Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / GM Ehlvest - Rybka. Game 5 (draw)
- - By Victor Zakharov (*****) Date 2007-07-07 23:39
[Event ""]
[Site ""]
[Date "2007.7.8"]
[Round ""]
[White "*Ehlvest"]
[Black "*Rybka"]
[Result "1/2-1/2"]
[Eco ""]
[Annotator ""]
[Source ""]

1.c4 c6 2.d4 d5 3.Nf3 g6 4.e3 Nf6 5.Nc3 Bg7 6.h3 O-O 7.Bd3 c5
8.O-O cxd4 9.exd4 dxc4 10.Bxc4 Nc6 11.Re1 Na5 12.Bf1 a6 13.Bg5
Be6 14.Qd2 Nc4 15.Bxc4 Bxc4 16.Rad1 Bd5 17.Ne5 Re8 18.Ng4 Nxg4
19.hxg4 f6 20.Bh6 Bxh6 21.Qxh6 e6 22.g5 fxg5 23.Re5 Qf6 24.Rxg5
Bc6 25.Re5 Red8 26.Qe3 Rd6 27.Rd2 Rc8 28.Re2 Rcd8 29.Ne4 Bxe4
30.Rxe4 Rxd4 31.Rxd4 Rxd4 32.Qxe6+ Qxe6 33.Rxe6 Rb4 34.b3 Kf7
35.Re3 Rd4 36.a3 Rd1+ 37.Kh2 b6 38.g4 Kf6 39.Kg3 g5 40.b4 Rc1
41.f3 b5 42.Kh3 h6 43.Kg3 Rc2 44.Kh3 1/2-1/2
Parent - - By Roland Rösler (****) Date 2007-07-08 00:42
First time I´m a little bit disappointed by Jaans play. He had the initiative but can´t go to a slightly better or even rook endgame with some imbalances. Then he will have very good winning chances. Up to now there are two rook endgames and Rybka without tablebases is leading 1,5 - 0,5. This isn´t the reality!
Okay, the first rook endgame Jaan was a pawn down and the positions was very unusual and complex and he has time trouble. Excused! But in the 5th game, there was a kind of resignition in his play in the ending.
I read the chat from Vas and Larry and now I´m down. They really believe that Rybka has chances to win this rook endgame (because the evaluation was about -0.2 or so?) and it will be a mercy to offer a draw to Jaan. I don´t like this! Okay, human blunder (because there is one move they do not consider in their plan) but in general you can be absolutely sure that Rybka blunders in rook endgames (because Rybka has no plan and go consequently in the wrong direction because depth doesn´t help).
PS: I hope, we will see an other rook endgame in the last game.
Parent - - By lkaufman (*****) Date 2007-07-08 02:36
I wouldn't say I believed Rybka had any real chance to win this endgame. I just don't want to offer or accept a draw for the computer until it is pretty obvious to everyone that Rybka is not trying to win. I like the idea of the "Sofia" rules, where every game is played until the draw is obvious to even a very weak player.
Parent - By Hetman (*****) Date 2007-07-08 09:42

if Rybka will win by too big margin  would it not have been the bigger problem to find next GM to demolish. :-)

Parent - By Hetman (*****) Date 2007-07-08 07:04
congratulations for GM Ehlvest for the recovery, we shall not demand too much from the human player :-).
He has found the power to play after 2 defeats in the row  !
He is not the computer and the story of the match is influencing him. When you are losing by a margin 1st is to retain confidence, then you will be able to play for the win.
The win against programm is very difficult . It is unbelieveable what tactical shots the programm could find to survive. :-)
The next draw is probable.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-08 08:45
Rybka was nominally better in this endgame, not because she had an eval score of -0.2 (although it's a good sign) but because she had a better rook and a slightly better king.

It was pretty obvious that the winning chances were more-or-less non-existent, but I agree with Larry's reasoning: let the human defend it a bit. If it's really easy, as I think it was, then it won't cost him much energy.

Parent - - By Sesse (****) Date 2007-07-08 09:38
Did Rybka really have an evaluation score of -0.2? On the quad here (although in 32-bit mode), with 5-man tablebases (that were never used, though), ISTR it was -0.06 at the point Ehlvest offered a draw.

/* Steinar */
Parent - By Vasik Rajlich (Silver) Date 2007-07-08 18:12
Yes, she did, although it's a bit optimistic.


*Ehlvest - *Rybka,  2007
8/1p3k1p/p5p1/8/3r4/1P2R3/P4PP1/6K1 w - - 0 1

Analysis by Rybka 2.3.2a 32-bit :

36.Rc3 Ke6
  ³  (-0.38)   Depth: 5   00:00:00
36.Rc3 Ke6 37.Kf1
  ³  (-0.38)   Depth: 6   00:00:00  3kN
36.Re2 Kf6 37.g3
  ³  (-0.30)   Depth: 6   00:00:00  4kN
36.Re2 Kf6 37.g3 g5
  ³  (-0.32)   Depth: 7   00:00:00  5kN
36.Re2 Kf6 37.g3 g5 38.Kg2
  ³  (-0.29)   Depth: 8   00:00:00  8kN
36.Re2 Kf6 37.g3 g5 38.Kg2 h5
  ³  (-0.29)   Depth: 9   00:00:00  16kN
36.Re2 Kf6 37.g3 Rd7 38.Kg2 Kf5 39.Rc2
  =  (-0.23)   Depth: 10   00:00:00  53kN
36.Re2 Rd1+ 37.Kh2 h5 38.Kg3 Kf6 39.Kf4 Rd7 40.Rc2
  =  (-0.24)   Depth: 11   00:00:01  75kN
36.Re2 Rd1+ 37.Kh2 h5 38.Kg3 Kf6 39.Rc2 g5 40.Kf3 h4
  =  (-0.22)   Depth: 12   00:00:02  117kN
36.Re2 Rd1+ 37.Kh2 h5 38.Kg3 Kf6 39.Rc2 g5 40.Kf3 h4 41.Ke3
  =  (-0.23)   Depth: 13   00:00:03  158kN
36.Re2 Rd1+ 37.Kh2 Kf6 38.g3 b6 39.f4 Kf5 40.Kh3 Rd5 41.Kh4 h5
  =  (-0.22)   Depth: 14   00:00:06  337kN
36.Re2 Rd1+ 37.Kh2 Kf6 38.g3 b6 39.f4 Kf5 40.Kh3 h5 41.Re5+ Kf6 42.Re2 Rg1
  =  (-0.21)   Depth: 15   00:00:09  531kN

(, Rybka 08.07.2007)
Parent - - By Hetman (*****) Date 2007-07-08 14:03
'let the human defend it a bit.'  ?! :-). Excuse me, it sounds for me a bit as the sentence  from science fiction novele. Larry and Vas are not humans ;-)?

It is the question who is accepting, giving  the draw proposal programm or operator ?. If operator then it is not the game between programm against GM but programm + IM vs GM. ?

Parent - - By nine castles (**) Date 2007-07-26 10:07
There is no reason to complain that humans are accepting draw offers for the engine since if you were to make a rule against that, the engine would simply never accept draws. Human operators accept draw offers as a courtesy to their opponents; there is no selfish advantage for them to do so.
Parent - - By Hetman (*****) Date 2007-07-26 17:05
Why the engine could not have the algorithm to reject accept the draw ?
Parent - - By Vasik Rajlich (Silver) Date 2007-07-28 07:51
Actually, it's outside of UCI, but if it weren't, Rybka draw offering and draw acceptance algorithm would be very simple: just say no.

Parent - By Ernst (***) Date 2007-07-28 09:25
Hehehe :-)
Parent - By Hetman (*****) Date 2007-07-28 15:04
Lets wait to the nearest Freestyle. We will see how does it work ?
Parent - - By Roland Rösler (****) Date 2007-07-08 21:27
In 1971 the Russians deprive Mark Taimanov the grandmaster titel because of his 6-0 loss against Bobby Fischer. This was bullshit! But: If a grandmaster will lose this position against Rybka 2.3.2a (without tablebases), he is no more a grandmaster in my eyes.
GMs are very good in playing rook endgames, because this endgame happens by far most frequently and if they don´t play this endgame very good, they will not be GMs. If Jaan would be a gambler, he will go on in this position and waits of the mistakes from Rybka (and they will come very probably). But he isn´t a gambler! In the whole match Jaan plays honest chess without any tricks against the weaknesses of Rybka. Maybe this is fair. But to make your points against Rybka, you have to be a a very good chessplayer (like Jaan) and a dirty gambler (like Nakamura?). And you have to test it in the preparation for the match.
Last statement: Vas, if you can´t see any human challenge for Rybka, let´s play theme matches. I suggest a rook ending match about 8 games (4 positions to play with white and black; Jaan selected the positions, he is the most honest guy I know) between Rybka without tablebases and an IM (maybe later a GM). I bet on the IM.
Parent - By Vasik Rajlich (Silver) Date 2007-07-10 13:29
I certainly agree that humans are relatively better than engines in rook endgames, but this effect isn't so strong that an IM would be favored in a rook-ending match.

Anyway, perhaps we could have shaved a few moves from the tail of this game .. no harm done.

Parent - - By Jim Walker (***) Date 2007-07-08 13:18
I have a very different take on this game and match.  First of all the GM offered a draw first.  I believe it should have been accepted on the spot!   After all the match is over!  He is playing for nothing now and that is very generous on his part.  You want to test Rybka under these conditions and he is doing his best even when all is lost and his money is already fixed.   At this point if the GM wants a draw it should be given unless there are clear winning advantages for the computer.  I believe it's disrespectful to do otherwise.  Thats my story and I'm sticking to it.
Parent - - By Fulcrum2000 (****) Date 2007-07-08 13:42
Yes, I think I agree with you on this one.
Parent - - By Hetman (*****) Date 2007-07-08 13:53 Edited 2007-07-08 14:20
Me, too. But we do not know the Rybka camp obligations to the sponsors.

I can appeal only to them.
Match is decided, lets try to be the humans.:-). I am very satisfied with Rybka but I am human.
Proposing,rejecting draw proposal it is very sensible matter.

GM has not using anti-computer strategy and was not building stonewalls, fortresses starting early exchanges etc.
Was doing himself in the good style. The games were interesting, we have not seen 6 times e4 e5 Nf3 Nc6 Bb5 a6 Bxc6 dxc6 d4 exd4 Qxd4 etc. GM deserves  treatment in similar good style but it is only MSO as the kibitz.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-08 18:14
All parties including GM Ehlvest agreed before the match that all 6 games would be played regardless of the score.

Parent - - By Hetman (*****) Date 2007-07-08 18:26
Thanks for reply but I have meant sth else... Has been agreed that he will not use anticomputer strategy ?
Parent - - By lkaufman (*****) Date 2007-07-08 22:04
Of course not. To some extent he modified his play based on his opponent, for example in the 3...Qd6 Scandinavian game he indicated that he would probably have offered a gambit of the b2 pawn against a human by Qd2 instead of Rb1 but thought it unwise against a computer. But if anti-computer play means trying to make draws by creating a totally blocked position, why should he do this? He would have to do it successfully every single time just to earn the modest extra money for a drawn match, when the big money was only for a match victory. I think his play made sense. He also told me that part of his motivation to play this match was to prepare for major upcoming high-level events; naturally he would not serve that purpose by playing for draws and not fighting with the White pieces. Basically he played his normal game but modified it modestly based on the identity of his opponent.
Parent - By Hetman (*****) Date 2007-07-09 15:11
Wow!For the draw was 3.100 $ >= 2 * 1500 $. It is a difference. It is the motivation. I am standing with that Ehlvest played like gentleman and refusal of his draw proposal in 4th game has been not the best move.
Building the fortresses could have been the kind of training either.
Parent - - By Jim Walker (***) Date 2007-07-08 19:46
So he agreed to play the whole 6 games.  I stand by my words.  When the match is already over and the GM ask for a draw it should be accepted immediately unless Rybka sees a significant advantage.  If you want to see if Rybka will blunder in the ending you can test that yourself.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-10 13:31
Well, we definitely played for the maximum score - of that we are guilty as charged :)

Parent - - By Jim Walker (***) Date 2007-07-10 14:17
Hello Vas,
After going for the maximum score and Rbyka having a performance rating of around 2850 are you satisfied with her performance?
Parent - - By lkaufman (*****) Date 2007-07-11 04:31
Calling it a performance of 2850 ignores the handicaps. You can argue about the exact rating equivalent of the handicaps, but that equivalent has to be added to the raw performance to get a true "performance" for the event. My estimate for the handicaps was 180, giving a performance of 3000; I could be off a bit either way, but surely not by as much as fifty points, so at the very worst Rybka performed at 2950.
Parent - - By Jim Walker (***) Date 2007-07-11 16:55
Hello Larry,
Actually it was around 2820 but I was not ignoring the handicap I was just expressing the performance under the tournament conditions.  I realize there were handicaps involved.  I just didn't try to put a value on them.  I was more impressed with Ehlvest than disappointed with Rybka.  I predicted 4.5-1.5 also but the fighting spirit of Ehlvest still surprised me.
Parent - By Vasik Rajlich (Silver) Date 2007-07-12 08:11
I kind of agree with this. Rybka stood worse well into the middlegame in three of the six games, which is an unusually high rate for her. Maybe it was partly due to the opening book.

Anyway, it's hard to make really watertight conclusions. The 4.5-1.5 score was exactly the one I predicted in our poll.

Parent - - By Hetman (*****) Date 2007-07-12 16:33
what was the point to play for maximal score ?
I hope not to made afraid the next human :-).

Parent - - By Vasik Rajlich (Silver) Date 2007-07-14 07:30
Rybka is missing a 'take-it-easy' parameter :)

Seriously, we want good fighting games. Otherwise, why play the last two games at all?

Parent - By Hetman (*****) Date 2007-07-26 17:07
I am not playing in thouse matches, fortunately :-).
Parent - - By Uri Blass (*****) Date 2007-07-08 16:50
I disagree about it.

I think that the game is also for the spectators and the spectators may complain about accepting a draw when the draw is not obvious and the program evaluates herself as better.

Parent - By Fulcrum2000 (****) Date 2007-07-08 17:55
The draw was very obvious, and most spectators watching the live game proposed the draw to Larry...
Parent - - By lkaufman (*****) Date 2007-07-08 17:57
That was exactly my thinking.
Parent - - By Hetman (*****) Date 2007-07-08 18:20
Has been that match for spectators or for marketing the programm ? I have thougt that for marketing.
Parent - - By lkaufman (*****) Date 2007-07-08 18:31
Most of the sponsorship was from fans, so mostly it was for spectators.
Parent - - By Hetman (*****) Date 2007-07-08 19:12
It reminds me old Roman Empire  and gladiator fights in Colosseum ;-). There were rule of thumb :-). Up meant leave the lost live, down kill him. Were the thumbs up or down ? I have red that draw was suggested ?
Parent - - By lkaufman (*****) Date 2007-07-08 19:57
Some said so, but not all. If there is significant doubt, why not play on? Ehlvest was not offended.
Parent - By Fulcrum2000 (****) Date 2007-07-08 20:03
Almost all people watching (and actually chatting) agreed that a draw was ok. They changed there mind after Vas said he would have a closer look at the current position. Then all of a sudden some said, ok lets wait a couple of moves...  I think it's not really fair towards Jaan. If he says it's a draw and offers one, take it. He showed the fans some exiting chess and had at that point already lost the match. Give the man a break...
Parent - - By Razor (****) Date 2007-07-09 04:10
That's incorrect; there were no thumbs down in the gladitorial ring!
Parent - By Hetman (*****) Date 2007-07-09 16:39
That I wanted to read, but here were thumbs down.
Parent - By Jim Walker (***) Date 2007-07-08 19:51 Edited 2007-07-08 19:53
Uri how much is better?  Anything less tha +/- 0.25 is not significant in the endgame.  What is the purpose to continue?  To see if Rybka will blunder?  They can do that in home analysis.  To see if the GM will blunder?  To what end?  The match is already over.  If the match is still in doubt I can see a fight to the end.  That was not the case.
After saying all that,  I must say I enjoyed the games and appreciate all who made it possible.  I also have more respect for GM Ehlvest after watching him fight in this match.
Up Topic Rybka Support & Discussion / Rybka Discussion / GM Ehlvest - Rybka. Game 5 (draw)

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill