkaufman (*****) Date 2008-01-04 05:03
Why didn't I think of that? Seriously, I have a serious concern about playing the Sicilian vs. Joel, but I'll probably try it anyway.
Parent - - By Banned for Life (Gold) Date 2008-01-04 05:29
Maybe the right idea here is to play a somewhat inferior opening that isn't drawish and hope Rybka can overcome the initial disadvantage? I'm wondering how a B01 would play out...

Regards,
Alan
Parent - By Graham Banks (*****) [nz] Date 2008-01-04 05:38
A Larsen might be worth a go.
Parent - By lkaufman (*****) Date 2008-01-04 18:27
We tried that approach against Ehlvest in the 3 move book match, and he drew half the games with White. I would try this if normal openings fail, but so far we're in good shape.
Parent - By FirebrandX (**) [us] Date 2008-01-04 05:54
Why be afraid? If there's some glaring weakness Joel exposes of Rybka in the Sicilian, it would serve well for providing info on areas to improve newer versions.

Anyway, if people are suggesting wild lines now, might as well have Rybka play the Icelandic gambit :P
Parent - - By Razor (****) [gb] Date 2008-01-04 06:14
You need to use unorthodox openings; Joel will not be so well prepared against these.
Parent - By lkaufman (*****) Date 2008-01-04 19:06
I guess Philidor was a bit "unorthodox opening". So far a mixture of openings is working well.
Parent - - By oudheusa (*****) [ch] Date 2008-01-04 17:22
Yes I agree, you should go for unusual openings. Play a6 followed by g6 after e4 and eg b5 after d4. Only a couple of book moves are enough and Benjamin will have to start thinking from the start. I believe Rybka will considerably benefit from getting unusual openings against an (unprepared) human GM.
Parent - By lkaufman (*****) Date 2008-01-04 19:14
Well, we played the Pirc with ...a6 (which could have happened with your move order) and that was the only game we did not win so far.
Parent - - By Ingo Althofer [de] Date 2008-01-04 08:49
8 games in a row, and all with black, is a long way
against a human grandmaster who is going for
draws only.

Maybe, it will help the Rybka team at least a
little bit that on each day two games are played.
You may exploit this by using a double hit strategy:
When in the first game of a day an opening worked well,
try it again also in the second game of that day. The
grandmaster would have only a few hours to look for
improved reactions.  Especially, after your revolver strategy
from day 1, Benjamin may not expect a double hit attempt.

Good luck,
Ingo Althofer

PS: In the mid-90's I used my 3-Hirn approach to play
several 8-game matches against grandmasters. From
there I am familiar with the problem of opening choice
against humans who try to "outbook" computer
opponents.
Parent - By lkaufman (*****) Date 2008-01-04 19:09
Yes, I remember those matches you ran then. As for double hit strategy, I may do this, but I don't think it matters too much what order I play openings in. With our schedule, after playing, eating meals, answering emails, resting up, etc. Joel only has an hour or two to prepare for any one game regardless of when it is played.
Parent - - By Debaser (***) Date 2008-01-04 11:06
Watching Rybka games yesterday I am still not able to see an anti-human play. On convekta broadcasting you can see the analysis by default and anti-human Rybka and they are very similar. That agresive Rybka you talked about some time ago would be better, but it is still not available?. Rybka should play sharper lines, looking for complications, though perhaps that magic move is not the better positionally, please Vas and Larry will we see here a move like the one in Kasparov-Junior match that chessbase publicite always that they can? ;)

http://www.chessbase.com/newsdetail.asp?newsid=777

IMHO GM Benjamin is playing very well, he remembers me the match Kramnik-Fritz, getting simmetrical lines and exchanging pieces always he can to get an "easy" endgame. Of course he is playing for a draw.
Parent - - By lkaufman (*****) Date 2008-01-04 19:12
Bear in mind that the "anti-human" and "anti-draw" strategies are mostly connected with contempt setting, which is used for the actual game but not for the broadcast "anti-human" version.
Parent - - By Debaser (***) Date 2008-01-05 10:53
So the anti-human version in the broadcast is not the same that is playing GM Benjamin?

In the second day appeared in the broadcast the output of an updated anti-human version, is this the right one that is playing?
Parent - - By lkaufman (*****) Date 2008-01-05 20:16
The version may be the same (I'm not sure) but the displayed version has no contempt set, so in effect it is quite different.
Parent - - By Debaser (***) Date 2008-01-05 22:26
OK thanks.

Sorry, a new question ;), and the differences between the first and the updated anti-human?
Parent - - By lkaufman (*****) Date 2008-01-06 00:49
We have so many versions that I can't keep track of which one is displayed. The difference is probably some small anti-human detail, like avoiding symmetry for example.
Parent - - By Debaser (***) Date 2008-01-06 13:53
Your engines folder must be a mess then, be careful and do not load fritz in the next game vs GM Benjamin ;)
Parent - - By Vasik Rajlich (Silver) [hu] Date 2008-01-06 15:01
Ok, for those of you who are morbidly curious:

Rybka 2.3.2e14 was the 'starting point'
Rybka 2.3.2f14 was 2.3.2e14 + more aggressive handling of completely locked positions
Rybka 2.3.2m14 was 2.3.2e14 + avoidance of symmetrical and near-symmetrical pawn structures
Rybka 2.3.2u14 is 2.3.2m14 + more aggressive handling of completely locked positions + a few technical changes

In the big scheme of things, the differences between these four versions are quite small.

Vas
Parent - By Debaser (***) Date 2008-01-06 16:07
Yes, really a mess. Thanks for the info.

With morbidly curious you are talking about somebody else, I am just lightly curious ;)
Parent - - By Uly (Gold) [mx] Date 2008-01-07 21:59
Why isn't there a 2.4 version, not even a 2.3.3!?

Are you just going to suddenly jump to 3.0 directly?
Parent - - By dareapa (**) [us] Date 2008-01-07 22:06
I think that 2.3.2a is basically the final update to "2.0", so 2.3.2a is really "Final Rybka 2.0" Am I correct guys?  So the next public release should be Rybka 3.0.  The rybka's that we see now 2.3.2xxx. Are really versions leading to 3.0. ( I think)..Am I on the right path guys??
Parent - By Vasik Rajlich (Silver) [hu] Date 2008-01-08 14:41
Yes, exactly. Rybka 2.3.2a is basically Rybka 2. The next one is Rybka 3.

Vas
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-07 23:44
Why not jump to 3.0 directly?  Must everyone always work in base 10 all the time?
Parent - - By Uly (Gold) [mx] Date 2008-01-08 00:18
Having base 10 versions between 2.3.2a and 3.0 would give us hints about the release date (And amount of improvement). Once we see 2.9.3, we know that 3.0 is very close.
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-08 01:41
But such things really aren't known, for one thing, and for another, there aren't always 10 versions between major versions.
Parent - - By Uly (Gold) [mx] Date 2008-01-08 04:28
Ok, you're right. I still don't like to see these 2.3.2xx versions' numbers, they give the illusion that we're stuck (at 2.3.2) and have made no improvement.
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-08 04:34
I think they do that so as not to give the illusion that there will be any forthcoming, currently non-market Rybka 2 version in the future.  If they said something like Rybka 2.4 or Rybka 2.3.3, then everyone would be asking, "where's my copy?!".  If it was called something like Rybka 3 alpha or Rybka 3 alpha xy, everyone would say, "Hey, I want to buy Rybka 3 early and get this alpha version!" when it might not be entirely stable or ready for the consumer market.
Parent - - By Uly (Gold) [mx] Date 2008-01-08 04:44
I still want my copy of Rybka 2.3.2e14 :)

Another thing I don't like is that the versions' names don't follow logic. For example, a previous version was called 2.3.2g7, and the g comes after the e that was used later. That means that what is more important is the number, but then the 7 of g7 is like version 7.0, but the next e14 version looks like a 1.4 version, which means that the g version should have been 2.3.2g07 (And, it's not about the year, because we're not yet at 2014.)

I find it a mess, as if version numbers did not matter and were just cosmetic.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-08 04:56
I guess I didn't go further on the "where's my copy?!" part, though I had meant to--the problem here is that Rybka 2 customers would basically be getting a Rybka 3 product for free and wouldn't have as much motivation to buy Rybka 3.
Parent - By Banned for Life (Gold) Date 2008-01-08 00:50
The 2.4 version has been playing on the CB server for quite a while now. It really isn't very good. :-)

Regards,
Alan
Parent - By Vasik Rajlich (Silver) [hu] Date 2008-01-08 14:40
The new public convention will be simple: Rybka 3, Rybka 4, Rybka 5, etc.

Internally, we have lots of versions with tiny changes.

Vas
Parent - - By Vasik Rajlich (Silver) [hu] Date 2008-01-05 14:17
Of course, we have been working on this, although it's not perfect.

Keep in mind that this type of .. Bxh2+ combination is a drawing attempt and wouldn't be suitable for this match. Keeping the tension is a tricky topic.

Vas
Parent - - By Debaser (***) Date 2008-01-05 14:55
Oh. I never thought that ...Bxh2+ was a drawing attempt. I always thought that the reason was to force a mistake by Kasparov (well, perhaps he was not the one you would expect a blunder). And so they got a draw only because Kasparov played the right continuation. But GM Benjamin is not Kasparov.

Anyhow what I meant is that with spectacular moves like this you should get a lot of interest in the match and a nice advertising for the future. You can read in several fora that Rybka is the best, yes, but her play is boring, expecting for a mistake, not attacking and so on, and perhaps this is not the best way to play a match with draw odds and to get publicity ;)
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-05 15:57
Ahh, that old ...Bxh2+?! question again...

It definitely wasn't the best move in the position--I've analyzed it a bit awhile ago, and black had an advantage before that move, and that gives him a draw at best, and may give white a significant advantage with correct play.  Of course, if it's against a human, the probability of perfect play in such a position is very tiny, even for Kasparov, so it might be worth the gamble, and it somewhat paid off: Kasparov had to settle for a draw in one of his white games.
Parent - - By Uly (Gold) [mx] Date 2008-01-05 23:35

> Of course, if it's against a human, the probability of perfect play in such a position is very tiny, even for Kasparov, so it might be worth the gamble


This is the main point. Rybka should try to play "unsound" moves if it's likely that they'll work against a 2600 player.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-06 00:38
In that case, a separate version would definitely have to be released specifically for GM play, in spite of the fact that an incredibly small percentage of Rybka's games are against such opponents.  I'm not so sure that such a thing would be worth it.
Parent - - By JohnL (***) Date 2008-01-04 12:16
I think a problem with this kind of match is that it is very much about opening book. This means that the skill and effort in preparation by the Rybka team plays a crucial role for the result.

In chess960(FRC) Joel would have much less chances to draw.

By the way, I would go for the sicilian vs e4 and perhaps Leningrad dutch vs d4 but it would require a really extensive non-balance book.
Parent - By lkaufman (*****) Date 2008-01-04 19:13
You are right about the book. Joel himself said he would have no chance in a FRC match. He probably would lose every game. As for sicilian, as you see I tried it with success.
Up Topic Rybka Support & Discussion / Rybka Discussion / GM Joel Benjamin vs. Rybka: Game 2
1 2 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill