1
2
Grandmaster Vadim Milov, currently world number 28 at FIDE 2705, has agreed in principle to play a handicap match in September with Rybka subject to our raising a reasonable prize fund for the match. We provisionally talked about using a variety of handicaps, for example two games of normal chess with White pieces for Milov and perhaps either large time odds or single-core restriction, two games at pawn and move handicap, two games at Exchange handicap (a1 rook and b8 knight off), and two games of normal chess with Milov allowed to use some chess program (details to be decided later) on a laptop. He is pretty much willing to try any reasonable handicap. Anyone willing to contribute to the prize fund please message me here. I think this match is likely to happen, he wants to play. Meanwhile we're still waiting to hear from a potential sponsor of the Nakamura match, we have the one-day Dzindzi match on Monday, and also a one-day match soon after that with an IM at pawn and two move handicap. Rybka 3 will be getting a good workout, it seems!
![[us]](/mwf/v12/flags/us.png "United States")
Giving a handicap to a 2700 player, is that not wise?
Rybka 3 is expected to be rated over 3200, so some sort of handicap is necessary to even try to validate this rating. As for the specific handicaps listed, I think Rybka is favored on one core with the Black pieces, Milov is favored at pawn and move or the Exchange if it's a slow game, and as for a laptop-assisted game everything depends on what program he is allowed to use. We'll have a little better idea about pawn and move after Monday's match with Dzindzi.
![[de]](/mwf/v12/flags/de.png "Germany")
We'll have a little better idea about pawn and move after Monday's match with Dzindzi.
First I would like to see a rapid game match Roman vs. Vadim. My bets are on Roman! :-)
Please, don´t understand me wrong. I´m not convinced of Vadims playing strength, but maybe I´m totally wrong. Vadim is 36 years old and his Elo performance was an up and down and his normal Elo performance was about 2600.
First I would like to see a rapid game match Roman vs. Vadim. My bets are on Roman! :-)
Please, don´t understand me wrong. I´m not convinced of Vadims playing strength, but maybe I´m totally wrong. Vadim is 36 years old and his Elo performance was an up and down and his normal Elo performance was about 2600.
Milov is at a lifetime high, for whatever that's worth. I know that Roman is much stronger in 5' blitz than his 2548 FIDE rating would predict; I've seen him play on the internet and it's pretty impressive. I'm also confident that in rapid chess he is also much stronger than his rating. But regardless of these considerations, they cannot make up for 157 Elo points, in my opinion.
2600? I don't totally believe in 2705, either. But he is a very strong player and certainly stronger than 2600.
I'd definitely bet on him against Dzindzi.
By the way, Milov is a very 'clean' and technically strong player. He may be a good opponent against a computer. (Unfortunately, I find these handicap games just horrible. It doesn't look like chess to me. The match I found most interesting - by far - was the white + draw odds + three-move book against Benjamin. I would surely prefer these conditions against Milov, although I see that it most likely wouldn't be fair.)
I'd definitely bet on him against Dzindzi.
By the way, Milov is a very 'clean' and technically strong player. He may be a good opponent against a computer. (Unfortunately, I find these handicap games just horrible. It doesn't look like chess to me. The match I found most interesting - by far - was the white + draw odds + three-move book against Benjamin. I would surely prefer these conditions against Milov, although I see that it most likely wouldn't be fair.)
I agree--I guess I'm more of a "purist" in the chess sense--the draw odds and low book match was quite interesting to me, and the match "in principle" with Nakamura also sounds interesting. I get the impression that Rybka 3 will also be the strongest engine in the opening phase of the game.
I feel the same way. I'd rather see normal chess. Why not give Milov move odds and set it so that Rybka has to win by X games?
This is interesting I was thinking of a sort of computer assisted match vs Rybka by some top player.A sort of one sided Advanced chess match.(Advanced Odds?)How about letting the GM use a computer AND give him odds on top of that?I think the only way to level the field is to let the GM use comp assistance,as for the NORMAL game the GM will be destroyed most likely and will lose the other odds set ups too it seems.But YES how about Advanced Odds match,that should be interesting.
In my opinion, a 2700+ player at a standard tournament time limit like 90'+30" is a favorite against Rybka 3 at pawn and move. To talk about giving him these odds plus a computer is just unreasonable, unless you have in mind giving him a Novag Superconstellation or some such machine from the early '80s!
> unless you have in mind giving him a Novag Superconstellation or some such machine from the early '80s!
Yes, for a 2700+ player starting from a winning position, he might need some entertainment during the game. :-)
I don't know Larry based on our preliminary results I think R3 can handle the challenge.R3 is destroying the competition right now with unbelievable results.
This doesn't tell us whether pawn and move handicap is too much or not. We'll have a better idea after Monday, although the games with Roman will be semi-rapid. Steinitz claimed that even God could not give him pawn and move handicap successfully, and perhaps he was right.
wasn't it Steinitz who wanted to give God pawn odds?
I've read that this was a misquote. I don't think he was that crazy. The statement as I gave it might be wrong, but it was a reasonable belief to have at the time. I'll bet that most World Champions in the past would have said the same if asked.
Yes, that's a resonable assumption, I think. But I also think that they would have been wrong.
He was probably wrong actually! I don't think that Steinitz would be rated above 2550 nowadays (maybe a very strong IM or weak GM is what I think his games were worth for). But those with a romantic view of the "Good old days" would probably have a difficulty accepting my claim.
I agree with you. Probably he could calculate variations as well as today's 2700, but a lot of chess knowledge was unknown in his day. Standards have just risen dramatically and steadily over time, which was Kasparov's basic thesis in his "predecessor" books.
![[be]](/mwf/v12/flags/be.png "Belgium")
on a bit of an unrelated .... i was just wondering if Rybka sees the legendary Bxf7 sacrifice of Bobby Fisher ? and how fast?
sorry just curious.... :)
sorry just curious.... :)
You guys are talking about a scenario where Steinitz plays "from the dead" without learning today's theory? Yes, you may be right about 2550, ballpark. However, until someone devises a tool to convert strength of moves of already-played games into ratings--an extremely difficult task, actually--it remains speculative. Steinitz in the 1870s was one of the most dominant players of all time, and the competition, talent-wise, was stronger than one might think. But certainly it seems unlikely his boast was justified.
"But those with a romantic view of the "Good old days" would probably have a difficulty accepting my claim."
The genuinely "romantic" thing about those days was that players had to improvise, opening (after a few moves), middlegame, and endgame. The very best players sometimes invented now-standard strategies on the fly. To my mind, this is far more impressive and interesting than memorization and learning how to handle position-types from others, even though it results in weaker chess.
"But those with a romantic view of the "Good old days" would probably have a difficulty accepting my claim."
The genuinely "romantic" thing about those days was that players had to improvise, opening (after a few moves), middlegame, and endgame. The very best players sometimes invented now-standard strategies on the fly. To my mind, this is far more impressive and interesting than memorization and learning how to handle position-types from others, even though it results in weaker chess.
![[il]](/mwf/v12/flags/il.png "Israel")
What I read about the "romantic" time is that people believed that they had to accept sacrifices unlike later time.
If this is correct then it means that the analysis at that time was analysis of lower quality also because of lack of ability to calculate variations(simply because people did not calculate correctly).
Uri
If this is correct then it means that the analysis at that time was analysis of lower quality also because of lack of ability to calculate variations(simply because people did not calculate correctly).
Uri
Way oversimplified, I think. It's when people COULDN'T calculate that a sacrifice would work that they felt that the gentlemanly thing was to accept it. They knew such longer-term sacrifices could work, but the concept of "compensation" wasn't clear in their minds. I'm sure there are cases where people did not accept them, but they usually did. It was taking the challenge to see where it would lead.
Look at some openings manual from a hundred years ago. You know what you'll find? The majority of opening theory was the King's Gambit. Seriously, the MAJORITY! Did people think it was best for White? More likely, they just thought (correctly) that it's a lot of fun to play!
In other words, chess was treated more like a game back then.
Look at some openings manual from a hundred years ago. You know what you'll find? The majority of opening theory was the King's Gambit. Seriously, the MAJORITY! Did people think it was best for White? More likely, they just thought (correctly) that it's a lot of fun to play!
In other words, chess was treated more like a game back then.
You may be correct in saying that the people who invented new ideas are more "impressive" than those who just copy the ideas of others, but the present discussion is about actual playing strength. Of course Steinitz deserves much more admiration than some randomly selected 2550 player today, but I don't believe he actually would win a match against such a player if he were revived, even with randomly chosen openings like CCRL/CEGT use and even giving him a few months to take opening lessons from a current theoretician.
I wasn't just talking about 2550 players...ALL players of today are copying, even when they are also innovators. Please don't get me wrong...today's top players are just as impressive (and more numerous) than those of yesteryear, playing under the conditions they face. I don't think what they have to DO is as impressive or interesting, is all...no disparagement to the players.
You're right that I got off topic.
A few months of study would change Steinitz considerably. It's impossible to predict where that would leave him. I don't think his chess talents were inferior to any active player's today, but that's another topic. But I agree that without being brought up to date, he couldn't handle God with a pawn and a move.
You're right that I got off topic.
A few months of study would change Steinitz considerably. It's impossible to predict where that would leave him. I don't think his chess talents were inferior to any active player's today, but that's another topic. But I agree that without being brought up to date, he couldn't handle God with a pawn and a move.
![[ee]](/mwf/v12/flags/ee.png "Estonia")
Excuse me, but what's this assumption based on?
Rybka 2.3.1 apparently doesn't agree with you... :) I think he would be somewhere near 2300 rather.
Rybka 2.3.1 apparently doesn't agree with you... :) I think he would be somewhere near 2300 rather.
I didn't want to be that hardcore, but ....... maybe ;)
at what time controls would the match be played ?
Probably I would allow Milov to choose any time control he wanted that is consistent with two games per day. He is quite used to playing in 2 round a day open tournaments. As with all of these matches, the two round a day format is necessary to keep the cost of an eight game match reasonable.
>...or single-core restriction
Would you really consider giving this handicap? In the past you have been pretty insistent that Rybka be played on the strongest hardware available to you.
Regards,
Milton
Milov asked that two of the games be normal chess. I thought that the single-core restriction would still qualify and yet give him at least some chance; I didn't discuss this point with him yet.
I'd like to see normal chess with the handicap being either:
1. time
2. the GM only needing say 1 or 2 points out of 8 for a win
or some combination of these.
1. time
2. the GM only needing say 1 or 2 points out of 8 for a win
or some combination of these.
We're only talking about two games of normal chess; if he is getting paid by the point (or for match victory) he would presumably aim for draws in these two games unless that's against his nature. Time handicap and single core handicap are very similar if the time handicap is large enough and ponder is off; theoretically time makes more sense because we don't waste everyone's time needlessly, but single core might appeal more to him (as it is not an explicit handicap) and to those users with just single (or maybe dual) core machines.
Yes, but even though logically it might be roughly equivalent, single core handicap is about as unappealing a match condition for spectators as is imaginable. I don't think it is much of an assumption to say that people on this forum want to see the strongest possible Rybka on the best available machine playing. It's like when I play people table tennis (I'm reasonably good)-- they would much prefer I play them left handed, or spotting them points, than if I told them "I won't be playing as hard as I can". People want to see Rybka trying as hard as it can, and in fact it is thrilling to see it overcome its handicap in odds matches.
I agree with what you say, but a time handicap (especially with ponder off) has the same effect as playing on 1 or 2 cores instead of 8. Either way the quality of Rybka's play is somewhat reduced; that is why we generally prefer material handicaps and would only do a time or hardware handicap for two of the eight games. I think on this point I would defer to Milov's wishes, if he has a clear preference.
Which is why I think pawn odds, move odds, exchange odds, and combinations thereof are much more attractive than silly time handicaps and core-handicaps. We all want to see high level chess played. Limiting the actual quality of play on the part of the engine is not very pure match-wise imo.
![[ca]](/mwf/v12/flags/ca.png "Canada")
I'd prefer not to to see Rybka "dumbed down" in some way. I'd rather see the human getting larger material or move odds instead. Using core restrictions and time odds seems to me like pulling a few spark plug wires off a car to make for a fairer drag race. Better to give the other car a few car lengths headstart, so you can still scream along at full throttle and pass him.
Peter
Peter
Yes, I agree with you, that's why only two of the eight games would be played this way, to accommodate Milov as well as those Rybka fans who have asked for this sort of handicap.
In opposition to a "hardware-to-the-max" philosophy (which also seems tantalizing to me), playing 1-core-handicaped is fine by me. A match with single-threaded-engine will reduce ELO-strength from about 3200 ELO (octal) to about 3100 ELO (singlecore). Will these 100 ELO points really matter in a match against a 2700 ELO human brain? Improvements on playing characteristic (based on advanced search and evalutation) will be obvious anyway. On the ohter hand, the audience could have the pleasure, to follow the match with live-analysis on equal hardware, using the same engine: in my opinion, that promises a lot of fun and a highl level of identification (with the owned rybka-version as well as with the match istself).
In continuation of this idea, it could be attractive, to let an GM play a match against Rybka 3 mobile on PocketPC. This tiny hardware will be 100-200 times slower than current PC hardware, but nonetheless it would reach something about 2800 ELO points - wich would be fair.
But maybe a GM will feel embarrased to seriously play again an unimpressive gadget, which has the appearence of a toy more than a technical monster. On the other hand, he could also feel an offence to his pride, to be given odds. (Who wants to start 2 seconds earlier in a 100 m competetion, to get passed anyway...)
(o dear, I beg your pardon for my foolish englisch - I didn't practice for years...)
In continuation of this idea, it could be attractive, to let an GM play a match against Rybka 3 mobile on PocketPC. This tiny hardware will be 100-200 times slower than current PC hardware, but nonetheless it would reach something about 2800 ELO points - wich would be fair.
But maybe a GM will feel embarrased to seriously play again an unimpressive gadget, which has the appearence of a toy more than a technical monster. On the other hand, he could also feel an offence to his pride, to be given odds. (Who wants to start 2 seconds earlier in a 100 m competetion, to get passed anyway...)
(o dear, I beg your pardon for my foolish englisch - I didn't practice for years...)
Yes it'll make a difference. It won't play as strong so it won't decimate the human player as efficiently. I don't mind a handicap of material or moves because that simply changes the nature of the game. I don't like a handicap where one player agrees to just "go easy" on their opponent. This seems way outside the realm of the spirit of chess.
Well, I have to admit that my objection to Rybka playing on one core still stands, but your "take it to its logical extreme" idea of playing on a pocket pc DOES appeal to me, but for a different reason. Here, if Rybka can win, it shows that it is a strong enough program that even when hampered dramatically, it can still beat some of the best in the world. It still feels like Rybka is trying its hardest, but it just happens to be trapped in an inferior device. I know it's completely subjective, but one core seems wishy-washy to me. If your are going to handicap a sprint, don't just wear old beat up shoes-- give the opponent a 10m head start.
I agree with your analysis. However there are some people who believe that these 3200 ratings for computers are nonsense, and so playing two games with Black on single core might help to persuade them otherwise if we win them. I would probably prefer to make all eight games have a more explicit handicap, but I agreed to this request to have two games of "normal chess" included. Also, your English is fine!
So the unsuspecting are going to the shambles, appeased, that the axe of the guillotine this time will weight only half of that, what's generally used. Not worrying, until the day is done... How lucky they are!
By the way: thank's for your friendly, foreign-language-related reply - http://www.leo.org helps a lot !
Before this day is done (it's now 2 o'clock a.m. in good old germany / Hamburg) one last question for today: a few hours ago I posted some results of an extremly difficult tactical testsuite, not without reason so called "Nightmare":
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=80221#pid80221
Maybe you or someone of the busy beta testers is willing to compare these results with the current R3!?
Good evening and further inspiring posts and discussions in these computerchess-historical times, whichover you and your splendid team left your mark on!
Axel Caro
By the way: thank's for your friendly, foreign-language-related reply - http://www.leo.org helps a lot !
Before this day is done (it's now 2 o'clock a.m. in good old germany / Hamburg) one last question for today: a few hours ago I posted some results of an extremly difficult tactical testsuite, not without reason so called "Nightmare":
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=80221#pid80221
Maybe you or someone of the busy beta testers is willing to compare these results with the current R3!?
Good evening and further inspiring posts and discussions in these computerchess-historical times, whichover you and your splendid team left your mark on!
Axel Caro
Regarding tactical testsets, I'll leave it to the beta testers to run them, as my responsibility is positional evaluation (plus things like contempt and time management). If you recommend any difficult positional testset, I might run that. But those are tricky, since sometimes the right answer is a matter of opinion.
![[mx]](/mwf/v12/flags/mx.png "Mexico")
If Rybka is never going to reach positions like the ones on these test sets I don't see a point.
That would mean you don't see a point in analysis.
Rybka is not only for positions she reaches in her own games... I think any commercial engine must be intended to find the best move in ANY position, considering that analysis is the most important purpose. A chess player will not analyse Rybka games (or not primarily :-) ), but his own, those of his next opponents, those of the GM he is a fan of, those he studies in training, etc.
So actually, Rybka's "own" positions are the least important IMO.
Rybka is not only for positions she reaches in her own games... I think any commercial engine must be intended to find the best move in ANY position, considering that analysis is the most important purpose. A chess player will not analyse Rybka games (or not primarily :-) ), but his own, those of his next opponents, those of the GM he is a fan of, those he studies in training, etc.
So actually, Rybka's "own" positions are the least important IMO.
In such cases in where the engine won't find the right move in reasonable time, I think advanced analysis methods like IDeA are needed.
Great! :-) I like this post very much.
I don't think Milov playing Rybka with a single core will make MUCH of a difference as R3 trounces 2.3.2a on a Quad vs R3 on a single.Larry didn't you say that a single cored R3 defeats 2.3.2a on a quad at bullet play?If this is true Milov still has to work very hard just to draw R3.
Of course this is all true. If we do this it will just serve to convince those people who don't believe the 3200 level ratings. I suppose he will understand that a draw would be an excellent result under these conditions.
Did he ask for a handicap in those games? I think we should keep the handicaps as clear as possible and let Rybka always play at full strenght, I think if 2 of those games are just normal chess, this would be ok and allow bigger handicaps in the other games.
1
2
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill