Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Joel Benjamin-Rybka; A Full Pawn!
1 2 Previous Next  
Poll Who are you routing for in the Benjamin vs. Rybka match? (Closed)
I would like to see Benjamin win. 16 32%
I would like to see Rybka win. 18 36%
I'm neutral. 16 32%
- - By lkaufman (*****) Date 2007-07-13 23:30 Edited 2007-07-13 23:51
     In response to the clear preference of voters in a poll on this topic, we have decided that the upcoming pawn handicap match with GM Joel Benjamin shall feature alternating colors, as well as a slower time control and larger increment! By alternating colors, we have removed the objection that this is not a full "pawn handicap" match. The time limit was chosen by Benjamin, who felt that it was about optimum for him given the two round a day format. Perhaps we are being too generous this time, Joel is confident that he won't lose the match and we're prepared for a possible match loss, but will do our best to win! Here are the terms:
1. Time Limit: One hour per side, 30" increment.
2. Eight games, two per day Aug. 6 thru Aug. 9.
3. Schedule: 10 a.m. and 4 p.m. each day (Eastern time). Match location: my home in Potomac, Maryland.
4. In each game, Rybka removes a different one of the eight pawns at the start.
5. Colors: In the four games in which Rybka removes a rook's pawn or a bishop's pawn, Rybka plays Black. In the other four games, Rybka plays White. The first game will be with "d" or "e" pawn removed (chosen by lot). If the first game is with "d" removed, the order shall be d,c,b,a, e,f g,h; if "e" is first, then the order is e,f,g,h,d,c,b,a, thus insuring alternation of colors.
6. This is a "friendly" match. This means that neither side will attempt to gain points or advantages by non-chess means. Rybka will stop Joel's clock on request for short restroom/beverage breaks, and Rybka may be rebooted if necessary due to "crash" or "freezing", with the clock restored to its time at the moment of the incident as closely as possible. Any operator errors may be corrected when discovered, with the times restored as closely as practical.
7. Joel's time is governed by his physical clock, whereas Rybka's time is governed by her displayed computer clock, so operator delay will not penalyze either side. It is understood that the operator will be broadcasting the games, and this may cause some delays.
8. Financial terms: $200 per point scored by Joel, $200 bonus for a drawn match, $600 bonus for a won match, and $150 to offset train fare.
9. The operator has the right to offer and/or accept draws, either with or without consultation with others, but will be primarily governed by Rybka's evaluation.
10. Rybka will run on a "Quadcore" computer, and the version will be a non-public one, which may contain a small "opening book" for pawn handicap.

     Joel Benjamin has been three times US Champion, six times World Open champion or co-champion, three times U.S. Open Champion or co-champion, and has played on the U.S. Olympic Team in six Olympiads and two World Team Championships, and also in one Interzonal. More to the point, he was the chess consultant for Deep Blue, having earned the job by defeating it 2-0 in a formal match in 1995. He is writing a book about his chess career called "American Grandmaster" due to be published by "Everyman" this December.

     Discussion: Although Joel is rated about fifty points below Ehlvest, his experience with Deep Blue may compensate, and he is likely to prepare much more than Ehlvest did, since Ehlvest plays tournaments virtually non-stop. By having Rybka play Black in the two edge pawn games, we are insuring that Joel will be at a substantial advantage in every game, and in the case of the f7 handicap, we will be giving the traditional "pawn and move" handicap, known to be very difficult for Black as normal development is difficult due to the weakened king's position. I would say that this match will be the most difficult challenge Rybka has yet to undertake.


Parent - By turbojuice1122 (Gold) Date 2007-07-14 00:46
This should be interesting, though of course we have the situation of these not being "real chess positions" due to the nature of the handicap, making the games a bit less of a pleasure to watch.  Nonetheless, it is nice that Joel has accepted this challenge, especially doing so with such a relatively small prize fund, though again thanks to those who have provided to the fund.
Parent - By Razor (****) Date 2007-07-14 07:24
Whilst I am sure this will be received well by some, I suspect many will still prefer to see a 'normal chess' contest played under the controls more common to the human opposition.  The result of this match adds very little to understanding where Rybka 2.3.2a+ stands in terms of playing chess at tournament level against human (GM) opposition.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-14 07:35
I just want to add one comment here - this is the first time that the odds on paper are extremely close. Rybka is in serious danger of losing this one. We hope that it will be an exciting event from the sporting point of view.

Parent - - By Carl Bicknell (*****) Date 2007-07-14 20:23
Why is this match so much closer than the first one with pawn odds? Is it the time control?
Parent - - By Gaмßito (****) Date 2007-07-14 20:58
In my opinion Joel is simply a stronger player than Ehlvest against computers. He was an expert in computer play some years ago. I strongly believe that he will not play for the audience as Ehlvest did. He will play to win this.   
I agree with Vas that the program is in more danger against Joel than against Ehlvest.  
He is enough dangerous, and even more with the handicap pawn advantage. Note that, if he play and find the right strategies against Rybka in this upcoming match, he could give the lunge. :-) Good, maybe so much in this way I don't believe it.. But it could be a very closed match. 
Parent - - By OleM (**) Date 2007-07-22 01:27
He may have been an expert a few years ago, but computers, and especially Rybka, has improved so much since then. I don't think Joel has a chance to win the match. But of course I hope he puts up a real fight! Those people who say they don't like the handicaps can't be true chessaholics in my view....a true chessaholic loves chess no matter what! I certainly will watch with great interest.
Parent - By Gaмßito (****) Date 2007-07-22 04:03

Of course Rybka is much better than any program of the past, by a big difference. You are clearly right there.  
Joel also was a very strong player against programs, but I think that, even in this way, if he plays correctly finding good plans and strong openings, he could have a good score against Rybka.  
Note that Ehvest did not learn from their first experience with Rybka. He played too aggressive in the second match and an experienced player in games against machines  know very well that these things are generally paid very expensive.  
One thing is of course to play for the audience and other to play in anticomputer style. If you play in a brave style, even knowing well that your opponent is tremendously dangerous and it does not forgive a single error, it is not very wise to play in this way.  
The computer can kill us if we play the most minimum error and we humans are very exposed to play mistakes. It is much better to play with good plans and find strong openings where the machine do not knows how to develop their pieces correctly.  
Still Rybka, has not faced an opponent of that caliber, but there are many and very strong players. The problem here is that a majority are too concerned and worried to find only the best GM's based only in higher ratings, but higher ratings do not means always a better play against machines. We need to make the experiment against true experienced players.  
Parent - - By Dadi Jonsson (Silver) Date 2007-07-14 21:52
Because Rybka has black in four of the games against Joel Benjamin.
Parent - By Vasik Rajlich (Silver) Date 2007-07-16 11:25
The colors and time controls are now neutral. (In the first match vs Ehlvest they favored Rybka.)

Larry's description of match ("Full Pawn") is quite accurate. The only remaining quibble is that two of the 8 games are with rook pawn handicaps. Some would argue that a rook pawn is not really a pawn.

Parent - By grolich (***) Date 2007-07-15 17:57
Most importantly, in the previous match, Rybka was white in all games, so the pawn down handicap was less telling.
In this match, in half the games Rybka is both black and with a pawn minus. That's a huge difference.

Also, Joel has a lot more experience with computer chess than Jaan Ehlvest.

And finally, of course, there is the issue of the modified time controls.

But the most important factor is having black with a pawn down...

This should be interesting.
Parent - - By Razor (****) Date 2007-07-15 14:46
Even without knowing what hardware you are running on I would say the GM will struggle to draw this event, let alone win it.  Sure enough, Benjamin is a decent player, but he is going to have to play two games a day (instead of one) and the time he has per move (on average) is around 25% of what he would normally have in a human v human contest.

Looking at many of the comments in this thread, it would seem to me that most answers point towards 121 chess tournament events, such as, Rybka v Fritz, Rybka v Hiarcs. Rybka v Shredder, etc.  For what it is worth I would prefer to enter Rybka in to human tournament where Rybka's result doesn't count in the tournament.  The tournament organisers would have less trouble agreeing this with those playing if they knew Rybka would not be taking any prize money.  What you and your customer base would gain is analysis on how far Rybka has come against GM opposition when playing usung chess rules governed by the organisers.  Again I would say that Rybka 2.3.2a+ should play on a middle of the range laptop, i.e., 1Gb Ram, dual-core processor technology, etc (about £750).  This makes the result so much more meaningful to the masses.
Parent - - By lkaufman (*****) Date 2007-07-16 03:25
     Your comment about 25% of normal time is just poor arithmetic. For a sixty move game, Joel with have 90' total, or 1.5 minutes per move. Standard tournament chess is considered to be 40 moves in two hours, or 3 minutes per move. This is 50%, not 25%. As for two rounds per day, almost all of the tournaments Joel plays in are two rounds per day, at much slower time limits than this one, so this is hardly a strenous schedule for him.
     As for having Rybka play in a GM tournament, it would have to be a 2700+ field for it to be of any interest to me, and it's not very likely that Rybka will be invited to such an event, though of course it's possible. Everyone would just play for draws, the question would be who would draw and who would lose.
Parent - - By Razor (****) Date 2007-07-16 04:45
No, not poor arithmetic, just a misunderstanding on how long Joel had for the game.  Nonetheless, you do confirm the difference between the match you have set up with the GM and what we both understand to be 'normal' tournament chess is significantly different and of no interest to me nor the majority of people responding to this subject on this website.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-16 11:26
Rybka did play in one official tournament, the "Torre Entel 2006".

Parent - By Vinvin (***) Date 2007-07-17 21:13
Results are here :

1     Rybka       =    1    =    1    1    1    1    1    1    1    9.0/10

2460(cat IX)
Parent - By tasdourian (**) Date 2007-07-14 22:33
Despite what some here have been saying about it not being a "real" match, I think they are missing the point.  Because the "embarassment factor" has been too potentially high for the human, we have never been able to see if a top computer program really could give pawn odds to a grandmaster in a serious match.  I might also add that it shows how Vas and Larry are clearly interested in the actual question of how Rybka can do, as it would not be really all that good publicity if Rybka lost-- the Chessbases of the world could conceivably try to use it as "proof" that Rybka isn't all that much better, as silly as that notion is.  I really like the idea that there is a real sporting challenge here!
Parent - - By Linus (***) Date 2007-07-15 08:04
So the race human vs car goes on. This time the driver of the car must use the first gear only. Yet, the car will still win. What next?

I am disappointed that this farce goes on. I thought the second Ehlvest match proved once and for all that humans (even GMs) are not a match for todays chess programs any more.  What is the use of trying to find more clever ways to handicap the computer so the human has a chance at all? What will come next? Knight odds? Rook Odds? All this energy would be better invested in trying to set up a match Rybka vs Fritz, Junior, Hiarcs, Shredder or whichever other program claims to be the best.

Just a thought: If Chessbase does not cooperate why not set up such a match without their consent. Are there any legal ramifications? Would it be allowed e.g. for Larry Kaufman to set up a match Rybka vs Deep Fritz at his place, maybe with a well respected independent referee. To avoid questions of unfair preparations such a match would have to use the standard versions of both programs as they come out of the package (or as downloaded, in Rybka's case). No tweaks, no special books, identical hardware and time controls.

Now that would be a match I would like to see.
Parent - - By gala.martin (**) Date 2007-07-15 08:57
"Now that would be a match I would like to see."

It is easy. Just buy rybka and fritz, and run a match on your pc. nothing more.
Parent - By Linus (***) Date 2007-07-15 10:06
You are missing the point. I mean a public match with live broadcast. With the goal to prove whih is the strongest engine. Since Chessbase obviously don't dare to agree to a formal challenge, make it an informal one. An independent referee should give the match the necessary credibility.
Parent - By nine castles (**) Date 2007-07-15 10:10
There is no point in running such a match since anyone can easily to go to CCRL or whatever and download hundreds of games of Rybka vs Junior.
Parent - By lkaufman (*****) Date 2007-07-16 03:32
No other program claims to be the best, at least on a computer vs. computer basis. Some claim that they might be better against humans; these matches help to refute such claims, though that is not the main reason for them. One point of the second Ehlvest match was precisely to prove that only with explicit handicaps like a pawn can grandmasters hope to win a match from Rybka. If we lose this match, we will then have a goal -- to improve Rybka to the point where she can win a similar match against Joel or a comparable opponent (GM Emanuel Berg is probably next up).
Parent - By Vasik Rajlich (Silver) Date 2007-07-16 11:27
A match against an unwilling opponent is IMHO not a good idea. I certainly wouldn't be too impressed if someone staged a match vs Rybka and our team wasn't around to make sure things were done properly.

Parent - - By plicocf (***) Date 2007-07-15 08:13
IM LK sorry, I have no interest in this match.
Despite this, I want good luck for Rybka team.

Paulo Soares
Parent - - By Chandan (*) Date 2007-07-15 16:39
Hi Vas and LK ,

To be very frank this match does not interests me or most other chess lovers. For starters, the pawn odd is an unreal situation. Also, I feel that Rybka should now take on someone of its own size. This match will not prove anything.

Parent - - By SR (****) Date 2007-07-15 18:32
You are entitled to your view, but I find it quite interesting to see how grandmasters are doing with this kind of handicap. It also teach us something about what kind of resources are available in positions most of us would have considered hopeless (against any top player) a few years ago.
Parent - - By Chandan (*) Date 2007-07-15 19:01

I am more interested to know how the best Human chess player fares against the best chess S/W at standard time control.

Parent - By Jim Walker (***) Date 2007-07-16 12:32
Chandan if you are truely interested in how the best human will do vs the best chess S/W at standard time controls I will tell you.  The human will lose.  No question about it.  Why bother? 
For the rest of you who say nobody is interested, you are wrong again.  Just go to the web site when the match is on and see how many show up.  You do not speak for the silent majority.  Everyone has his/her own idea of what is a fair match or what is most desirable.  At this point in time I think Larry and Vas are doing their best to get both.  Multiple matches under different conditions to find where the humans are capable of competing is about all that can be done at this point.  Face it, humans have been left behind.  Yes even the top GMs don't want to be embarrased unless they get big money for it.  Kasparov got his big check and still cried like a baby when he lost complaining of the conspiracy theory.
Anyone who wants Anand to play Rybka should feel free to contact Anand themselves and offer whatever prize money they have to offer and see what happens.  I feel sure if you put up the money Vas and Larry will make Rybka available.
:)  :)  :)
Parent - By Juergen Faas (**) Date 2007-07-15 19:17

>>> To be very frank this match does not interests me or most other chess lovers <<<

To be very frank, I´m glad *I* am not "most other chess lovers". :)

Parent - - By lkaufman (*****) Date 2007-07-16 03:33
"Take on someone of its own size"?? Who else is rated over 3100??
Parent - - By Chandan (*) Date 2007-07-16 04:40
Hi LK,

Even though there is no one else with 3100 rating, I do not think that Rybka is actually 300 elo stronger against a 2800 human like Anand, kramanik or Kasporav though i am very sure that it is stronger than them.
The point remains that 3100 computer against 2600 human is not interesting particularly with pawn handicap as the game will be an unreal one. Also Rybka has proved its superiority by beating Elvest in pawn handicap earlier. I feel that its energies are best conserved for organizing a match against the best human player today. Though I know it would be difficult, I do not think any other match will do justice to Rybka's strength and name.

Parent - By lkaufman (*****) Date 2007-07-16 05:47
Rybka does not lose energy playing a match! Seriously, we would love to play a match with Anand, but it's very unlikely to happen, so we will play the most interesting matches we can arrange with the available funding until someone bestows half a million dollars on us to play him.
Parent - By Quapsel (****) Date 2007-07-20 08:20
I'm interested in this Event!
Shurely I would prefer a match between Rybka and Kasparov, Topalow, Anand, Kramnik and so on.

But while this can't be done, I am very interested to see, what Joel and Rybka can do in such a Situation.
Unreal or not.

Parent - - By JhorAVi (***) Date 2007-07-15 22:12
Even if Benjamin wins this match, this event cant be recorded as human win against computer because it's a handicap.  Unless you can organize a match format where the human has better chances without the word HANDICAP... like longer time controls.

Id rather prefer to watch a quality TWO-game-match with longer time-control rather than 8 games lost due to time trouble, blunder & fatigue.

Sadly it's impossible because I almost forgot that the organizers are in Rybkas side therefore humiliating mankind is prefered :)
Parent - By Carl Bicknell (*****) Date 2007-07-15 23:07
The only way out of this deadlock is with a player in the top 5.

Kramnik V Rybka may be expensive, but it would be the ultimate human V Comp showdown.
Parent - - By lkaufman (*****) Date 2007-07-16 03:40
     There is only one way to accomplish what you want. A human grandmaster would play email chess, taking for example one day per move, while Rybka would be set for 40/2. Of course this could not be broadcast live, and the grandmaster would have to be someone of such recognized integrity that we would trust his word not to use any engine. Naturally there could not be a significant monetary prize without creating doubt in people's mind as to possible engine use. Now if anyone knows a grandmaster of high integrity who would like to take on this experiment with only a token prize, please let me know. This might be reasonably fair, if it's a strong GM who actually spends a couple hours per move on the game. Simply changing the time limit to 40/3 for example won't do any good, and anything longer than that just runs into problems of fatigue and adjournments.
Parent - - By Lee Ma Hong (**) Date 2007-07-16 14:48
There is only one way to accomplish what you want. A human grandmaster would play email chess, taking for example one day per move, while Rybka would be set for 40/2.

Hmmm ... Isn't Arno Nickel a GM in Correspondence Chess?

or if you want a FIDE GM, maybe you can try getting in touch with Ulf Andersson. He has played a few email matches afaik.
Parent - - By Uri Blass (*****) Date 2007-07-16 15:21
Arno nickel used computers to help him so he is not relevant here.
Ulf Andersson is not active in correspondence chess and did not score well in his last tournament.

Parent - - By FICGS (**) Date 2007-07-17 11:40
Definitely, the time control is too short...

Parent - - By Sepultura Date 2007-07-19 22:19
I'm a little disappointed. In my humble opinion, this match is not interesting because of the following reasons:

1) It's not classic chess.
2) The opponent is weak and unknown.
3) It's almost the same kind of match played a brief time ago.
4) The final outcome has not significance: if Rybka win the match, nobody will be shock due the restricted
   opponent's strenght. Otherwise, if jugador win the match, everybody knows he would be crushed without odds.
   In adition, he can't proudly say "I beat Rybka", so the victory is not real.

Come on, Rybka's fans want to see her in full strenght mode against a top 20 player...
We know it$ difficult, but not impossible.

Regards :)
Parent - By turbojuice1122 (Gold) Date 2007-07-20 03:22
I agree with (1), but definitely not with (2) and not really with (3).  Joel Benjamin is not at all unknown--he was one of the main grandmasters behind the Deep Blue team, and he is an expert at playing against computers.  Such was not the case with Ehlvest, even though Ehlvest was higher on the FIDE rating list.  Joel Benjamin has made it a science to know how computers play chess and has an incredible amount of experience in this capacity.  As such, I would say that his chances in this match are at least those of what would be for an ordinary player rated 2700 on the FIDE rating list, and possibly more like 2750, depending on the player.  As for (3), we have not seen such a match with "pawn and move" odds, in particular with the f7-pawn missing. 

However, interestingly, Rybka seems to be quite adept at defending the position with the f7-pawn missing.  I have played a number of 120'+60" games on my computer with Rybka as black and strong chess engines as white, and the results are as follows:

Fritz 10: 1-0 (quick win in just over 30 moves)
Hiarcs 11.1: (white lost on time in a better position)
Fruit BetaX1: 1-0 (long win in endgame)
Fruit 2.2.1: 1/2
Rybka 2.2 Winfinder: 1/2
Spike 1.2 Turin: 1-0 (quick win in just over 30 moves)
Naum 2.1: 0-1
Ktulu 8: 1/2
Loop 10.32f: 0-1

I will post these games and more, probably in a separate post, in the future.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-20 09:42
In my view it will be noteworthy if Benjamin can win.

Parent - By Jack-Ireland (**) Date 2007-07-20 19:05
Sure, this will be an interesting match. I am looking forward to it.
Best, Jack
Parent - - By Sincefalastrum (*) Date 2007-07-24 00:15
Hi, just a simple question, how many moves can Rybka plan ahead?, at least generaly speaking.

Parent - - By turbojuice1122 (Gold) Date 2007-07-24 02:56
You will need to be a bit more specific and give a time control for the game of which you speak and the hardware on which you want Rybka to run.
Parent - By Vasik Rajlich (Silver) Date 2007-07-24 07:47
Maybe 30 half-moves or so in a normal middlegame. Obviously, it depends on many things.

Parent - - By tturgut (**) Date 2007-07-29 01:01
Rybka has no chance in a correspondence chess match. I would expect it to score % 30 or less against me.
Parent - - By Vasik Rajlich (Silver) Date 2007-07-30 10:17
Welcome to our forum Tansel :)

Your figure sounds about right - that's 150 Elo or so.

Parent - - By tturgut (**) Date 2007-08-01 14:37
Thanks Vas. Hi Uri.

1) Even though Rybka is by far the strongest program, it still has major weaknesses.

2) In the 17th Olympics, I am scoring 9/10 which will be the highest individual score of the Olympics. I had no difficulty beating Rybka aided players in my last 2 tournaments ((Finally, Rybka will be able to get half a point from me in the Candidates tournament, but so far i have won every game- both with white and black)

3) As time increases, human+computer combination become much stronger. The trick is to guess (or feel or know) which program will do best in different positions. So, even though i analyze every move with Rybka, often I chose moves that are not Rybka's first recommendation. (Rybka gives the absolute best move may % 60-70 of times, but that is not enough). Also in certain type of endgames, Rybka is still not the best engine. Also Rybka has a tendency to underestimate the attack on his own king sometimes.

4) For Uri's question:
     a) Time: I would say iccf time controls (10 moves/ 60 days).
     b) hardware:  In my opinion hardware is not as important. A 32 processor machine may give 5-6 plies more than a dual processor machine, but I can jump my analysis a couple times seing 4-5 moves deeper than even that computer. (See how Rybka team won the Final round of the last FreeStyle Tournament even in that time control. (even other players did better than the 32 processor computer)
     c) Opening books: I don't believe in opening books yet. I think that they are not ready. I don't like copying other people's moves. The quicker one goes to his own way, the better it is. (especially with white). There are so many uneplored sidelines that are very rich and exciting. Somehow this technique is working well for me (9.5/10 with white in last 2 Olympics, and (probably) 5.5/6 in Candidates))

5) By the way, even though GM Benjamin is a very dangerous opponent for computers, I still think that Rybka will win easily. Too much tactics in chess to have control in a short time limit.

Good luck to the Rybka team!
Parent - By turbojuice1122 (Gold) Date 2007-08-01 15:34
I agree with number 5, especially after seeing results on my computer with long time controls and strong programs playing against Rybka with f7-pawn missing.  I'll post these games in a separate thread.
Up Topic Rybka Support & Discussion / Rybka Discussion / Joel Benjamin-Rybka; A Full Pawn!
1 2 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill