Mig 
Greengard's ChessNinja.com

Linares 07 at the Half

| Permalink | 80 comments

I'm still alive, but check back tomorrow. My left ear has decided it doesn't like me anymore and has invited a horrible ringing and scratching in to share my headspace. If van Gogh had this, now I understand. I won't bore you further, at least not with my health.

Round seven in Linares, if you still remember Linares, certainly shouldn't have bored you. Not only did we get three decisive games for the first time but they jumbled the standings and included one of the messiest top-level games you'll ever see. Anand beat Leko to join Carlsen in first on +2 when the Norwegian held Svidler with black. Ivanchuk swapped places with Aronian by beating him with white. Chucky is in clear third on +1. Aronian and Svidler are on 50%, Peter the Swede having drawn all seven of his games in the first half. Topalov's phenomenally wild and sloppy win over Morozevich moved him up to -1 to join Leko. Moro dropped to -3.

If you want exciting, tactical chess you have to be prepared to accept a few blunders along the way, but Topalov-Morozevich took this proposition to the extreme. Morozevich's fabulously provocative play with his trademark ..g5 in the French failed to discombobulate Topalov, who steadily fortified his position on the kingside. After 19.fxe5 White was solid and it was clear Black was going to have to sac something in the center or eventually be overrun due to his bassackwards development. 19..Nxd4 was playable but doesn't confuse the issue sufficiently. Moro went for a knight sac on e5 that put Topalov under tremendous pressure, although it's important to remember that Black also has to play very accurately in the attack to have a chance.

Of course this is just the sort of insane tactical position computers excel at and humans often fumble, so I got into it with the roving packs of afritzionados on the ICC. Watching patzers, and I use that term with affection, instantly saying that a super-GM has blundered, or is "totally lost" drives me insane. I promised to stop boring you so I'll skip my well-practiced rant on chess being a human game. (More immediately below in the comments, however.) That a computer says +1.72 for white after 23.Qxd4 in a boggling position with both kings in danger is trivially interesting and the lines can be amusing -- even revealing on occasion -- afterwards. They are essential for serious analysis, of course. But if you have to find 23..Bd7! 24.Nb6! Bc6+ 25.Kh2 Rd8 26.Nc4!! to call Black lost (and he's not), you really need to turn off your engine and enjoy the game.

That said, it didn't take Fritz to see that Topalov was soon getting the better of things. His king had a shelter behind the black f-pawn and the black queenside was still undeveloped. But quite uncharacteristically, Topalov pardoned Black again and again. Part of this was because he tried to play quickly in Morozevich's time trouble, but overall he was just very inaccurate. Moro climbed back into the game with his own threats and it looked like he might have a miracle draw in the making. Even as late as 35..Kc7 (instead of one move later) Black is battling despite playing with a rook on a8 and a bishop on c7. Crazy. Topalov could have ended things immediately with 38.Qg7 but Black got yet another life. Black could have met the worst of the threats by covering g5 with 40..Qf4 (or 40..Qe3). After he missed that he got no more chances. Ironic that the killer turned out to be the g-pawn-turned-e-pawn. White still needed precision against Moro's sharp defense. Hard to call such a blunder-filled game a great one, but it was lots of fun. Twenty years ago it would have taken months to analyze this one, if not years. Now the engines cut through the tactics in minutes. No wonder some spectators don't understand how difficult this game is.

As he did against Morozevich, Anand once again slithered out of an inferior position to win. Anand got a miserable opening with black and Leko could have put him away with a spectacular combination pointed out by Jon Speelman. It's long, but quite forcing. 25.Bxc6+! Rexc6 26.Bd8! Qf8 27.Qe5+ Be6 28.Qxb5! Bg4 29.Bf6! Bxd1 30.Rxd1 Bb6 31.Qe5+ Re6 32.Qd5 Rd6 33.Qxd6 Qxd6 34.Rxd6 with an easy endgame win. After that reprieve Anand defused Peter Leko's pretty 23.e6 blow and was better long before finishing things off with a perfunctory exchange sac. I dunno how he does it, but let's hope he keeps doing it.

Ivanchuk was trying to find a way to pressure Aronian in the endgame when the Armenian went to sleep and allowed his rook to be trapped. He had to give up a pawn to save it and after that Ivanchuk made quick work of it. Carlsen played my dear Accelerated Dragon, even with the unusual 8..e6, which I picked up from Hugo Spangenberg in Argentine semi-lento tournaments. Malakhov played it against Carlsen last year and Carlsen must have liked what he saw. It looked like Svidler was going to make him suffer for a long time but he hallucinated with his 17.Nc2 move and Black found the ugly but effective 18..f6! and held on.

Considering that Topalov is actually doing better at the halfway point this year than last year the only real shocker on the crosstable is Carlsen at +2, a truly fabulous result no matter what happens in Spain. The four rest days can't but help the teenager. Everyone has a loss except for Svidler and if things continue this way +4 will be clear first and +3 a shared first, as expected. Things kick off in Linares on Friday, March 2. The games start at 15:30 local time, 9:30am EST.

The local Morelia paper has had some interesting color items (on Henrik Carlsen and Aruna Anand in particular) I'll get to later. Their editorial page has had an ongoing debate about whether or not the tournament is doing anything good for the city or even for chess in the city. Some of the editorials were critical of the funding, much of which is apparently undisclosed. Another said that Morelian chess hasn't shown any progress over last year. But many city and state officials came to the final day and one said he hoped to have the tournament back in Morelia next year.

80 Comments

Anyone have any suggestions about how draws should be handled?

Yer a laff riot. Now quit it!

Now you won't hear anything in stereo for a while, you might try the Stereophonics ;-) Get well soon.

Indeed, how to use computers in general (for preparation, during live commentary, for writing commentary after a game) is a far more interesting subject than cheating and quick draws and all. (Btw John Nunn covers the subject in his very recently published second edition of Secrets of Practical Chess. Recommended.)

For example: on Chessbase Mikhail Marin focuses on explaining games with words, but doesn't seem to use a computer. He misses many important moves in Topa-Moro, some of them mentioned by you, Mig, and some mentioned by Sergei Shipov. Can we blame Marin for not mentioning those moves? Can we blame Shipov for mentioning lots of computer moves?

I think it depends who the audience is, peter. Surely, people who really want to improve their chess will realize that the added value of computer-analysed variations will always be less than explanations in ideas and strategies. They won't be the ones calling Topalov lost when he has missed a +1.47 computer move. Fortunately ;-)

As I understand it the analyses by Marin are just short notes for free. He gives a deeper analyse later in chessbase magazin.
I myself find the computer very helpful in opening preparation.

I don't like the term "blame" either way. If you want to present the objective truth of a tactical game and you don't have a week to analyze it, using an engine and presenting those lines is essential. Marin is in a hurry and instead of having the key moments of the game dictated to him by a machine, he uses his knowledge and strength to pick out the crucial moments as understood by a Grandmaster.

This is a key point. When any dude with an engine goes over Topa-Moro with Fritz, he'll find blunders and let the objective opportunities to change the evaluation define the key moments. That, as opposed to a strong player understanding the actual turning points, the committal moves, the choices between, say, continuing with risky play or bailing into an endgame, etc. This is how an actual chess game is played, not by when +1.1 rises to +2 or drops to -1. It's about decision-making and evaluation and practical chances and competition, not just crunching numbers. It's not outrageous to discard those things and analyze dispassionately and objectively, move for move, but you still can't ignore the element of practical chances, something a computer doesn't understand at all.

Lastly, the idea that engines are the end-all god of chess strength is far from the truth. They frequently misevaluate positions of all kinds, from the opening through to the endgame. Material imbalances, transitions to the endgame, etc., they can be a mess. Even the sharp tactical positions like Topa-Moro aren't immune. They can cut through the tactics fine, but that doesn't mean they understand what's left after the tactics. You often see them play "brilliant" sequences to reach endgames they misevaluate entirely. That they almost never blunder the way humans do (and that they beat humans) should not confuse the fact that they routinely play positional chess at a 2000 level or worse. Grandmaster analysis is not only more instructive and useful, it's also simply better most of the time.

Live commentary is yet another matter. Having a GM watching an engine while trying to talk intelligently about a position is not just hopeless, it would degrade the quality of the analysis and kill its instructive value. As I often say about my favorite old books, that there are a few tactical mistakes in the analysis easily found by computers almost never lessen the quality of the instruction being presented. Sure we want accuracy as much as possible, but again, most of the time natural and logical (explicable) chess is what counts. The two or three times per game an engine is going to cough up something interesting (obviously more in games like Topa-Moro, but imagine what an engine would do to Kramnik-Anand from Corus) isn't worth the distraction.

Hey Mig - the annoying sound in your ear - that`s called a wife ...

Actually I think having someone around to apply ear drops is a pretty good reason to get married.

Mig,

In Leko - Anand, in the Bxc6+! variation suggested by Jon, instead of 28 ..Bg4, Black could play Bb8 which keeps the position still double-edged. As you said before, the +1.5 engine evaluation appears to mean less here ..

Leko-Anand would have been just a brilliant game, really brilliant. The whole idea of Ra1-c1-a1 just so that while it was on c1 White could play b4 to wall in his bishop on a5 is amazing. The funny thing is that Bxc6 and especially Bd8 are exactly the two main tactical ideas he must have been working with for the past hour or more when the critical moment arose. He just didn’t manage to put them together in the right combination at the right time. Must have been an agonising game especially as Leko must have felt in his waters that if he had to play 25 Rd2 things were bad. I imagine he thought …fxe6 had been forced (as Larry C did on the commentary) and hadn’t been thinking much about …Rxe6. So typical of Anand: he never forgets he might win even when he’s almost lost.

A computer would never find 23...Rxe6 or understand why it was a good move in a game between humans, but still I don’t agree at all with Mig about the commentator having an engine running. Of course having nothing but computer variations is pointless, but equally pointless is throwing about fun tactical variations which are refuted at move one, or, like Jon Speelman did in Svidler-Ivanchuk, groping for ten minutes to find out why taking the rook on a8 in the opening simply loses (as had already been published). It ought to be possible for the GM to do both – imagine he was at the board and ask himself what would be natural, and at the same time run an engine and use the better tactical vision that presents. In fact having the GM whistle through this teeth and say the engine’s line here is good but very hard to see, or alternatively simply impractical, would be a useful corrective to the Fritz kibitzers.

Remember Mamdeyarov-Sokolov from Essent? The pawn ending both GMs cocked up? The computers see the truth of that instantly. Meanwhile on ICC one or two GMs were (it has to be said, rather arrogantly) lecturing the kibitzers, including their fellow GMs some of whom did have it right, on how it should be done, and making complete fools of themselves. Now OK, that’s a laugh, but it wouldn’t make great commentary.

I hope organisers continue to invite Moro to these tourneys even if he finishes on minus big. He's rarely involved in a dull game and his audacious style keeps the crowds hooked.

Mig: discombobulate!? have you tried calling to the 'horrible ringing' in your headspace? i think it will answer to the name of Nigel. Get well soon

It's easy to prove that "engines aren't the end-all god of chess strength," to use Mig's phrase. If they were, then they would always give one of three assessments for any possible position, namely - forced win for white, absolutely equal, or forced win for black. Of course neither the engines nor humans have any chance of assessing most positions with such accuracy, but in the ultimate sense, there is no position that has a true assessment value of .22, or -1.06 or plus-over-minus or unclear.

28..Bg8 29.Bf6 Bd7 30.Ra6! and the pins work for White. Black keeps the piece but is bound after 30..Bd6 31.Rxc6 Rxc6 32.Qb7 (or Qd5). I suppose the c-pawn keeps Black alive, but it looks rather excruciating. Better than 28..Bg4 tho.

The problem with having an engine running live is, as Kasparov put it, you can end up becoming a slave to the machine. In theory it would serve only as a blunder checker to glance at on occasion, but in practice it becomes very hard to analyze when you have a constant supply of computer waffles being served at you. I do regularly toss the positions into Fritz when I think it might have something useful to say. And I have time for that since I'm not the one analyzing steadily. I'll mention the computer corrections or suggestions on occasion if they are interesting or relevant.

Chessbase: "Veselin Topalov played well below par, with a performance of 2691, 92 points less than one would have expected."

Wrong!

2690 is exactly the performance all leading GM say we expect from Topalov. Where he belongs to.

Ellrond: "i don't give a damn about what the majority thinks. i am used to thinking with my own brains." - Mikhail Botvinnik

You think GMs aren't susceptible to "group think"?

One thing we all seem to agree with, is that a game with commentary should have some instructional value. I agree that engine evalution shifts (e.g. from -0.50 to +1.60) are not necessarily the important turning points in the human struggle that took place. But it doesn't mean an annotator should ignore those moments. When the evalution changes dramatically, he should be warned and have a closer look at the position. It's the annotator's decision whether he thinks one of the players did miss a small tactical idea or whether it's just nonsense what the engine is saying.

My point is that nowadays it's not too much to ask an online commentator to check his annotations with an engine. It may only take ten minutes extra and in most cases helps getting closer to the truth.

For live commentary it might be better just to ignore the engines - I agree with that.

Relax Ellrond, there's no conspiracy - just a minor formal mathematical statement that you can safely ignore since you seem to be inert to concepts like probability, distribution, proof or deductive reasoning in general anyway.

>>2690 is exactly the performance all leading GM say we expect from Topalov. Where he belongs to.>>

Actually, I don't think I've heard *any* leading GM say that. I kinda doubt that you have either, Mr. Pantsonfire.

Who can define 'patzer'? (Those folk whom Mig views with affection).

The idea of cheating with a watch (chessbase.com) is not so unrealistic. For example you could implement a thing that vibrates and so gives the coordinates of the chessboard. IMHO they should not be allowed to wear watches or any other electronic devices when they play. NO, im not joking :-)

Anyone who isn't Kasparov.

Both Mig's observations about the flaws of engine analysis, and rdh's corrective, are well worth pondering.

When annotating games, I've had to confront this dilemma of just how much emphasis to assign to Fritz's judgments.

Mind you, the dilemma does NOT disappear in those cases where I think the engine's judgment is objectively correct (i.e., setting aside those other instances where the computer exhibits faulty thinking, such as Mig described). Indeed, it can be even more acute.

In one recent situation, I resolved the dilemma by assigning a (?) (question-mark-in-parens) to an opponent's move that was evidently both strong and so obvious it's hard to imagine anyone would consider any alternative. I followed with the comment: "Natural and thematic, but inferior to the computer move 16.Nc5! If Black then pressed ahead with 16...h4, computer-aided analysis shows that White could snatch and then return material and emerge into an endgame where his passed pawns are just too strong..."

I then gave the 29-ply (14.5 full moves) variation that Fritz came up with, which I checked extensively. (Chess Life, December 2006, page 42)

I find it a little bit disconcerting giving a question mark to a move that not only was chosen by someone hundreds of points stronger than I, but that appeared perfectly strong on its face and would probably pass muster with most GMs who examined the position unaided for a minute or two. Sure, the engine at 30-ply depth found a superior alternative; but how realistic is it to use that finding as the standard when judging the game and writing about it for the edification of a human audience?

I rarely analyze GM games for publication, but ran into similar feelings when annotating the famous brilliancy Tate-Yudasin, in the same article. In that game I had to place negative marks (?, (?), ?!, etc.) after no less than 4 of Yudasin's moves. In that case, my task was made a little easier by the fact that it primarily Tate's wizardry over the board, rather than Fritz's conjuring after the fact, that laid bare the weakness of the super-GM's play (the game took place in 1997, only a few years after Yudasin's 2 Candidates matches).

Noun 1. patzer - a poor chess player

Mig, I really hope your ear sickness is temporary and nothing serious. It must be terrible to hear the ringing all day long... a la Robert Schumann.

Mig wisely says:

"This is a key point. When any dude with an engine goes over Topa-Moro with Fritz, he'll find blunders and let the objective opportunities to change the evaluation define the key moments. That, as opposed to a strong player understanding the actual turning points, the committal moves, the choices between, say, continuing with risky play or bailing into an endgame, etc. This is how an actual chess game is played, not by when +1.1 rises to +2 or drops to -1. It's about decision-making and evaluation and practical chances and competition, not just crunching numbers. It's not outrageous to discard those things and analyze dispassionately and objectively, move for move, but you still can't ignore the element of practical chances, something a computer doesn't understand at all."

Given this, I would say the idea of occasional (as in, not every move, "go/don't go for it") signalling from somebody who has access to an engine is not a very good idea. It is one thing to rely on radio transmission of full analysis where you can reliably receive signal of what computer is seeing. It is another to hope to see a wave or a gesture without knowing what's behind it on the chance of getting a +1, +.5, instead of a move you do understand. Can't imagine anything worse than a computer telling me that a perfectly reasonable position actually is weak without knowing why. "What am I not seeing?"

Sorry, I hope this doesn't hijack the thread, so let me also switch and comment on the major subject, which is analysis quality.

When I was a kid, one of the fascinating things appealing to me about chess was how well some of the books written. There were appropriate anecdotes, questions to ponder, entire chapters devoted to very simple openings or positions.

Against this, something like "computer thinks 25. h4 is strong", followed by a thousand more variations means absolutely nothing and provides no information to a viewer unless the commentator also explains why each move is strong or weak. Any player for whom the engine analysis is sufficiently self-explanatory probably has an engine on his computer and can run it himself anyway.

I favor the idea that a so-called "perfect analysis" is more appropriate for a post-game evaluation, and live commentary is more like watching a game with a much more chess-skilled buddy who explains to you what's going on. A very appropriate computer-use in the latter scenario is to test a line you thought of yourself as well as the running position to see how strong it is for each player.

Kramnik-Fan. They could do like Garry and take the watch off soon after game started and when he deemed the game over put the watch nack on ;-)

Come on now Mig, Patzers? Quit holding yourself in the same esteem as those who are commenting with you on chess.fm...

One of the characteristics of a really strong player is the ability to constantly seek complications in an inferior position and to outplay his opponent in these complications. According to the computers, those complication-seeking moves are not the best, they may hasten the defeat against perfect play, but against humans, it is the best way to turn a loss into a win.

I think the commenatators should ideally operate without Fritzy et al but the current computer evaluation could be in a window in the corner

Mig,

If the time used to setup positions in the application running the engine is an issue, then why don't you find a way to automatically feed the relayed games into the application as the games progress.

From a patzer's perspective it's often beneficial if the commentary can provide an explanation of why computer line X is inferior to human line Y, and in what way that line gives better practical chances.

Patzers tend to think that the GM commentator might not have seen the preferred computer line, and they most certainly do not understand how the two lines compare (as far as practical chances go).

Thus I believe accounting for what the computer "thinks" in the commentary, is a way to get the patzers more interested in the commentary.

Hate to say it, but Carlsen has been very lucky so far. He still has to play black against Morozevich, Anand, Topalov and Leko, so if he manages to stay at 50% I think it would be a great tournament for him. I expect him to lose at least twice in Linares.

Annotaters should give the Fritz evaluations with only one digit to the right of the decimal point (good = +1.2, bad = +1.23).
And these evals change so much depending on the PLY DEPTH. So it is good to stick with one stated depth for most moves.

There is room for improvement over the traditionally vague statements like "Black has the better game here". The problem is this same statement has been used in situations Frtiz evals at +.9 and +2.2.
While 'objective' precision is impossible, 'relative' precision should be possible.
Appending the wordy phrase with a Fritz eval like +.9 adds 'relative' clarity, without overriding the phrase.

- - - - - - - - - - - -
A great annotator will highlight the aspects that are intangible to computers:

(a) Making an objectively weaker move in order to steer into a style of game disliked by your opponent.
(b) Making an objectively weaker move in order to create complications.
(c) Steering toward a type of endgame.
(d) Taking risks to avoid a draw, in situations where a draw does not help win prize money.
(e) Playing extra quickly due to clock considerations.
(f) And so on...

Actually, I wonder how long this list could be, or what exactly is meant by "and so on"?

I think having the super tournaments in smaller areas such as Morelia is a good idea. It would be a benefit for the game to let these different regions experience the joys and excitement of chess.

Hmm, seems like many are being very critical of 'the kid' Carlsen. He is having a great tournament and it is not due to just luck. I'd like to see him pull a stunner on everyone and take the Linares title this year!

Hi Mig. Two comments:

1) Many of the things you say about engine weaknesses don't really apply any longer. It's true that Fritz 10 suffers from lots of these diseases, like mis-evaluating endgames. However, Rybka is in completely other ball-park concerning positional evaluation. You can take the Fritz 10-Kramnik match as an example. Fritz 10 clearly played itself into problems in a couple of the first games. If you go over those games with Rybka, you will see that that engine knew exactly what was going on and had very good end-game evaluation all along (for example, understood that white was better in the night vs. bishop ending). Rybka would have played those positions very well and not have made the same mistakes as Fritz 10.

Rybka is still not flawless in endgame evaluation (no engine is flawless), but I firmly believe that most of your criticism of engines is outdated and mainly applies to over-optimistic engines like Fritz.

2) While you may think (and correctly so) that there's more to human chess than +2.00 changing to -1.00, I think that engines may provide just the right tool to take human chess into the mass market realm. As you mentions, games like Topalov-Moro would be analyzed for months and we would wait for months for publicized analysis in the pre-computer era in order to "learn" what was going on. That we no longer need to wait for strong GMs to present their conclusions like this certainly takes some of the mysticism out of chess, and that is a very, very unfortunate consequence of the advances of computer chess. However, what about this way to present chess to the mass market:

Chess is broadcast, and up in the corner of the screen, there is an engine evaluation. I suspect that this is actually quite interesting to the "patzer" or the complete chess-illiterate. I have this picture in my head: Some guy is sitting in a bar, watching a chess game on the TV. Another guy comes over and asks about the game. The answer is then something like "yeah, Carlsen was up 0.45, but then he blundered, Anand is going to run this one home". That's really how chess could reach the mass market, just as pool reached all the women in Great Britain. They may not have a clue about the variations, but they understand the fight and the personalities and their trustworthyness. "Moro just entered this crazy line, giving the opponent a +1.40 advantage, but he couldn't find it. Now Moro is winning. Man, this Moro is a genious!".

You get it? Engines allow patzers and chees-illiterates to relate to the action on the board. Build on this, and you may have a receipt for success.

I rather think Carlsen makes his own luck. Very hard to walk out of seven rounds of a cat XX tournament with +2 unless you have something more than luck in your pocket. Or do you mean that whenever a top-rated player resigns it's because of their poor luck?

"Leko-Anand would have been just a brilliant game, really brilliant. The whole idea of Ra1-c1-a1 just so that while it was on c1 White could play b4 to wall in his bishop on a5 is amazing. The funny thing is that Bxc6 and especially Bd8 are exactly the two main tactical ideas he must have been working with for the past hour or more when the critical moment arose. He just didn’t manage to put them together in the right combination at the right time."

Yes, I also found the Ba5 manouvre spectacular. The same idea was played by Kramnik against Anand not long ago.
Leko seemed to be on the right way. After Qxg7 there must have been possibilities with Rd8+ I guess. Unfortunately this was the time he lost control when Anand made two active moves (Rxe6 and Bg4).

lucky or not seeing a 16 year old kid win at this level is just utterly amazing to me.

Is that your term Mig? 'afritzionados'? Joyce would be proud.

I understand your doubts, Alkelele, and I am aware that Rybka is considered some sort of holy grail by many in the computer chess community. I agree it displays greater understanding of human chess than most engines, perhaps more than any previous engine. When it first came out I noted how it played with pawns more intelligently than other engines. I have worked with and tested every version of the program, although I don't use it much because it is tactically slower than fast engines like Fritz and Toga. I'm mostly interested in using engines to quickly find blunders and tactics since I trust my own positional evaluation far more than that of any engine. Its evaluations are conservative and sober and therefore less prone to gross misjudgment.

Even Rybka it is still many hundreds of Elo points away from any IM in positional play. Of course there is no purely positional or purely tactical position so the immense tactical strength of an engine compensates in many ways and they generate decent moves without planning or comprehension with the occasional notable exception.

Your description of the masses watching chess via a computer evaluation neatly summarizes the horror I feel watching people shout out "blunder!" every time their Fritz (or Rybka, if you prefer) drops half a pawn in eval. I actually suggested something similar during the ESPN broadcast of the Kasparov-Fritz match, using a graph (similar to the histogram Fritz generates). But we in the supposedly knowledgeable community shouldn't be fooled into thinking such things are realistic.

simsan also makes the good point about amateurs simply finding computer evals interesting in their own right. Partly because they consider them their personal all-seeing oracle, but the differences can be interesting. But that's really a parallel analysis game, trying to explain computer ideas, which are very often inexplicable even when they aren't bad. And I would hate to encourage the perception of the sort of competition between GM and computer analysis many seem to desire, or believe exists. It adds a serious distraction for both the GM and the audience.

Perhaps I'm particularly sensitive to these matters because I regularly go over analysis with strong players, mostly Garry these days. I often check over his NIC column with an engine. I occasionally bring up a move or a line to him and try hard to make sure it's something conclusive or "normal" because I've been stung more than once with, "Come on, Mig! HUMAN chess!" Meaning, of course, that it may or may not be a decent move but it doesn't really affect the evaluation and it's nothing a GM would consider.

"trying to explain computer ideas"

there are no computer ideas. There only exist concrete variations.

One common error about "human chess move finding algorithm" is

1) abstraction leads to good moves.

The opposite is true:

2) good moves lead to abstraction.

the theorem 2 fits well in brain analysis, where it has been proved, that elo is very highly correlated to the ability of memorizing chess positions.

so where come good moves from, it it's not abstraction? It's memory.

Yah, I called them afritzionados during the commentary the other day. Just a little joke. I'd hate for another of my silly words to be deemed a grave insult. It beats "computer chess enthusiast" all to heck.

re: afritzionados.

Damn. Wish I'd have come up with that one myself.

As you know, I'm a very serious person when it comes to chess and related matters; but humor is good for the soul. I like afritzionado's, I hope you won't mind my using it from time to time down the line. I mean it's not like it's migstradamas or something as ominous. Now that would be silly...I mean can you imagine?

>>2690 is exactly the performance all leading GM say we expect from Topalov. Where he belongs to.>>

Actually, I don't think I've heard *any* leading GM say that. I kinda doubt that you have either, Mr. Pantsonfire.
-- Posted by: Charles at March 1, 2007 09:02

Just another mindless Topalov basher who conveniently overlooks the fact that Topalov was a 2730-2770 player before the cheating allegations began.

NO let us look at facts. He never was hanging around 2770. He never got above 2745 before the big spike started. Here is a view of his elo chart on the fide site.

http://www.fide.com/ratings/id.phtml?event=2900084

Everybody throws out opinions and numbers and never looks at the actual facts.

Mig's point, that some class level players give too much credibilty to Fritz' eval numbers, must be true. Yet...

All this Fritz eval bashing has yet to explain why Fritz' evals of +1.1 or higher, made early in middle games when material is even, tend to occur in games White later wins.

When I run Fritz' Compare Analysis on grandmaster games (ply depth = 11), I rarely see the eval swing from say -1.0 to +1.0 and back for purely positional reasons (clear tactical mistakes aside). But if the evals were as unreliable as has been written in this blog today, wouldn't such swings be more common?

I admit I have not rigorously gathered data about this into an MS Excel worksheet, but one tends to get a feel for the numbers by using it a lot. If you say I am wrong about the numbers, I won't argue.

Of course, the best annotations are human based, with error checking and supplemental information from a chess engine.

In the USCF's Chess Life magazine 2007/01 (page 6), the editor said it was tolerable that some columnists do not bother to computer check their submissions, despite the fact avoidable errors are getting through. Humm.

The other flavour of computer chess aficionados might be called Rybyonok.

A patzer is a player 200 or more points below the imagined strength of another player who chooses to view his fellow players in a deprecatory way.

Can a mid-career change of style increase one's Elo rating?
Has there ever been a grandmaster that raised his Elo in a short period of time, just by changing his style? Skills take years to grow, but maybe style can be changed almost overnight?


V. Topalov: To those who imply Topalov may have cheated (not me): Could Topalov's supposedly sharp & sudden rise in Elo be explained by an increase in his aggressive playing style?
Most analysts believe Topalov's aggressive style is one big reason he won in San Luis 2005.


V. Kramnik: Some people bash Kramnik for what they say is his "draw-ish" playing style.
I presume they are implying that this World Chess Champion could have more success by applying his skills to a more aggressive -or- speculative style. How much higher can the WCChamp go?


Bobby Fischer: Fischer said his results with Black improved when he started playing for the win with the Black pieces.

Gene, the rise of Topalov has a lot to do with his work with Cheparinov as well as improved work ethics. Of course there will always be idiots who think there's something wrong with the whole picture but FIDE, ACP or any other authority will not pay much attention to their whining anyway.
Kramnik can be successful when he tries hard. Proof - he won games against Leko and Topalov when he absolutely needed to. It's a pity he doesn't always play that way, but his style is just more drawish in nature. Not as bad as Svidler or Leko though. And obviously it's a great match style.


Miguel wrote: the rise of Topalov has a lot to do with his work with Cheparinov as well as improved work ethics.
---

By coming up with mostly dubious novelties that create unbalanced positions, the kind of positions in which a computer will invariably outplay a human, in which Topalov proceeds to outplay his opponents with extraordinary accuracy. A real genius that Cheparinov.

Mig,

I'm pretty sure computer h/w has reached a point where a chess engine can beat the strongest chess player. You might want to see the following video before you make up your mind about chess engines. The following video is a panel discussion that includes one of the main programmers of IBM's Deep Thought - Murray Campbell - that beat Kasparov in '96. He said that they had created a special evalution function for closed positions which was why Kasparov was completely baffled by the computer's "behavior" in game 2. I'm not saying Fritz plays strong positional chess, but I think if programmed properly, computers can beat the strongest human players even in closed, positional games (I'm not sure how well Rybka fares in positional situations). As Prof. Feigenbaum (who is an AI legend) says in the video, Chess is a surprising AI problem in which brute forces seems to work.

Here's the link...

The link didn't show up in my post, so I'll give it a try again... here we go.

http://video.google.com/videoplay?docid=-1583888480148765375

"Can a mid-career change of style increase one's Elo rating?Has there ever been a grandmaster that raised his Elo in a short period of time, just by changing his style? Skills take years to grow, but maybe style can be changed almost overnight?"

It depends on how much the player's rating was raised, and how brief the span of time.

Tal flirted with a 2700 rating in 1973 (and again in 1979), and showed a dramatically different playing style than when he won the World Championship in 1960. So, that represented a gradual, but effective, evolution in style. Kortchnoi's rating and results improved fairly dramatically after his defection. Did that reflect a change in his playing style?

Nigel Short showed a stylistic change in the early '90s, where he experimented with Classical Open game positions, and achieved success by simply outplaying his oppenent in the complexities of the Middlegame, rather than having to rely on grinding out TNs. It would be interesting to correlate sustained increases in rating by veteran chess pros with a change in opening repetoire.


Topalov's raise through the rankings is neither quick, nor
surprising. Don't your professional ratings (if you work at all,
rather writing to the blogs all the time) fluctuate with the
situation, inspiration, environment, possibilities?? Here is another
revealing interview by Topalov from 2002:

http://www.chesscenter.com/twic/topalovint.html

D.


Consider the pro futbol (soccer) player before a throw-in. His positional accumen, built on years of his own practice, and centuries of human play, suggest it is safest to throw the ball back, to one of his own defensive backs, to reset play. He always glances goal-ward first, to see if there is some offensive opportunity, but most often turns and throws the ball to one of his own defensive backs.

On this throw-in, our pro has already glanced toward the goal, didn't see any tactical opportunity, and has pivoted to throw the ball to his safe defensive back, who is open and waiting. Out from the stands rushes a fan, his laptop in hand! The fan is running Fritz-futboller, a new application. The fan shouts, "No! No! Throw in upfield to Baumgirddler, who will tap to Taddy, and he to Jose, and Mack, and then Grabble will hit it in the goal! A guaranteed goal in just six touches of the ball! Against the opponent's best defense! Fritz-futboller sees every variation here, and it's goal in six!"

The pro will just have to say, "I don't see it," and throw the ball in to his defensive back.

"Oh no, another draw!" says the fan with Fritz-futboller.

(Oh, futbol as we know it is over! Fritz-futboller sees every variation.)

tjallen,

Funny one! If it's a joke, ignore the subsequent lines.

IMO, the game of chess can only be compared with checkers, othello, and Go or gomoku (and maybe other similar games).

Definitely not with football. The factors that influence the game are 22 players, 3 refs, coaches, spectators, weather, pitch, et al.

Compare that with an 8x8 board, 32 pieces, and two players. Close but no cigar :)

Kenny Foster:

The rating list you have is from 2001 ionward.

Before he had higher rating. He reached 2750 in 1996 (or 1995 can't remember) and he was number 3 (or 4) that time after Kaparov, Karpov... And that time the ratings were less infalated than now. (less than 10 players were above 2700..)

So there is a question, why did he droped after that? Was this result of stopping using computer for a while??

Be serious.. He has some fluctuations in his performance, but nothing particularly odd, and we all know that he can play at least at 2750 (as he did in the 90's)

Changes of lifestyle are much more likely to account for peaks in rating than changes in playing style - see Korchnoi, Sosonko (defection), Tal (health peaks, basically), Short (really put the effort in 1990-92 or so). Topalov went through just the change in circumstances which might well produce such a peak in 2004-5. Unfortunately, it's also the exact same change which would make computer cheating possible, hence all this stuff.

While Mig's Luddism is admirable and even rather touching, I too think he's been rather too influenced by working with Kasparov. Kasparov is not 'a GM', he's the Man, and his judgment still is worth taking notice of over a computer's. However, I can definitely assure you that Fritz's positional judgment is far superior to that of this IM, and I suspect any others. Give it a collection of twenty positions and a minute each, and the same to me, and if I'm closer than it is once I'll be lucky.

Alkalele's suggestion is very interesting. He did rather fool me with his suggestion that all UK women love pool (Footballer's Wives came to mind) but on reflection I think he meant snooker. But whether you regard his vision as heaven or hell, it's what already happens when folk are viewing Linares and the like on the internet. Commentators need to wise up or they'll become irrelevant. Mig is trying to hold back the tide. Remember Canute?

Zakki wrote:
------------------
Miguel wrote: the rise of Topalov has a lot to do with his work with Cheparinov as well as improved work ethics.
---

By coming up with mostly dubious novelties that create unbalanced positions, the kind of positions in which a computer will invariably outplay a human, in which Topalov proceeds to outplay his opponents with extraordinary accuracy. A real genius that Cheparinov.
-----------------

You imply Topalov's cheating at the board, but don't you think it would be enough to have analysed the position in depth beforehand (with computers), and have worked out some of the possible lines? That would give you a big advantage at the board (and players tend to play more accurately/blunder less in familiar positions).

The good thing about dubious novelties is they have suprise value and, as we saw in the WC match, you can get away from theory early without giving your opponent a chance to get in his own novelties. It's a tactic I'd expect others to adopt (sadly).

Censorship on Susanpolgar Blog! She can't stand any critics.

This item has been deleted:

"Ellrond hat gesagt...
don't believe Susan. She supports Topalov. A person who does this can do everything. "

I posted it again Friday, March 02, 2007 6:00:00 AM, look yourself and see how it will be deleted.

Free speach!


I am in no relation to Sloan, I am german and don't even know the name Sloan.

the critics have been posted in Susans USCF election thread

"My Statement for the USCF election "

Susan has lived in her youth in a dictatur without free speach. Perhaps that explains her affection to totalitarian methods?

I think Mig and Susan Polgar http://susanpolgar.blogspot.com/ presently have the best chess sites on the net. I must have a hundred chess links, but these two are the ones I check lately on a daily basis. Lots of entertainment in both. I'm hoping Susan takes the USCF elections by storm btw.

lwolf123, hehe, after Bush as head of the country, are you daring to have somebody similar at the head of USCF? A despot or so. I can understand that. Makes things easier.

Look, how Susan acts on critical opinions now. No question, what she will do with the money of USCF then.

Does somebody have a good site to look up FIDE ratings through the nineties? Foster's link certainly demonstrates Topalov as never achieving 2750 and never even coming close to 2780 before 2005 and derida insists that Topalov's rating used to be much higher.

As you can see from this paper: http://members.shaw.ca/redwards1/ after his initial rise Topalov peaked once at around or little under 2750 in the late mid-90s and then quickly plummeted back down to 2700 where his rating stayed till 2001 or so. At this point it gained about 30 points (FIDE website) and stayed level for roughly 3 and a half years. Which brings us to the current peak.

Comments, advice appreciated at www.gettingto2000.blogspot.com

Posted by: Miguel at March 1, 2007 22:38
Miguel wrote: the rise of Topalov has a lot to do with his work with Cheparinov as well as improved work ethics.
---

By coming up with mostly dubious novelties that create unbalanced positions, the kind of positions in which a computer will invariably outplay a human, in which Topalov proceeds to outplay his opponents with extraordinary accuracy. A real genius that Cheparinov.


Posted by: zakki at March 2, 2007 02:08

As I said in my previous post: Of course there will always be idiots who think there's something wrong with the whole picture but FIDE, ACP or any other authority will not pay much attention to their whining anyway.

lwolf123, hehe, after Bush as head of the country, are you daring to have somebody similar at the head of USCF? A despot or so. I can understand that. Makes things easier.

Look, how Susan acts on critical opinions now. No question, what she will do with the money of USCF then.

Posted by: Ellrond at March 2, 2007 09:09

Well, Ellrond, maybe you should put on a turban, become a Taliban and go live in Afghanistan. You'll be free of them evil despots.

Of course there always be idiots who think their opponents are idiots.

Of course there always be idiots who think their opponents are idiots.

Posted by: Vlad Kosulin at March 2, 2007 11:44

Or idiots who who call their opponents idiots to try to belittle the fact that they are idiots.

And let's not forget idiots who think idiots who call their opponents idiots to try to belittle the fact they are idiots are not idiots, while their opponents certainly are not idiots but rather think that the idiots are idiots, even though they have read Fedor Dostoevskiy's novel "The Idiot".

What's wrong with the idea of Susan Polgar in charge? We know the clocks would run on time.

I still think there's just too much animosity, name-calling, and personal attacks in the chess world, and that Susan and Sam should just set aside their silly bickering and try to get along.

nowadays censoring seems to be acceptable in the U.S. and a sign of good leadership.
In that meaning Susan is a good leader.

>> lwolf123, hehe, after Bush as head of the country, are you daring to have somebody similar at the head of USCF?

You've lost the debate before it started Elrond, via corallory of Goodwin's law.

http://en.wikipedia.org/wiki/Goodwin%27s_law

Better luck next time friend!

This thread has become too idiotic.

the base of this phenomenon is the tendence of every discussion to go from concrete to general. That is a human basic.

>Susan has lived in her youth in a dictatur without free speach. Perhaps that explains her affection to totalitarian methods?

Free speech is a guarantee of protection. For some reason weak minds treat it as an entitlement to have their dung published. Let's not blame the system, as it can be seen in this case that the German system is not idiot-proof either.

D.


Chess is originated in India. The Indian origin of the game of chess is supported even by the Encylopedia Britannica according to which, "About 1783-89 Sir. William Jones, in an essay published in the 2nd Vol. of Asiatic Researches, argued that Hindustan was the cradle of chess, the game having been known there from time immemorial by the name Chaturanga, that is, the four angas, or members of an army, which are said in the Amarakosha (an ancient Indian Dictionary - S.B.) to be elephants, horses, chariots and foot soldiers. As applicable to real armies, the term Chaturanga is frequently used by the epic poets of India. Sir William Jones' essay is substantially a translation of the Bhawishya Purana, in which is given a description of a four-handed game of chess played with dice." "Sir William, however, grounds his opinions as to the Hindu origin of chess upon the testimony of the Persians and not upon the above manuscript," He lays it down that chess, under the Sanskrit name Chaturanga was exported from India into Persia in the 6th century of our era; that by a natural corruption, the old Persians changed the name into chatrang; but when their country was soon afterwards taken possession of by the Arabs, who had neither the initial nor the final letter of the The Indian origin of the game of chess is supported even by the Encylopedia Britannica according to which, "About 1783-89 Sir. William Jones, in an essay published in the 2nd Vol. of Asiatic Researches, argued that Hindustan was the cradle of chess, the game having been known there from time immemorial by the name Chaturanga, that is, the four angas, or members of an army, which are said in the Amarakosha (an ancient Indian Dictionary - S.B.) to be elephants, horses, chariots and foot soldiers. As applicable to real armies, the term Chaturanga is frequently used by the epic poets of India. Sir William Jones' essay is substantially a translation of the Bhawishya Purana, in which is given a description of a four-handed game of chess played with dice." "Sir William, however, grounds his opinions as to the Hindu origin of chess upon the testimony of the Persians and not upon the above manuscript," He lays it down that chess, under the Sanskrit name Chaturanga was exported from India into Persia in the 6th century of our era; that by a natural corruption, the old Persians changed the name into chatrang; but when their country was soon afterwards taken possession of by the Arabs, who had neither the initial nor the final letter of the word in their alphabet, they altered it further into Shatranj, which name found its way presently into modern Persian and ultimately into the dialects of India."

word in their alphabet, they altered it further into Shatranj, which name found its way presently into modern Persian and ultimately into the dialects of India." The great Mahabharata War is the result of Chess. There Srimad Bhagavd Gita originated in the battlefied preached by Lord krishna to motivate Arjuna to do his duty as a warrior.

One should add that cheating in Chess also originated in ancient India. The Danailov of those times was Shakuni, uncle and mentor of prince Duryodhana (Topalov). When the aging king Dhritarashtra (Kasparov) was about to retire, Shakuni/Duryodhana played a Chess tournament against the rightful heirs to the throne, who bet their entire kingdom on the outcome. Shakuni swindled them out of the match (the exact cheating mechanism is unknown, much like today), and it took a war of near-total annihilation to set everything right.

I won't go through the whole story, which takes 90,000 verses, but you can look it up under 'Mahabharata'.

Twitter Updates

    Follow me on Twitter

     

    Archives

    About this Entry

    This page contains a single entry by Mig published on March 1, 2007 12:22 AM.

    Linares 2007 r7 was the previous entry in this blog.

    Linares 2007 r8 is the next entry in this blog.

    Find recent content on the main index or look in the archives to find all content.