Flushed

Oh yeah, right. Danailov was the one spending his time in the toilet instead of the chessboard. I thought it was Kramnik.

Respectfully, I would like to say ... is not time to move on? There is a lot of interesting chess news lately that people would like to talk, but squeezing a topic in which there is nothing else to say besides the same opinions (Kramnik ... Topalov ...), would be nice is there is no chess at all to talk.

On the other hand, Chessbase (which is far for being objective in the whole issue), continue squeezing the topic by showing crappy statistics that compare Kramnik with other champions. People who have nothing else to do get a shot to make Kramnik publicity. They start with the assumption that currently, in every position the best move can be determined (which is absolutely false; not even in computer analysis, because move evaluations can change after certain time of running the machine) and say that the percentage of times that Kramnik does the "best move" is highest than most of other champions, bla bla bla. Of course, they don't make specific (... best moves according to computer ... running it for a time...), because that's the idea of publicity, right? But Mig can be independent of Chessbase once in a while.

PS: Of course, Mig answer would be: "If you don't like it, create your own blog", but I guess he is smart and understand what is the point; I apologize if he take the comment in the wrong way. I will move to the other topics he might open to discussion.

Giannis--

The toilet visits are on Kramnik, just as they were in London 2000 when no one besides the arbiter heard anything about it. Blowing toilet visits into an international incident, and continuing to act the fool with a little toy toilet, is on Danailov.

Sand--

Your concern is undoubtedly addressed in the research paper. Chessbase can hardly be expected to summarize or publish the whole thing.

With the introduction of databases/computers, the expansion of opening theory, and the benefit of experience, it's not surprising that later champions are, on the whole, more accurate than earlier ones. Rated second-most accurate, Kramnik is also the only player whose entire WCC career took place during the computer age.

It's thus somewhat unfair to compare him to anyone else on the list. It is possible to appreciate Kramnik's career accomplishment and remarkable accuracy without arguing that he was greater than his great predecessors. Karpov and Kasparov did it longer. Fischer beat down the Soviet chess empire single-handedly.

Capablanca's ranking is remarkable.

As much as I hate Danailov for the way he acted/commented etc., it's the actions of the Appeals Committee's that should be reprimanded.

If they acted discretely and correctly, maybe none of this would have come out.

Duncan

Before the advent of the strong computer programs, it was exceptionally difficult to figure out what the "best" move happened to be.

Fischer excelled at finding the "best" move. that is one reason he was so incredibly good.

Let us look at the history of chess opening theory. Some positions have been argued for 100 years or more. Of course they are still argued. But if one looks into some of the detail, we find that there was really no way to make a determination in many positions.

Now we have a computer program in everyone home that gives a "best" move. Of course it is not really the ultimate best move. as we see so recently as Rybka coming up with a different way to assess the position, the computer was assigning different positions to the "best" spot.

We tend to forget that the computer simply gives us an evaluation of "best" according to the way we program the computer. If we change the value of the bishop pair, the king safety, the center, etc then the ranking of the positions changes immediately. The trick is to have a very good evaluation function. and people like Kramnik have a very good evaluation function.

Chess programmers actually try to copy the moves of the strongest chess players.

I will end with a new idea. I am trying hard to impliment this idea in my life.

THE ONLY WAY TO HANDLE TROLLS IS TO IGNORE THEM COMPLETELY.

before responding to rediculous statements first read this completely.

http://members.aol.com/intwg/trolls.htm

It really is the only solution.

The link above only gives me 4 seconds of the TV stations promo-intro and then stops. Is the video posted somewhere else? Thank you.

My take on Crafty analysis of World champion is here (I am Gushingbeer):

http://www.chessninja.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=8;t=000238;p=1#000003

To summarize, the Crafty analysis is flawed, because Crafty is not as strong as some (or most) of the world champions. Had it been Rybka, I would trust the analysis more.

But I think it is quite obvious that Kramnik's level of play is much higher than Steinitz's. I think top chess players have improved a lot compared to past world champions. Steinitz was objectively considerably weaker than other world champions, even the early ones, but even if we ignore Steinitz and compare, say, Lasker and Capablanca with Kasparov and Kramnik, than it is no doubt that Kasparov and Kramnik were objectively the stronger players. The argument can be made that compared to their contemporaries, the earlier champs were were dominating than the present ones, but as far as actual playing strength goes, I think it is no contest. Todays players are much better than in the past, pretty much like it is the case with basketball and football. Or take physics - one hundred years ago only 1 or 2 people understood theory of relativity, and these days an average physics major graduate knows more about physics than Einstein knew. That doesn't change the fact that Einstein was ahead of his time, of course. And I think it is kinda like that in chess - a GM rated 2650 today is not as a great a player as Lasker or Capablanca were (dominance wise), but he plays objectively better chess.

Russianbear--

Well said. It comes down to unanswerable questions such as:
--How would Kasparov and Kramnik have performed in an era without computers?
--How would Capablanca and Fischer have performed in an era which had them?

Sorry, I cannot understand Dutch, but, if I can judge from the TV channels of my own country, Mig, please don't feel humiliated : news are often about a lot of nonsense and rubbish, or news given by a very low-level point of view. Just to attract viewers..

I thought the paper on computer analysis of world champions was an excellent piece of work. Thanks to chess base for publicising this. The comments linking this to bias towards kramnik are bizarre - the research showed Capablanca coming out ahead in virtually every category performing strikingly better than later players some of whom played substantially in the computer era. As the paper points out this to some extent reflects differing styles but still I thought it was amazing to see Capablanca's results. The paper is not saying who is the best player ever - its benchmarking the play. Of course there are problems with the approach but the paper wil provide a great basis for future work using other programs adjusting the methodology and so on. n the meantime. Its food for thought and I think that Capablanca's performance on this basis was not predictable and is surprising.

The lovefest between Russsianbear and Greg Koster is as heartwarming as their posts are mirth inducing.
A few gems from Russianbear: "Steinitz was objectively considerably weaker than other world champions".."but even if we ignore Steinitz and compare, say, Lasker and Capablanca with Kasparov and Kramnik, than it is no doubt that Kasparov and Kramnik were objectively the stronger players".."the argument can be made that compared to their contemporaries, the earlier champs were were dominating than the present ones, but as far as actual playing strength goes, I think it is no contest"..
Do tell us, what is your "objective" measure?
A few more: "Or take physics - one hundred years ago only 1 or 2 people understood theory of relativity, and these days an average physics major graduate knows more about physics than Einstein knew. That doesn't change the fact that Einstein was ahead of his time, of course."
How noble of you to conceed this point. Einstein is acknowledged to be one of the foremost intellects in the history of the human race because he thought of special and general relativity FIRST and also because of his contributions to quantum mechanics. That some first year student can understand the concepts of special relativity and solve some example developed 20 years after Einstein's death does NOT mean he is better in physics than Einstein. Anybody can train their minds and learn something somebody else has developed. To strike out in territory nobody has explored before and come up with mindboggling results is the test of genuius, not in applying rules and methodologies others have developed. The fact of the matter is, no definitively objective measure can be developed to compare the achievements of people in different areas. It is a subjective description to say so-and-so was a genius in his field, and usually the more qualified and knowledgable one is in one's field, the more insight one has into what constituted true genius. For example, in physics the opinion of Richard Feynman would be more valid than that of Russianbear. Oh wait, I guess you understand all about quantum mechanics which has moved on since Feynman died, so you must be cleverer than him.. Please stop the drivel, it hurts my head.

different eras not different areas

new Fischer interview, part 1 and 2 about Bobby's problems with his Swiss bankers, part 3 some chess
(Capablanca, Alekhine and the old masters, the problem of openings and the power of memorization, etc.)

http://bobbyfischerpage.tripod.com/index.htm

d, it is all good and well, except that you are wrong. I am not saying that physics graduates or professors of today are smarter than Einstein. Of course Einstein thought of the relativity first and that proves he was a genius. But that is besides the point. Whoever first realised that two times two equals four probably was a brilliant man with IQ of 180. He may have been one of the math greats. But that doesn't change the fact that 5th grade student of today knows more math than that guy. This fact doesn't make a 5th grade schoolboy a greater matematician, but it does mean he knows more math. For all Einsteins intelligence, it is hard to argue that people who are in physics today, even the ones who are not as brilliant, know more about the physical world than Einstein. That is kind of an obvious point, and I don't see how you can dispute that. That doesn't mean I don't think that Einstein was a genious, but it does mean that a lot of people today know physics better than he did - because not only are they familiar with Einstein's work, but also with 80 more years of work in physics, they have more data, etc.

As for what constitutes a better chess player objectively - do you really want to argue that players of today are inferior to early world champions? Will you also claim that Bill Russel would outplay Shaquille O'Neal? It is kinda obvious to me that a lot of players of today are better than Capablanca and Lasker were, for the same reason that a lot of scientists today know more physics than Einstein. They happened to be born later and benefited from greater knowledge (including all the games of Lasker and Capablanca), as well as better training methods, tougher competition, better books, computers, etc.

d, and one more thing:

unlike many other things, chess is objective. We know the moves and we can measure their quality, So an objective examination of the quality of play of the world champions IS possible. The chessbase article we were talking about was an attempt to do just that - examine quality of play of the world champions by using objective criteria. But the approach was flawed, as they used Crafty, that is a relatively weak engine. Crafty is only rated 2618 on this website: http://web.telia.com/~u85924109/ssdf/list.htm

So, had we examined the world champions' games with an engine that is rated 3000 (or 3200, the higher the rating, the better the results), now THAT would tell us something. On that same list, #1 rated engine is Rybka that is rated over 300 points higher than Crafty. Rybka may still be overrated there, but had we used it instead of crafty, we would get much more meaningful results, and I think those results would be quite different from what Crafty gave. But as the experiment was, it was basically a test of which world champion played the most like an engine rated 2618. If anything, the test such as that would identify the WEAKER of the world champions, not the stronger ones. Well, of course the world champions who played more simple positions were likely to get higher scores, as was the case, as well as those world champs that were relatively free of blunders. Anyway, the results were interesting, but the method was flawed. I would wait until some 2900-3000 rated program examines the games, and I have a feeling if it happens, Capablanca won't be on top.

I wonder whether these guys started their analysis at Move 1 or at the first "new move." Our modern dudes might be getting undeserved "credit" for replaying thirty moves from the book.

This debate on the analysis using Crafty is to some degree missing the point. The article on Chessbase, by the way, does make for an interesting read. There was an extensive discussion on the similarities and differences between human players and computer algorithms on this website a few months ago, and some of the points raised at that time are also relevant here.

My interpretation of the results was that Capablanca and Kramnik are the players whose mental processes came closest to paralleling the results of the algorithms used in the program. In chess and in other fields, such as econometrics and applied statistics, one finds human practitioners whose thinking is very similar to the output from computers. This is not necessarily an overall evaluation of their strength as players, or outside of chess, their performance within their professions.

In general, one would expect the ability of humans to think in ways that parallel computer algorithms to correlate with overall ability. But the correlation may not be exact. As one would say in statistics, it is not coterminous with overall ability.

From my point of view, it was interesting that the computer selected players who were noted for the clarity and logic of their play. These are properties of most computer algorithms. It is not surprising that the computer -- or more specifically, the Crafty program -- would not select players like Alekhine, whose play was often impenetrable, or Lasker, who showed an intuitive understanding of complexity. Most likely, the computer did not select Kasparov for the same reasons: his play was inherently complex, and often involved calculating long variations.

It is probably somewhat surprising that the computer did not select either Karpov, Botvinnik or Fischer. Botvinnik was known in his time as a scientific player, while Fischer has also emphasized clarity and accuracy in his play.

In sum, the Chessbase article will not settle the debate on the strongest players in history. Rather, it provides interesting evidence on the parallels between human thought and mathematical algorithms.

Fischer (in that interview) believes that the present GMs are objectively stronger but this is so precisely because of the "opening" theory that nowdays analyses the game and teaches the strongest ideas to be played out well beyond the opening phase.

In his view chess has become more and issue of study and memorization rather than one of game and creativity.

The most relevant comparisons are not those of Fischer to Steinitz or Capablanca to Kasparov. The most interesting ones are comparing GMs of similar historical eras to each other. Regardless of your interpretation of a stat such as the 5 percent difference between Capa and Alekhine or the similarity between Kasparov and Karpov it is very informative.

One stat I would find interesting is the percentages of other notable GMs. It has often been argued that, for example, Kasparov and Karpov were shoulders above the rest of their contemporaries or that Spassky triumphed in an era when several GMs were roughly at the same level.

Russianbear, I will try again..(for the last time). The question of who is "better" in physics, can perhaps be solved with this thought experiment. Imagine that you could transport Einstein in time, sit him together with a first year major in physics, where both have the same background, assume high-school in the US, ALs in the UK etc. Because of his amazing intellect, Einstein would soon outstrip our hypothetical undergrad, unless he was one of a tiny fraction of the percentage of people who have populated this planet, such as say Maxwell, Heaviside, Gauss, Newton, Leibniz etc. That is the only sort of comparison that makes sense. In this comparison, Einstein is "better" in Physics than your average joe undergrad. Its something in his genetic makeup, something that allows him to make connections and leaps of intuition in his chosen field of Physics, that makes him "better" in Physics than a million billion trillion others.

Hi guys, I'm happy that the article initiated many interesting and constructive discussions around various forums. I would like to kindly remind you that you could find a lot of answers to your questions in the original article, which is available at: http://www.chessbase.com/news/2006/world_champions2006.pdf .

Evaluation of each game started on the 12th move, without the use of an openings library, of course. This decision was based on the following careful deliberation. Not only today's chess programs poorly evaluate positions in the first phase of a game, but also analysing games from the start would most likely favour more recent champions, due to vast progress made in the theory of chess openings. Starting the analyses on a later move would, on the other hand, discard too much information.

As we stated in the article, we needed an open source program in order to slightly modify it. In particular, instead of time limit, constant fixed search depth was applied on every move. With such an approach we achieved the following:
1) Complex positions, which require processing larger search trees to achieve a more accurate evaluation, automatically get more computation time.
2) The program could be run on different computers without fear of not getting the same evaluations for a given set of positions on each of them.
The latter enabled us to greatly speed up the calculation process by distributing the computation among a network of machines.

It would be interesting of course to see analyses of Rybka or any other of the strongest commercial chess programs, hence our encouragement to other researchers (or actually anyone) that might have access to their source code. However, from the scientific point of view, our main goal was to offer a carefully chosen methodology for using computer chess programs for evaluating the true strength of chess players. It will most probably improve in time and actually it's quite possible that similar approaches will be tried for some other games as well...

Nicely done, Matej!

Yuri Kleyner raises an important point -- the quality of the competition. This clearly affects any comparison based on the outcomes of games. At various times in history, chess was dominated by players who were significantly stronger than the competition -- Morphy and Philidor were early examples, Karpov and Kasparov are more recent cases. At other times -- including the present day -- the situation at the top has been more collegial. There are several players of comparable skill, who are rather evenly matched. Another instance of this situation was 1948-72. After regaining the championship from Smyslov, Botvinnik described himself as "first among equals" -- probably an accurate assessment.

Arguably, this makes titles like "the strongest players of all time" problematic. More realistically, at given points in time, certain players were significantly above their contemporaries, while at other times, the competition has been much closer.

Returning to the chessbase article, computer programs will inevitably prefer the players whose mental processes are closest to the algorithms, i.e., players who favor logic and clarity. For the same reason, however, algorithms will tend to give lower ranking to players whose ideas were more complex, original, and more creative.

In sum, there are several conflicting criteria at work in this debate. Possibly, one might want to use a weighted average of these criteria, for instance quality of moves as evaluated by the computer programs, results against other players, strength of the competition at the time, etc.

Let's say that I have an incredible talent for curling. I am amazing at everything that's involved in the game, etc. Yet I never played curling in my life and don't even know the rules.

Using d's standard I am much better in curling than an average "major league curling" player. Using russianbear's standard, the average "major league curling" player is better in curling than me.

I will let the readers decide which standard more accurately reflects the everyday use of the term "better in".

I'm afraid Chess's Computer Era has inevitably become the Toilet Check Era will assists from cell phones, ear buds and other technological "advances." But it's hard for me to understand why someone would want to resort to computer assistance during a tournament, as has been documented too many times to mention and too many times for the good of our game. That's not like a baseball player bulking up with the help of banned substances, it's more like a minor leaguer sending a better player in as a double thinking no one will notice. Maybe tournament organizers should just let people who want to use computers during the games play in separate sections for their own prize pools? That or establish lifetime bans for the offense?

Dave50,

Needless to say chess algorithms do not reflect likelihood of success. I may choose a flawed move, but it may take you a lot of energy and time to find a refutation. So as result you blunder on next move or lose on time control. You on the other hand chose a simple accurate "best" move. I respond with an equally accurate move and the game goes on.

Now, which ones of us made a better move?

Hi,
I also just get 4 seconds of generic animation from the link above, no matter how I open the link. Is it posted elsewhere?

d--

Russianbear speaks of "ACTUAL playing strength." How well DOES Capablanca 1920 play? How well DOES Kramnik 2006 play? Opening, middle game, endgame. Start a match between them tomorrow and see who wins.

You are talking about "POTENTIAL playing strength." How well COULD Capablanca play if he had a few years to bone up on the modern game. Russianbear doesn't deal with that issue. Don't quarrel with him over a point he never made.

Yuriy--

Nicely put.

DJN - I had the same problem when the link was opened by WinAmp (which registered itself as default).

I saved the .asx file (right-click and Save Target As...) and then opened it in MS Media Player. That worked, although I'm not sure the movie was worth the effort.

Yuriy Kelyner, I am actually in agreement with you. My earlier comments here were rather nuanced, and were intended to indicate that computer analysis is not completely objective, because it ignores factors like the inherent complexity of chess.

For example, as pointed out earlier, the computer would tend to downgrade players like Lasker, who had a deep understanding of psychology, as well as tactical players like Tal, whose combinations were both inspired, and difficult to refute over the board. This does not imply that Lasker or Tal were inferior players. Rather, it is a limitation of the computer algorithms.

Or to put it another way, the extent to which the moves selected by a human player parallel the moves selected by the algorithm is only one measure of the players' ability.

Matej Guid, I read your paper, too. I think you did a good job with what you had. And you had a nice methodology. However, I think it is a bit misleading to say Crafty could tell us who was the best player of all time. I mean, Crafty is only a little over 2600, and it may even be a little overrated. I understand that you needed source code so you could insert the code for calculating the complexity of the position. But still, I think Crafty is a bit too limited to be a reliable judge in something like this. Shredder would be much better and Rybka would be even better, because they both have a much more human approach (especially Rybka) and would understand human moves better.

So I think instead of finiding who made the best moves, it was more like finding who played the most like a computer engine rated 2600 - and that is a VERY different thing. I've seen you cite the (low and good) scores that Crafty assigned to other engines, whose games it examined, but to me it doesn't prove that Crafty is qualified to do the job of checking the quality of play of the world champions. All it tells me is that certain computers programs think more or less alike.

I was wondering, if the thing you needed source code for was to figure out the complexity of the position, maybe one could try this: Use Rybka for analysis, and use Crafty scores that you got for calculating positions' complexity. And then we could multiply the scores the champions got (as per Rybka) by complexity factor (as per Crafty). While Crafty is not reliable enough to judge the actual moves of the world champions, maybe it would be just good enough to be the judge of complexity? What do you think?

Oops, didn't notice this had become a haven for discussing the computer article. Don't tell me you guys are tired of talking about toilets already! Anyway, there's an item on the computer article up now, so move on over.

Indeed it would be interesting to see how would using another (better) engine affect the results. Frederic Friedel of Chessbase.com even suggested us to ask readers to extend the research so that we get a wider base of games. I think it really wouldn't be a bad idea to offer a website with appropriate tools so that the whole chess community would be able to contribute to "the final verdict", even if this approach reflects only one measure of the players' ability.

The solution you propose, Russianbear, unfortunatelly wouldn't be sufficient, since we need more data from the engine. I'll give bellow exact instructions about what would actually be required.

First, the search should be limited to a certain fixed depth d (plus quiescence search). Then, for each position, the program should iteratively search to depths from 2 to d and for each depth the following data needs to be obtained:
- best move (as suggested by computer) and its evaluation,
- second best move and its evaluation,
- evaluation of the move played.
Besides, for each position (from the first move on) the material state of each player should be given.

However, that's not all. Once the analyses are obtained, the appropriate database should be set up and the whole set of scripts or programs should take care of all the required calculations.. That's quite some work, but if anyone has the required time and feels ready to do the job well for the sake of the whole chess community, I'll offer my help. Anyway, if you happen to know the programmers of Rybka, Schredder, Fritz etc. - now it's time to ask for a favour! ;)

From the scientific point of view, our main goal was to offer a carefully chosen methodology for using computer chess programs for evaluating the "true" strength of chess players. The methodology will most probably improve in time and actually it's quite possible that similar approaches will be tried for some other games as well. Note that an analysis that would give us the ultimate answer about who was the strongest player ever in all aspects (including psychology) will never be possible, since we don't have much more information from the past games than the played moves and the results obtained (it would be interesting to know at least the times spent for each move not to mention a myriad of other factors).

Anyway - what is the strongest move in a given position? The one that will win against your opponent? Or maybe the one that would provide the best statistical score in thousands of games? One can easily get lost in such questions. However, in the search for the truth, some methods are more objective than others and even if computers can't give us all the ultimate answers, they can surely answer quite a lot, if used appropriatelly.

Koster it restores my faith in humanity to see you defending Russianbear so staunchly, however you are as always clueless. Loud, but clueless. Its about a profound understanding of the underlying principles, which is what makes somebody bettter.

d--

Russianbear does not need defending, least of all by me. But I sensed that in explaining his straightforward argument to you, he'd run out of patience.

"It's about a profound understanding of the underlying principles, which is what makes somebody better." This view was refuted by Yuriy in his "curling" argument, which you never got around to addressing. I can't say it any better than he did.

"Oh yeah, right. Danailov was the one spending his time in the toilet instead of the chessboard. I thought it was Kramnik.

Posted by: Giannis at October 30, 2006 06:03"

You're so cute when you pretend not to understand.

Yuri K, the third of the musketeers in your little trio btw, has said something so incredibly asinine, that there's no point talking about it. There comes a point when you have to recognise that you should never argue with fools, because they onlty drag you down to their level and beat you wit experience.. would be funny if it weren't so sad :-(

since I'm a sucker for punishment, I'll have one more go. The reason that Yuri K's argument is asinine, juvenile and probably wholly representative of his intellectual ability, is that he has never demonstrated his profound understanding of curling. Einstein demonstrated his profound understanding of Physics by advancing the borders of Physics as it was known then. Hence he has conclusively demonstrated his profound grasp of Physics. Incidentally, Einstein vs a first year Physics major is completely the wrong example to bring up as a parallel. Even if Einstein had no chance to assimilate, his knowledge of Physics would blow the first year major out of the water, simply because it hasnt moved on so much yet, and the Physics major could not possibly accumulate the knowledge of a lifetime over 1 lousy semester. A better example would be Einstein vs a modern physicist, somebody who has demonstrated considerable achievements in the field. However this is beside the point.

Warning to all posters on issues requiring at least a shotglass of thought and reflection: if you are in disagreement with "d", you are a "musketeer" and a "fool" who produces "mirth-inducing drivel". Not that there's anything wrong with d's argument, but if he honestly believed in it, would he be so quick to call out Koster and Yuriy K and others as "asinine" and "clueless"? Or perhaps he is disagreeing only because he dislikes these guys and hates Kramnik? That's okay too, but his contemptuous and abusive posts stain a very interesting argument.

It's nice to know that with my asinine juvenile and limited intellectual ability I still manage to understand the idea of hypothetical scenario, one in which it is unnecessary to prove a postulate, because it is part of the scenario itself. Now, if you will excuse me, I am going to go demonstrate my curliing skills.

Matej Guid, I replied to you in http://www.chessninja.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=8;t=000238;p=1#000018

I think it is easier to have this conversation in a forum instead of a blog entry comment. Please let me know what you think. Thanks.

Ah clubfoot, I was missing you my friend :-) I have come to enjoy your posts, but this is a poor effort.
Yuri K, unfortunately you dont understand. Curling is a good idea, stay away from Chess!