Kramnik-Fritz 06 g2

Oh, give it a rest, Mig, with your ‘basically pick the opening’. Or at least explain why – isn’t it true that the computer’s opening book shows only the probability of certain moves?

Mind you, Kramnik looks to me to have made a shrewd choice here. To my eye this would be a good type of position to have against a computer, saying nothing about the objective merits, you understand.

I looked at Kasparov's idea after 30.e3 (followed by Ng2 at some point) myself and so far I wasn't able to find a refutation -white indeed seems to win in all lines. So I won't bet against Kasparov, but it sure gives incentive to look further! The position after 30.e3 is deceptively drawish looking so it is almost hard to believe that black loses no matter what he does.

I'll take that bet Mig.

I bet Kasparov $5 he cannot win that position as white against me.

Please forward this message to him and I will be glad to take him on Fact to Face any time he is in the NY/Boston area....

:)

Massive blunder!!!!!!!!!!

Confirm on Qe3?

Wow, that's gotta rank up there as one of
the worst moves ever by a World Champion.

Going a few moves back, allowing a Q+N
vs Q+B endgame was risky, since everyone
knows Q+N creates more mating threats...

Oh my God! Can anyone think of a worse move by a sitting world champion?

Throwing away an almost win, then missing mate in one right on the next move...

It's true, yes.
And I am very sad.

THE worst ever, if it's true. I don't see what Fritz had in mind after 34...Kg8. But the site seems to have crashed; is it really right, I wonder? My team mate in the 4NCL had this happen recently - sensory board started playing the post mortem moves and before you know what history's recorded a whole different game to what was played.

Un-freaking-believable. I'm embarrassed to say it took me a few seconds to see White's mate threat. I can't imagine how Kramnik could have missed it. I guess it must be nerve-wracking trying to play perfect chess, but DAMN!

> Oh my God! Can anyone think of a worse move by a
> sitting world champion?

Definitely not. Closest I can think of was Petrosian leaving a queen en prise. I'm not even certain he was champion at the time.

To be _slightly_ fair, I can see how one (if they were me) miss such a mate. But seriously...

> But the site seems to have crashed; is it really
> right, I wonder? My team mate in the 4NCL had this > happen recently - sensory board started playing
> the post mortem moves and before you know what
> history's recorded a whole different game to what >was played.

I hope you're suggestion is true, God help me.

Site seems confident enough now. What a shame. Still, it will give Topalov's gang something to be happy about, which is nice. They haven't had so many laughs recently.

OH MY GOD.

This can have a much worse effect on Kramnik than resigning in a drawn position in game 2 had on Kasparov in his Deep Blue match. I just hope this doesn't turn into Adams-Hydra II.

And yes, this is the heavy favorite in the "Worst move ever by a world champion" contest.

He played what?????

I feel so sorry for him. This must be the worst moment of his career. How very sad. He had such an impressive game going there.

On the bright side, it was fun :) Crazy humor. Looking forward to some pictures of a red faced Kramnik. Let us hope he does not get dispirited.

“See what happens when you DON’T go to the toilet 50 times. (Soon to be quoted by Danailov by some rag of a paper)

Kramnik WILL recover, no worries from me.

“See what happens when you DON’T go to the toilet 50 times. (Soon to be quoted by Danailov by some rag of a paper)

Kramnik WILL recover, no worries from me.

Well, missing mate in one in an even position will stand as the worst ever until a world champion resigns in a winning position, the ultimate blunder.

It is easier to get a feel for the game when there is a living, breathing opponent sitting across from you. A computer operator sitting there is not the same (that guy does not have to think about the game at all).

In a kind of justice, Kramnik is experiencing chess the way he made Topolov experience it.

Mig said: "Well, missing mate in one in an even position will stand as the worst ever until a world champion resigns in a winning position, the ultimate blunder."

Well, I don't know, Mig. If I world champion resigns in a postition where he or she can give mate the next move, then *that's* a worse blunder. Missing getting mated in one is second only to that, I think.

Kramnik said: "FRITZ WILL PUNISH YOU IMMEDIATELY!" and it did!

This will be a very known position in Checkmate in 1 move for beginners!

It's reassuring to know that Topalov's supporters can always find new depths to go to.

Was it really equal? I'm sure everyone's right, but what was White's next move going to be?

Well, having established that it is the
worst move by a WC ever, it should be
interesting to discuss why?

I propose that the human expectation was the check
..Ng6+, which blocks Qh7...

I truly do not believe Kramnik missed the checkmate. What I believe happened was that he was thinking 34... Kg8 35. Ng6 Qe3 draw, but instead played 34... Qe3?? (Thinking the moves 34... Kg8 35. Ng6 had been [played)
Simple mental transpositional error. It happens to the very best!

I wonder if there is video footage of Kramnik making his move and Fritz's response. Maybe it is sadistic to ask to see it, but I think it would be of interest.

I wonder if there is video footage of Kramnik making his move and Fritz's response. Maybe it is sadistic to ask to see it, but I think it would be of interest.

Well, having established that it is the
worst move by a WC ever, it should be
interesting to discuss why?

I propose that the human expectation was the check
..Ng6+, which blocks Qh7...

On a conspiracy note . . .

Maybe Krammnik discovered that he really can beat this program with the rules as they are and decided to "drop-one" fight for a drawn match and then call for a rematch. We "only humans" would love it-- he would make a few more bucks

Probably not

On a conspiracy note . . .

Maybe Krammnik discovered that he really can beat this program with the rules as they are and decided to "drop-one" ---fight for a drawn match and then call for a rematch. We "only humans" would love it-- he would make a few more bucks

Probably not

Well, having established that it is the
worst move by a WC ever, it should be
interesting to discuss why?

I propose that the human expectation was the check
..Ng6+, which blocks Qh7...

On a conspiracy note . . .

Maybe Krammnik discovered that he really can beat this program with the rules as they are and decided to "drop-one" ---fight for a drawn match and then call for a rematch. We "only humans" would love it-- he would make a few more bucks

Probably not

To err is human...

too bad the computer is unforgiving!

A blunder is never just a blunder. Humans just do find it psychologically very difficult to play against computers. Kasparov famously resigned in a drawn position, then fell into that absurd opening trap – the two worst blunders of his entire career (certainly pre-Khuzman). Kramnik too resigned in a drawn position and now makes the worst blunder of his career.

It’s a strange thing, but there’s enough evidence now that it’s not so surprising. Remember Fischer's inexplicable ...Bxh2, as well. That was even stranger in a way since it was such a well-known pattern.

It is hard to imagine what Kramnik was expecting, since White's next best choice after Qh7 was resigns.

I hope to see the video in next CBM or, even better, on the ChessBase web site.

sorry for duplicate postings

"Kramnik said: "FRITZ WILL PUNISH YOU IMMEDIATELY!" and it did!"

Well, the irony is that this error was so elementary *anybody* would have punished it. Topalov is probably thinking, "Why didn't he give ME one of those?"

Actually, this is almost a mirror image of the Topalov match, in which the score after 2 games should have been 1.5.-0.5 for Topalov, but instead was 0-2 for Kramnik.

If there's a silver lining for Kramnik, it's that he achieved better positions in both games. Although he's staring at 0.5-1.5, it's a safe bet that he'll be able to generate winning chances in the remaining games.

I wonder what the Fritz team was thinking with 3.e4? As far as I can tell, it's not a commonly played move. I suspect they won't want to play that one again.

Was it really so clearly a win for black
with correct play?

He should have traded the Qs in the opening.
Anyway, well spotted by the comp.

what was low about what I posted?

It is relevant that Kramnik treated Topolov impersonally in the last match, and now he is the one not playing a person.

It is relevant because now it is Kramnik who is blundering. To point out this parallel is not 'low'. Quite the opposite. I am pointing out the value of human interaction. Human beings are what is great in this world.

That was UUUUGGGGGGGLLLLLYYYYY.

He'll bounce back. By far the stoutest match player in the history of the game.

No worries.

Mig: did you give Kaspy my bet? Still waiting....

IMHO, the blunder was caused by psychological issues.
Kramnik lost concentration, because he got feeling, based on how first 2 games proceeded, that he found the way to get very pleasant positions, and to beat Fritz in the match without taking risks to lose a single game.

It is very easy to over-look the opponent's choices in so pleasant situation, when the game proceeds under your 100% control, you understand what and how your opponent thinks, and you play for 2 results only, after initial psychological stress of not knowing what to expect.

I am not Kramnik, of course (but I am Vlad ;-), and I happened to lose mate in 1 twice under similar circumstances (a much better simple position with no chances for my opponent)

Good call, Morrowind. The best explanation I've seen, anyway. I suppose Black has no good move after 35 Ng6 - I sort of assumed he would have some way to stop perpetual and get control, but I guess he doesn't.

Yes, awful. That Kramnik is a disgrace to the human race. Fancy making a mistake -- of all things! Should be shot. Do they still have guns in Germany?

Kramnik is strong in simple positions but after 32.Nxe6 the game became a bit tense and with the Qs on board the rest just followed....

What a bad joke this match is.

I just now saw the position?! Yes, he will recover from this, but he'll never forget it. I'm sure Kasparov still thinks about that Caro-Kann; in fact, I'll bet it crossed his mind when he saw Kramnik's mate-in-one blunder.

"Drop one" and play for a draw to get a rematch?

No way. If a human wins, it's a guaranteed rematch. A draw would be unnecessary. These matches will be sponsored as long as the producers have software to sell (and the humans don't get pummeled ridiculously).

And once computers surpass humans in every stage of chess, the publicized matches will then be held between computers to determine which program is the best.

No, there is no need to "drop one" to get a rematch.

Going back to the first comment by rdh, by "pick the opening" I refer to his ability to know in advance all the options at Fritz's disposal without having to remember anything other than the ones he wants. With such a guide and limited probabilities it's almost impossible to get an opening position he doesn't want to be in. There are no surprises. So while that doesn't entirely nullify white's typical opening advantage, it completely alters the traditional landscape of opening work. Pick a line with relatively few options, like he did today, and he can be close to certain to get a position he has already analyzed with the engine at home. That's usually going to be more important than playing the theoretical best.

I just told Kasparov about it; he wasn't watching. Not much of an opinion, really. He considers the whole match a joke because of Kramnik getting the engine in advance. I'm sure he'll have a few thoughts on the blunder later, but I'm not really sure there's much to say.

Yes, Mig, Kramnik got again and with utmost ease an excellent, ideal, postion in the opening.
It all looks as if this is an "opening odds" match.

The unusual rules rob the event of interest, no matter how it turns out.

Should Fritz win, I think we can safely say that big-money human-comp matches are over. If the world's best player can't beat Fritz with all of these extra advantages, then nobody can.

Should Kramnik win, you can't really take it seriously, because he the usual rules had to be modified to make it a fair fight.

Human v. human chess is a poker game. Bluffing, acting confident, keeping a poker face. An opponent sitting on a mate-in-one probably acts or even smells slightly differently. And in My Sixty Memorable Games Fischer said something about tuning into his opponent's brain waves.

When you play against a computer, however, all these odd little "tells" vanish. Kramnik and Kasparov have played only a handful of computer games, but this handful contains some of the worst blunders of their careers.

Kramnik just did not see it.

There is a similar game in chessbase.com data base : Danailov-Marin ( yes that Danailov, he is IM)

Danailov got mated in 1 on board by Marin in similar position ( Nf1, Qh2# mate)...such mates are not "pattern"-like, not as [Ng5,Qh7#], or [Ng4,Qh2#] respectively for Black, which are easy ,"automatically", to spot and fend off

i am important, i am important i can call kasparov!!!

To anon,
Mig worked for Kasparov at the former eponymous website, and they are pals. Now that Gary has retired, Mig is an important conduit to the chess world of Kasparov's opinions. Although one might be jealous about this, one shouldn't begrudge Mig, who has paid his chess-journalist's dues. I, for one, welcome Mig's access to Kasparov and his opinions.

tjallen

Mig, I wish you'd answer the question about what exactly Kramnik can see, because you keep saying something different to what chessbase seem to say. According to them, there is no advance knowledge of the opening book. All that happens is that during the game the screen gets turned round so long as the machine is in its book. I am too stupid to turn on Fritz's opening book myself, but as I understand it, what this will show is a list of moves with a weighting which show the probability of the computer playing that move, and also a number denoting the computer's assessment of it. Am I wrong? If so, this doesn't seem to me to give anything like the possibilities you're suggesting.

Anyway, you can see why Chessbase are happy enough with that. From the point of view of testing how strong the programme is and its strengths and weaknesses, this is a far superior way to conduct the contest. It doesn't tell us in a sense whether Kramnik or DF is the stronger, but perhaps that's not really as interesting a question?

In reply to Marc Shepherd: you're kidding, right? 3 e4 is virtually the present day main line of the QGA.

Ovidiu is right of course, this is a non-standard pattern. But I still find it very hard to believe Kramnik just missed it, because if the mate weren't there Fritz would just have collapsed in exactly the fashion you or I do all the time under pressure but computers never do. Kramnik could hardly have been thinking that that was what had happened. Still, he'll tell us himself at the press conference, I imagine.

"All that happens is that during the game the screen gets turned round so long as the machine is in its book. I am too stupid to turn on Fritz's opening book myself, but as I understand it, what this will show is a list of moves with a weighting which show the probability of the computer playing that move, and also a number denoting the computer's assessment of it. Am I wrong? If so, this doesn't seem to me to give anything like the possibilities you're suggesting."

The computer's opening book is the electronic equivalent of a GM's memorized openings. Now, as in everything it does, the computer's memory is flawless and more capacious than any GM could manage, but the basic idea is the same.

Now, imagine you were playing a human opponent, and he was obligated to divulge his preparation -- which openings he was favoring, how he intended to respond to each possibility. Doesn't that seem like a pretty significant advantage to you?

rdh
When it is the computer's move, in the opening, Kramnick doesn't see anything relevant on the screen, because the computer makes the book move immediately.

So it must be when it is Kramnick to move, that the screen becomes interesting. What can it show, other than the moves it expects from Kramnick? And that is the key. Kramnick can see the lines in the computer's book, among which he is expected to choose, if he is to remain in the computer's book. If he has a novelty, he will note that the novelty does not appear in the computer's book-based list. He knows which moves of his will provoke calculation, vs. which will provoke a booked response.

I'm not a good enough player to say how much that's worth knowing.

tjallen

I watched the game unfold on ICC, while listening to the commentary. Although I had work to do, it was fascinating to see the game unfold, the subtle threats and counter-measures...I was getting into the logic of the game as if I was playing it myself. And then for Kramnik throw it all away, it was heartbreaking to see. There is little point to continue the match now, how can anyone come back from that?

It also means that your memory isn't taxed as much because you can see the options yourself. It's not as if the opening book for this match is going to include hundreds of novelties that improve over mainline theory. More to the point, they had to go through it closely to chop OUT many lines they need to avoid. It's a much smaller book, though some lines might be deeper in some areas.

rdh, since when is 3..b5 the main line of the QGA? I believe we were discussing Kramnik's choices, not Fritz's. That move was actually brought up by Dlugy for a clock simul game against Kasparov in 1993 and virtually forces transposition into a narrow line with few options for White. And of course Kramnik could see if the book had any non-standard lines added other than those he had prepared. If you can get to a position with fewer than a dozen elite games in the databases by move four and you have the engine, you can prepare the game very deeply indeed. Obviously there are too many possibilities to do this successfully every time, but even if you only get it half the time that's a massive advantage. Heck, just the clock advantage you get from having an opening book to look at is far from insignificant.

Anon, I work with Kasparov full time for a living, not to impress you or anyone else. Get a life. If you don't want to hear about him you're at the wrong blog.

For me, this defeat has nothing to add to the human-computer struggle series.
If was a man vs man overlook this would mean something. Now it does not.

However, Kramnik has overplayed the computer since now, if ones tries to be objective.. (yes, I agree with Karpov's statements regarding the match).

Moreover, it's totally ridiculous for the computer to use all those books created by chess players and to consider its victory as "belonging" to it (again agree with Karpov).
It's rather a humans(top chess players, chess programmers, Microsoft programmers, Intel h/w CPU developers etc etc)+computer vs human alone (Kramnik) struggle.

Any transcripts of the press conference out there?

Hahahahahahahahaha:)
Well obviously Kramnik deserves the world champion title for blunders.
Why the organizers did not invite Topalov to play ageinst Fritz? They do not want Fritz losing:)

Mig: Marc Shepard said "I wonder what the Fritz team was thinking with 3.e4? As far as I can tell, it's not a commonly played move," to which rdh's response was "you're kidding, right? 3 e4 is virtually the present day main line of the QGA." Neither was talking about 3...b5.

marca, I thought that according to Topalov (and at least some of his fans), Topalov had already played a match against Fritz in Elisa. We all know how that ended.

Mig, I was replying to Marc Shepherd's 'I wonder what the Fritz team was thinking with 3 e4. As far as I can tell it's not a commonly played move. I suspect they won't want to play that one again.' Of course 3...b5 isn't the main line, although my better-informed Slav spies tell me it's a known opening trick and has been becoming very slightly fashionable.

Marc; yes of course it's an advantage. But it's not anything like as significant an advantage as Mig is making out, unless the conditions are different from what I think.

Since Mig still hasn't answered, I'll ask again:

1. Chessbase suggest that Kramnik only gets to see the opening book once the game has started, not in advance of the match. Is this correct?

2. Is it correct that all you see are the replies the machine might make to any move you have in mind, a number indicating how it rates that move, and a number indicating the probability of it playing that move? Or do you see something else, and if so, what?

3. (I haven't asked this before). Does anything in the rules prevent the Fritz team changing the book between games?

As usual, the devil is in the detail, not in anti-Kramnik rantings on blogs. We saw the same thing during certain other recent events.

Does no-one else think that this is quite a clever device designed to answer a more interesting question than the one we'd have answered with the usual rules when you or I take on Fritz, then?

Looking at the Fritz book, and assuming for the sake of argument they didn't add anything special to this obscure line, Kramnik could be 100% sure of the position reaching move 9 after Fritz played 3.e4. There's a 50% chance between 9.Nf3 and 9.Ne2. Then, after 9.Nf3 he would know that Fritz wouldn't be in book after 9..a5.

As tjallen points out, just knowing what moves will and won't take the computer out of book is a major factor too. Kramnik can simply try to play a tricky order to encourage some dumb computer opening play. For example today he played ..a5 instead of the natural ..Be7, which was in Fritz's book.

Anyway, while there is no way to define the absolute value of these advantages, they are undoubtedly substantial.

> Oh my God! Can anyone think of a worse move by a
> sitting world champion?

Definitely not. Closest I can think of was Petrosian leaving a queen en prise. I'm not even certain he was champion at the time.

-- Posted by: tg at November 27, 2006 12:12

That is the famous Petrosian v Bronstein 1956 game in which Tigran simply missed ... NxQ in a completely winning position.

And you are right, Petrosian was -not- the WC when he blundered his queen in this game.

One might reply, george, that you and I use all these books written by other players and yet still count our victories our own.

I can't get excited about questions of what's fair or creating a 'level playing field' or any of these - to me spurious - arguments about trying to balance various advantages computers have. It's a question of what you're trying to achieve with the match and what will make the most interesting contest, not of what's 'fair' (an absurd concept, frankly).

IMHO, the opening book complain is not legitimate. Fritz team could easily analyze all games played by Kramnik to better understand his playing habits and to prepare an anti-Kramnik book with novelties targeted at Vlad's style of play. Therefore, giving Kramnik access to the engine and to it's book is fair. Also, after every game the Fritz team can adjust the book.
By the way, giving engine access to 5-man tables is also not that fair, because Kramnik does not have access to endgame reference manuals. I can understand the ChessBase fear that Fritz will play dumb in endgame without these tables, therefore, we could say that giving Vlad access to Fritz opening book is equalizing the Nalimiv availability for Fritz, and availability of hundreds of Kramnik's games to adjust Fritz to play Vlad.

Kasparov's comments that giving Kramnik access to the engine makes match a joke are also pointless, IMHO. During match preparation it is usual to analyze the opponent's style. Everybody have access to hundreds of games played by Kramnik vs. top opposition, and Vlad's analysis in CI, NIC, etc. to understand his style, his advantages and weaknesses. The only way to understand how Fritz plays, is to play it, because there are no games published to serve this purpose at all.

Engines became too strong to play them as black box. To succeed you need to know whom you play with. Which became the case in GM chess at least 50 years ago. Now engines join the club.

But Mig, let's be clear - he only gets to see this during the game, right? Not in advance of the match. He saw Fritz's White opening book for the first time today. Or is that wrong?

Let's hear exactly what the conditions are and then we can decide for ourselves what degree of advantage they convey. The first duty of a journalist is to inform the audience properly before commenting, wouldn't you agree?

If I'm right, then the main advantage it gives the human, I would say, is the ability to see any TN coming and avoid it. But then as you say I don't imagine Team Fritz has been busting a gut to prepare any of those anyway.

One might reasonably suggest Petrosian's blunder was worse - it was a whole point rather than a half, after all. But I don't see that it's very important.

The rules are on the official site and they make clear what Kramnik gets and what he sees. They also spell out the limited way the Fritz team can alter the book between games (10 half moves, I believe.) Feel free to read them yourself.

And no, I don't think it's more interesting to show the computer's opening book to the human. Give Kramnik access to his own opening book perhaps, but why show him the computer's? It's bizarre.

rdh,

yes, without mate Fritz was lost after the "forced" trade of Qs ( immediately or after Ng6+/Kh7/Nf8++/Kg8)

perhaps Kramnik thought that Fritz did not "see" that the N vs.B ending, albeit equal materially, was lost for white. Comps are thought weak in the assesment of endings. He thought he was winnng when he got mated.

I will consider this result today a blessing in disguise. Now when I blunder in a tournament game, and it will happen, I can now say it's not so bad. After all, I know of a world chess champion who once was mated in one. Then I can (nonchalantly)leave the tournament hall, return to my hotel room, and proceed to break all the furniture.

1. Forcing Fritz out of the book at move 9 have only limited usefulness, because it is harder to predict future Fritz moves even after deep in-home preparation, and you take more risk of spending much time OTB and getting tired earlier. Staying in the book as long as possible is not bad if you know the book engine uses. Avoid the book lines only if they serve to engine advantage.
2. Fritz could easily go out of Vlad's preparation playing King's gambit, for example. Whatever you say, it is healthy opening, which lets engine play wild chess where human can't stand equal.

Indeed they are, Mig.

http://www.rag.de/microsite_chess_com/regeln.html. for anyone interested.

However, I am too stupid to know what they mean. They say that the 'final match version of Deep Fritz' shall be made available before the match, but I don't know whether that includes the opening book or not. They seem to suggest elsewhere that DF shall be regarded as 'using' the opening book, which might suggest they are seen as different entities. I had the impression though from what Chessbase say that the openings book is only shown on the day.

Do I gather that you don't know either? If so then, hey, why not say so and admit that much of your ranting about the rules has been based on ignorance? Or alternatively perhaps you could tell us all.

Ovidiu; no doubt Kramnik will tell us himself. You won't believe him, of course - I have the impression your opening book isn't set to allow you to do this.

It's not so much about predicting Fritz's moves in the middlegame, it's getting the type of position you want out of the opening. Kramnik wants to avoid sharp open lines and is happy to let Fritz out of book if he has already achieved a rigid structure, as he did today.

As for the King's Gambit, which I recommended to the Fritz guys in Bahrain and which I play myself, there are many forcing lines that would be trivial to follow while looking at the computer's book of options. If they cut off the book after five moves to avoid that I don't think it would play well. Computers don't understand gambit play. They see all the defenses and are just as likely to liquidate into an inferior ending as they are to play on sacrificially. If the ending is .01 better than the sharp attack they'll play the ending. This is why programmers don't play such things against humans. Gambit play is a state of mind, not just a variation.

And Karpov's loss in 12 moves against Christiansen was bad - but not as bad as this.

There is a very interesting idea on why Kramnik made the blunder. Yusupov said that the knight on f8 is unusual for the mating pattern, and the knight's head turned on to the left could force Vlad to overlook the mate! According to GM and professional psychologist Helmut Pfleger (not sure the spelling is correct), this have some ground; he also pointed out that Seirawan in his comments made a similar mistake.

Perhaps this is not the best time. But since I like to link to Tim Krabbé as often as I can, take a look at this.

http://www.xs4all.nl/~timkr/chess2/resigntxt.htm

"Moreover, it's totally ridiculous for the computer to use all those books created by chess players and to consider its victory as "belonging" to it (again agree with Karpov)."

Kramnik, as brilliant as he is, didn't invent all of opening theory himself. Just like all of us, including Fritz, he is standing on the shoulders of those who've gone before.

IMHO, it's unfair that Fritz gets human assistance in the first place (in the form of an exogenous opening book). If it is really that unfair that Kramnik can get a peek into that book, then switch the damn book off and we'll see what's what...

It should be pure human against pure machine (Kramnik don't get Fritz in the first place, and Fritz don't get human help either). Fair and square.

Look, I play with a friend of mine which perhaps has a lower understanding of the game and I often beat him playing good chess myself.
Moreover, I have played hundreds of games against Fritz unleashed (~2700 in my case). Only 1% of my games were good (of course I always lose). I am always playing stupidly and hastily.

Seems as if playing against a human and against a machine is different in psychological aspects.
Kramnik would never make such a bad move against Topalov, or Kasparov, but he did it against a impersonal machine.

Maybe is the anxiety of the errorless opponent ?
I don't know ..

Kramnik would never played so badly against a human? what about the second game against Topalov? he allowed a mate in 2 or 3 there. seems he has problems seeing mates.

"In reply to Marc Shepherd: you're kidding, right? 3 e4 is virtually the present day main line of the QGA."

Well, I think 3. Nf3 is still the more commonly played move. But as Mig suggested, what I meant by my comment is that I don't think the Fritz team is going to want to allow 3.e4 b5 again. I was not suggesting that 3.e4 itself was novel.

>"Yusupov said that a knight in f8 is an unusal for the mating patter">

I have argued here for this idea.

Rremember the Qc7 missed by everybody during the Elista match.
Not only Topa and Kramnik but also the other GMs ( Azamaiparashvili, etc.) analyzing without computer's help.

It was a different psychological issue then.
The focus on K-side makes one to ward off
from analysis the
Q-side choices as "irrelevant", "beside the point", etc. which is OK in 99.9% of cases, but that was the 0.1 case when it was not OK).

During my university years some teachers let us use any resources we thought we need to solve an advanced problem during exams, because the study main purpose was not in remembering as much as we need for a specific exam, but in ability to solve problems we face. This is very different from the "pick the right answer" approach, and this is what makes human's only advantage compared to computer.
Do you want to see chess becoming another trivia competion?

The main difference between human and engines development during last 5 years is that engines become better and better in chess understanding (pattern recognition and position evaluation), but human's memory does not advance much if at all.
Therefore, to equalize their chances, and if we can't avoid usage of "reference knowledge" (the opening preparation, Nalimov tables, etc), I'd let humans and engines use any print references and databases they wish during a game. This would make real competition in game understanding and creativity, not in remembering known facts.

During the rule negotiations for the 2003 Kasparov-Junior match we talked about allowing Kasparov to access an opening database of his own to level the playing field in that department. (It didn't happen of course.) To me that makes more sense than Kramnik seeing the computer's.

Handing over the exact engine is the worst though. Why do that? It's not like the opening or tablebase rules that attempt to address computer advantages that can be perceived as unfair.

Regardless, four more openings like these last two and Kramnik will certainly score at least one win.

"IMHO, it's unfair that Fritz gets human assistance in the first place (in the form of an exogenous opening book). If it is really that unfair that Kramnik can get a peek into that book, then switch the damn book off and we'll see what's what..."

It's a little silly to talk about "Fritz without human assistance." That's a contradiction in terms. Besides, Kramnik has human assistance too. Do you think he got to be world champion without other people's help?

After all, when Kramnik sees "1.d4," he doesn't say "Hmmm...there's a novelty. What shall I do?" He is relying on centuries of opening theory that he has memorized, and many years of his own training. The computer's opening book is simply the electronic equivalent of that. And Kramnik is clearly not working alone. He has seconds to help prepare lines for him.

I did hear two pretty good ideas. First, let Kramnik bring his own opening book into the room. He wouldn't see what Fritz has prepared, but he wouldn't have to rely on memory either.

The other is to play Fischerandom, so that both sides would have an equal amount of opening theory at their disposal: namely, zero.

Just like Kasparov's self destruct against Deep Blue by playing a losing book line in the Caro Kann, Kramnik makes an obvious blunder. These mistakes are made to keep these non-event high dollar extravaganza's close and to keep the money rolling in into the future. If the human were to slaughter these machines the GM's wouldn't get a return match for years. Keep the score at + or - 1 and these non events will continue.

"Yusupov said that the knight on f8 is unusual for the mating pattern, and the knight's head turned on to the left could force Vlad to overlook the mate! According to GM and professional psychologist Helmut Pfleger (not sure the spelling is correct), this have some ground; he also pointed out that Seirawan in his comments made a similar mistake."

Good one, Vlad. Some truth in it. I'm a lefthander and make a lot of those faults myself, not in chess, but in normal life (ever tried to close a water tap? Not so easy for lefthanders since you work in the wrong direction.
Since Kramnik is a right hander (thus in the book) it is a bit surprising. I agree with Jussopov that the position is unusual. The mating thread on h7 is not clear in the first moment, but one should "get" it if looking some seconds towards the scenery. Kramnik did not do it and sent the game to the dustbin. Obviously Kg8 saved the game easily since White cannot mate with queen and knight without making "silent" moves which in the meantime would give black chances to attack himself the white king.
I think Kramnik saw he was in a winning position and tried to play "good" instead of solid which would have given him the full point.

Someone has got to have a video of Kramnik playing Qe3 and the audience's reaction. I would kill for this video.

Yes Shepherd, let both players come to the match with a floppy disk containing their own opening book up to an agreed given move. This would level the field. As it is now Kramnik gets to choose what is more advantageous for him from what Fritz team has to offer, from what they have prepared for Fritz.

Up to now Kramnik has done reat in making such choices. He got almost perfect postions in both.

I thought it has been interesting that Kramnick has been in control both middle games so far, at least equal if not pressuring and winning slightly.

After the first game with Kramnick playing white, and seeing how close everything was, I know some humans feared the worst today as Fritz wielded white. But today, Kramnick was again safe in the midgame, and better at the end, until his last move, of course.

This is better than I thought he'd do, and I'm impressed. Surely this human is performing better than Adams did.

tjallen

What if Vlad got his health problems back?
I hope he just got tired, because in both games during approx 3 hours his play was almost perfect, and after that he missed the target.

More on the striking resemblance with Kramnik's blunder in 2002 against the same opponent, as well as the famous Petrosian and Reshevsky blunders, over here: http://www.doggers-schaak.nl/?p=615&lp_lang_view=en

To clear this up once and for all, Kramnik was NOT given the book Fritz is using in Bonn. This request was met with "what, you want us to give you our preparation?!" Of course no self-respecting GM could say yes to that! As has been explained already, what Kramnik sees on the screen are the moves for HIM that are in Fritz's opening book because Fritz makes its own book moves instantly. That is, the list of moves and probabilities represent what Fritz is expecting. (For example, today Kramnik would have seen that Fritz consideres 3..b5 a "no play" move rated as a mistake.)

The statistics he sees show how many games the book has evaluated with the moves on the list, which give an idea of the depth of the line in the book. The probabilities of Fritz playing a given move are weighted according to these frequencies. So if move A has been played 16 times and move B 10 times, Fritz would play move A 60% of the time. But that's not really relevant because the moves on the list are for Kramnik to play, not Fritz. What does matter is that he knows in advance if his move is in the Fritz book or not, making it much easier to safely push Fritz out of book at an opportune moment. But it doesn't help him avoid any novelties prepared by the Fritz team or to know what Fritz will play in response to his moves.

The ChessBase guys in Bonn tell me Kramnik hasn't spent much time looking at the screen yet, although they feel he has "clearly outprepared us so far." He played the first 17 moves today at speed.

Kramnik outplayed DF in both games, no questions about that. His blunder with 30 minutes on his clock is unbeliveable. Kasparov in his match against 3D Fritz also blundered in a much superior position. I suggest humans should have the right to one take back per game when playing such matches in order to avoid blunders like the one we saw today. What do you guys think? Would this reduce the appeal of the match?

This getting close to Nimztowitch's observation that he never lost a game when in good health.

in his interview in "Der Spigel" Kramnik said :

"On his health: His doctor wants him to swim a lot. "So I plough through the water every day, for two kilometers. But honestly I hate it. Swimming is so boring!"

2km every day !, he is a sick man indeed..and btw, look who's talking about being boring, lol

Ed: yes.

Charley,

I am not sure. I would be OK with it and I believe most knowledgeable chess players would agree with that. The problem would be to explain this to the general public. Frankly Kramnik's loss today was simply not fair since he showed much more understanding of the game against a machine that calculates 10 million positions per second, therefore a take back rule looks reasonable to me.

Mig wrote:

~~~~~~~~~~~~~~~~~~~
...I refer to his ability to know in advance all the options at Fritz's disposal without having to remember anything other than the ones he wants. With such a guide and limited probabilities it's almost impossible to get an opening position he doesn't want to be in.

~~~~~~~~~~~~~~~~~~

I'm rooting for Kramnik, but I must agree with Mig on this one.

~~~~~~~~~~~~~~~~~~~~~~
...that is, he has little fear of a novelty by his opponent (at worst he sees it first) but he can prepare his own with near certainty his opponent will be "surprised." That is huge.

~~~~~~~~~~~~~~~~~~~~

So big a handicap, any Kramnik victory is meaningless.

~~~~~~~~~~~~~~~~~~

...pick a line with relatively few options, like he did today, and he can be close to certain to get a position he has already analyzed with the engine at home. That's usually going to be more important than playing the theoretical best.

~~~~~~~~~~~~~~~~~~~

It's the biggest proof that computers are deemed stronger by the top players: even the World Champion is asking for a handicap.

~~~~~~~~~~~~~~~~~~
...I just told Kasparov about it; he wasn't watching. Not much of an opinion, really. He considers the whole match a joke because of Kramnik getting the engine in advance.
~~~~~~~~~~~~~~~~~~

I agree: the match is a joke under the current conditions.

So he saw nothing at all of the programme's opening book before the match. And during the game he can only see what it thinks of what he's about to play? And according to the Fritz camp he's spent very little time looking at it anyway.

Much ado about nothing, I suggest. So he may get the chance to take it out of book - whoopee doo. We can all do that; just takes bad moves.

You're right for sure that seeing the engine is more useful (less the hash tables, whatever those are). But I thought this was fairly normal in these matches, no?

Like I say, it all depends what you're trying to investigate. If you want to know whether humans or computers play more strongly (not an interesting line of enquiry, in my opinion), then fine - black box. If you want to know whether there are still certain areas where humans can play better and which they are (more interesting to the manufacturers by far, I suspect, and also to me) then the more chance humans have to direct the game whither they want, the better.

Kramnik outplayed DF in both games, no questions about that. His blunder with 30 minutes on his clock is unbeliveable. ...

-- Posted by: ed at November 27, 2006 16:33

Now Kramnik knows what Topalov felt like after Game 2 in their match: better in both games but trailing in the match.

Instead of being ahead 1-0, Kramnik is down 0-1.

I just love hearing the Kramnik supporters wail "But he was better in both games; he's outplayed DF twice".

Ha ha ha. They mocked us Topalov supporters when we said that after Game 2 and now they're singing the same tune!

LOL

Jeez, Victor, get a grip. Can't you read? He hasn't seen the opening book until today, and if Mig's right he can't see the moves Fritz is going to make anyway, just what it thinks of the moves he's about to play.

You see what I mean, Mig? You'd have done better to present the facts instead of your views, especially since your views bore no relation to the actual facts, because most of your readers, will go away with completely the wrong picture. At least you'd have been better to do that unless you merely wanted to smear Kramnik, and as you've assured us several times you'd never want to do that.

Maybe if you weren't such an insulting jerk to everyone you'd get better results with your requests. Amazing how that works.

As for me, I presented both facts and my opinions.

As far as I can see my first five or so requests were perfectly polite, Mig. Sure you presented the facts - in the end. But meanwhile the likes of Victor have run off into the night thinking, for example, that Kramnik gets to "know in advance all of the options at Fritz' disposal without having to remember anything in advance other than the ones he wants" (quoting you, I believe). That statement (attributed to you; I haven't checked whether it's correct) bears very resemblance to the facts as we now seem to have established them.

I'm sorry you think I'm an insulting jerk. What's a nice polite way of saying 'irresponsible journalism'?

Going back to game 1 - are you going to share the outcome of your/Kasparov analysis of the endgame Mig?

I don't see Victor having the wrong picture. Other than the part about seeing Fritz's novelties, it's accurate. His opinion of the same facts differs from yours, obviously, as does mine. Not seeing Fritz's moves doesn't mean he doesn't see the options at Fritz's disposal when he can see which moves will and won't push the computer out of book. Seeing the options in the computer's book is huge even if you only use it to tip the machine out of book at will. As others have pointed out, imagine seeing a list of the moves your opponent has studied on every move. You can pick one, or play something else.

On a related note, it's not as if they build these books from scratch. The book they used in Bahrain was 95% the same as the book that shipped with the commercial version with most of the changes being deletions to avoid drawish stuff. Making a new book would be incredibly risky. Kramnik looked at the commercial book extensively in his 2002 preparation and we can assume he did the same here. Today, for example, the line we saw was the same in both the commercial and Bonn books.

I don't see why it would be silly to ask that Fritz gets no human assistance at all DURING the game (i.e, no opening books, no tablebases).

If you think it's fine to give Fritz human help for the opening and the endgame, why not giving him human help as well in the middlegame? Where is the difference?

One could as well argue that it's unfair that Kramnik understands middlegame ideas that were developped throughout history, whereas Fritz doesn't because he cannot even read books. So, it would only be too fair to have a GM helping Fritz to pick up the lines that actually make sense in the middlegame. That would be fair.

If we want the match to be fair, we could argue that it's unfair that Fritz can think 8 millions moves per second, while Kramnik can only think at best of a few. Why not adress this point as well and restrain Fritz?

My point is that this idea of having a fair match between a man and a machine is nonsensical in the first place. How would you set a 'fair' match between a Formula1 car and a human sprinter?

If a Man-Machine match made any sense at all, we would be able to find a consensus on 'fair' conditions, wouldn't we? I think it's time to give up the idea of such matches. Today's blunder is a very clear illustration of how absurd this man-machine match concept is.

cadlag said: "If you think it's fine to give Fritz human help for the opening and the endgame, why not giving him human help as well in the middlegame? Where is the difference?"

The difference is that its help in the opening and endgame is completely static and hands-off; the information has been made available to the program and now it is up to it how to use it.

Middlegame help, on the other hand, would require human intervention in the middle of the game, unless you're thinking about some other kind of help that I'm not considering.

cadlag,

Agreed! That's the reason I believe a take back rule would be fair for such a match (say one take back per game). Kramnik showed much deeper understanding of the game today and yet lost due to stupid one move blunder. Blunder today was probably caused by fatigue. However we are trying to measure chess understanding in such matches, not endurance, thus my take back proposition.

If you think that being able to see all of his moves which the computer has a book reply to do is the same thing as 'seeing all the options at Fritz's disposal', Mig, then we do indeed differ, not so much in our interpretation of the relevant facts as in our understanding of what words commonly used in the English language might mean. At least we can agree that Fritz could have had the most devastating novelty ever invented ready for 3...b5, and Kramnik would have had no inkling of it other than the fact he would have seen that the machine had studied the move (which is what he did see).

The fact that the programme is commercially available and, as you rightly say, is likely to have the same opening book in the commercial and match versions with only a few small changes, makes this whole matter even more irrelevant, I should have thought.

Cadlag: I agree; 'fair' is a stupid notion to use. Either the match is interesting for what it teaches us about computer chess and human chess, or it isn't. The organisers evidently feel the present rules will achieve the greatest level of interest in, and arising from, the match, and I suspect they're right. Talking about the match being a joke for this reason is a bit childish, whether you're the greatest player in history or not.

A free takeback? Please say you're joking. If there is indeed some merit to this idea of psychological issues when playing against a machine, it is up to humans to overcome it, not be pampered by a get-out-of-jail-free card. You really think most knowledgeable chess players would support it? I think almost all, especially GMs, would consider it a degrading, laughable notion.

No I am not joking. Is there anything more degrading to a World Champion than to lose a game like Kramnik did today? Kramnik played better, showed much deeper understanding of chess and yet lost due to stupid one move blunder. His opponent does not suffer from fatigue, does not feel psycological pressure, has unlimited access to opening books and endgames tablebases. The same happened to Kaspy in his match against 3D Fritz a couple of years ago. What this tells me is that humans are losing because of fatigue and/or psycological pressure, and not lack of chess understanding. Therefore a takeback rule would make sense to me and I would support it in such matchs. I dont see anything degrading or laughable about it. We are just levelling the playing field.

Out of the question, ed. There's nothing degrading about what's happened today. World Champions do blunder horribly. In addition to the ones mentioned previously, Capablanca himself allowed Qa4+ winning the bishop on b4 against Saemisch (or was it just Qa4, and the knight on c6 was lost?).

http://www.chesspro.ru/_events/2006/fritz7.html

Here's some stuff about the press conference. I don't have the confidence in my Russian to say what explanation he gives if any - I have the impression Ovidiu was right and he just says he thought he was winning and didn't see the mate, but I'm sure someone more capable can translate/summarise it properly.

Suppose they draw the next 4 games uneventfully. Final score will be 3.5 vs 2.5 in favour of DF. The general public will once again believe comps outplay the best GMs. And yet for a knowledgeable chess player it is clear that Kramnik is much superior in his understanding of chess based on the firt 2 games of the match. Therefore, as a knowledgeable chess player myself I dont see anything wrong with a takeback rule in a man vs machine match.

I don't think allowing takebacks would be a solution either. Nobody would ever agree on the result of a game. If Kramnik had been given one takeback today, he might have been able to eventually win the game and we would be arguing forever to decide who is the true winner of the game. A game should have a clearcut result.

"That's the reason I believe a take back rule would be fair for such a match (say one take back per game). Kramnik showed much deeper understanding of the game today and yet lost due to stupid one move blunder."

Sorry, but that's how chess works. There have been plenty of human games in which the clearly better player lost due to one stupid blunder

Kramnik losing a winnable game today by a simple blunder tells us nothing we did not already know.
Humans occasionally make mistakes, and one mistake can ruin an entire chess game.
It would have been more interesting to see the game reach its natural endgame conclusion.

So again, I hope next time the human gets to choose a teammate. The teammate can act as an advisor, and as a blunder checker. This advisor could be any IM, it need not be another expensive GM.

The growing strength of computers has revised the most interesting unanswered question remaining:
Can an affordable chess computer defeat a TEAM of humans?

"I don't see why it would be silly to ask that Fritz gets no human assistance at all DURING the game (i.e, no opening books, no tablebases)."

Fritz is not getting human assistance DURING the game -- the opening book is prepared beforehand. Obviously Kramnik also has a "book" (in his head) of openings he is prepared to play. He is not allowed to get assistance DURING the game, but he has plenty of advisors who work with him ahead of time. Would you ban such advice?

I'd have thought correspondence chess was the more obvious last redoubt for humanity. Not very sexy in PR terms, I agree.

jcm

rdh:
Thank you for bringing out the facts. Keep the good work.

On the subject of the opening book, Kramnik has one additional advantage. Quoting from the rules:

"During the match, the opening book may not be modified, except that up to 10 ply of additional moves may be added in the opening variation of the game which has most recently been played ... and the weightings of specific moves may be modified so that the different variations, already present in the opening book, will be preferred by the program."

Here again, imagine if a human opponent were told that, once the match begins, he is not allowed to expand any of the opening lines in his repertoire, except in the variation most recently played. Obviously this would be impossible to enforce in human play, but it's yet another limitation on Fritz that would be absurd to demand in any other setting.

The only question we're answering here is, How much of a handicap does the world's best human player require, for him to have a chance of winning?

The game today had a clearcut result. DF won due to one move blunder by Kramnik. So what? I dont see this match only in terms of the final score. I would like to see the winner show clearly a better chess understanding than his opponent. In this respect I am not impressed with DF. Even though he is behind in the score Kramnik has outplayed comp by a large margin so far.

Marc,

In answer to your question: all that is needed is a takeback rule. One takeback per game.

rdh wrote:

Jeez, Victor, get a grip. Can't you read?

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Yeah, sure, RDHY. I can read. And I can see, too.

Based on what I read here from people like you, I can see that you'd like to make the conditions such that Kramnink (or any human) can beat the computer. Today, it is "take the opening book away"; tomorrow it will be "have it running on a slower processor", then it will be "don't let it calculate more than 7 moves ahead", etc., etc.

Somebody even suggested allowing take-backs!

Let's face it: Kramnik is not playing the computer on even terms. Even terms means each player prepares in secret and they come to play without having to disclose their preparation, knowledge or thinking process. (Remember Topy-Kramnink?)

If we have reached a point where top GM's believe that they can't beat the computers without crippling the machines, they should not play any matches against computers, the same way we don't race against cars. Pretending that we can "beat" the computer by getting a handicap is (or should be) quite demeaning for a top player like Kramnik.

It is my opinion that there NO way Kramnink can claim a legitimate win in this match.

Ed is absolutely right. What matters above all is chess understanding and I do not think that there is anyone who would argue that DF has had the better of that so far. No one-move blunder can change that.

An alternative to a takeback rule
would be a "hand calculator" for
tactics. It shouldn't be too hard to
write code that only checks for
simple mates, leaving pieces en prise,
or some reasonable laundry list of
"stupid human errors". The calculator
wouldn't flag any "strategic errors".

Then the games would be more reliably
decided on strategy; the analogy being
that being a good mathematician doesn't
have a whole lot to do with how well
you add and multiply.

Surely (some form) of this idea has
floated around, but with losses like today,
maybe it should have some support?

Also, it is quite poor of Kasparov to dismiss the match as a joke on the ground that Kramnik had access the programme in advance. How can we see whether computers have a deeper understanding of chess, unless we have the opportunity to test them where we feel that they are weaker?

"Ed is absolutely right. What matters above all is chess understanding and I do not think that there is anyone who would argue that DF has had the better of that so far. No one-move blunder can change that."

I agree that Kramnik has shown superior chess understanding. But I disagree that this is "what matters above all." Whether your opponent is a computer or a human, what matters is the moves you play on the board.

Blunders happen in human-human games too. No one has demonstrated that this particular blunder was caused by the extra pressure of playing a computer. Although this blunder was egregious, Kramnik has blundered against human opponents too. All a "take-back" rule would do is to bring further discredit to the match. Either humans can beat computers straight-up, or they cannot.

The whole issue is the opening book part which taints this match badly.
It allows Kramnik to "pick up" the opening, to steer the game into positions where the comps, any comps, are weak.

He has done great so far but what's the point ?

Fritz plays as comps do in such postions and Kramnik blunders as monkeys do when a bit tactics eventually happens to appear.

We have known all that since many years.

Following from Emmanuel: Especially since, if my memory of Mig's ICC commentary serves, it wasn't for lack of Kasparov trying to attain the same sorts of things in his own matches with computers.

Ok, most agree that Kramnik has had the better of the play (is it virtually certain he was winning today BTW?) and you've got to feel lousy for the guy, but the takeback idea is still ludicrous. People seem to be piling on the bandwagon of handicapping the machine under the guise of "leveling the playing field". Ok, whatever floats your boat. But I fail to see how takebacks have any place in serious a chess match. Do you even realize the monumental breach of protocol and ethics you're proposing Ed? No GM with an atom of respect for the game would ever consider it.

Marc, Fritz has access DURING the game to some external help in the form of an opening book and tablebases. Kramnik is not given access to such external help DURING the game. If you don't understand the difference, I'm sure you won't mind if you're opponent uses Chessbase 9.0 next time you play a game in a tournament.

In any case, you missed my point. I was not proposing new solutions to improve the 'fairness' of man-machine matches, I was pointing out that these matches are absurd precisely because they can't be fair.

ed, in terms of understanding chess, Kramnik is of course way ahead of the silicon beast. Fritz is just a monkey playing chess nearly at random, with a ruler and an incommensurable amount of time. He also has an opening book and some tablebases so that he can pretend he understands what he is doing...

You worry too much, Victor. If Kramnik doesn't feel demeaned, why should we on his behalf? Computers play better than us. Face it. If we need a handicap to make matches with them interesting, what of it? I don't see that as being demeaning. Of course Kramnik can claim a legitimate win in this match, if he wins. He can't claim he was stronger than the computer on equal terms, equally obviously, because that's not what the match is set up to test.

Sure I'd rather have a handicap to keep it close. Who wouldn't? You enjoyed Adams-Hydra, maybe?

But let's at least state accurately what the handicap is. If the information we have is correct, it is nowhere near as great as you, Mig, or Ovidiu would like to suggest.

I wonder what the purpose of not allowing changes to the book during the match was. I don't see much point in that if Kramnik hasn't seen it beforehand anyway.

I'm not sure there is any such thing as chess understanding, as opposed to strong moves. The merciless fact, culminating in checkmate, contradicts the hypocrite, and all that.

Given that of the ten worst blunders committed in history by world champions, round about four of them have been in computer matches, I think it's pretty clear that humans are more likely to blunder against computers. Although to be sure Kramnik has been fairly fallible against humans too lately - remember him leaving a piece en prise in the Russian final (I forget who against).

Kramnik at the press conference: “It was not actually only about the last move. I was calculating this line very long in advance and… I don’t know… It’s actually very strange, some kind of black-out. Because I was feeling well, I was playing well before and… I think I was pretty much better… and I calculated this line many many times, re-checking myself, and in all this Qe3 and Qxe4 I was calculating Ng6+, and okay, I already did it when I was playing …Qa7, I already calculated this line. After each move I was re-calculating it again. I was thinking it’s all fine, then again and again and finally I blundered it, mate in one. It’s really actually the first time it happens to me and I cannot really find any explanation because I was not feeling tired, I was feeling well. I think I was calculating well. It’s just very strange. I cannot explain."

Thank you, Cynical. I think you are right (and fervently hope so).
And ed: I am not by any means saying you are not a knowledgeable chess player (I wonder whether I am one), but I think Cynical is right here, and you have a misconception of what chess players I consider knowledgeable think.
I do not want to disparage you in any way, please note.

To Marc Sheperd,

when you say that Topalov must be thinking something like "why didn't he give ME one of those?"...

It's true that today Kramnik did give a mate in one, while Topalov had only a mate in three at his disposal.

We all commit blunders. But this one was awesome. Being rated 2200 myself, I've played between almost 300 classical games, and I can't remember blundering more than a piece in 3 moves...

Next time you play 1 min games on playchess.com, even when you play the most horrific moves, don't feel sad : the world champion himself can do even worse.

> I think I was calculating well. It’s just very strange. I cannot explain.>

Poor Kramnik, he seems disoriented, he was hit from nowhere and made KO. Now it is time for that French lady to use her charms and restore our hero to self confidence.
It would be disaster for Kramnik if he loses faith and begins playing without nerve, "Fritz punishes you immediately" as we saw.

Here's a neat coincidence. Kramnik's blunder in this game has one thing in common with his blunder against Deep Fritz in Game 5 of the 2002 match: both blunders happened when Kramnik offered a queen trade. I wonder if Kramnik's anti-computer strategy has affected him to the point where he relaxes as soon as the queens are off the board.

Allowing takebacks just seems a terrible idea - not because the human gets insurance against error, but because it can be abused to great effect. At a crucial juncture, when the human is unclear about where a line beginning with a certain move leads, he can punt a guess and gauge its "goodness" based on the response.
I'm not sure what the hand calculator idea is getting at either - if its main aim is to provide trivial checks against mate/material loss etc, the GM can surely do just as well in short time without any assistance. I'm guessing that errors like Kramnik's today come more from complacence and *not* bothering to perform a sanity check before making the move. And that's the equivalent of not bothering to punch in your move on the hand calculator before making it.
It is probably a better idea to equip the human with more time during the endgame - the amount of time could again be dictated by how long a tablebase unaware program takes to converge to the definitive line given in the tablebase. That seems more like a Man v/s Machine metric. And I'm totally against letting the human have *any* sort of insight into the machine's opening book. Why? Does the GM divulge his opening prep to the DF programmers?

Dear bs,

> I'm not sure what the hand calculator idea is >getting at either - if its main aim is to provide >trivial checks against mate/material loss etc, the >GM can surely do just as well in short time without >any assistance.

Your point is interesting. I would say that perhaps
the human would no longer make the move directly on the board but would type in his move into a calculator, which would then make a check. If it
find something, it blares out "Kramnik, you are
about to get mated in one, see Qh7++", otherwise
it says "Looks good". Either way, it concludes with "Final answer?" asking the human to confirm his/her move.

This certainly eliminates the question of whether the human would use the device (by this set up, he has no choice).

I agree that takebacks are rather silly. In fact, the motivation for them is ill-thought. No-one benefits if computers offer humans any kind of odds. In fact, Kramnik's performance so far shows that humans still have a better grasp of the game. A couple of bad moves cannot change this.

Having access to the program in advance is a great advantage, that practically cripples the machine. Kramnik's team will have surely played hundreds of games in certain lines to get into dry, predictable positions where there's not much chance of the machine complicating things.

I think Kasparov has a point in calling the match a "joke" (even if he, too, tried to get some advantage himself).

Victor, you are missing the point. You are asking what use the match is if Kramnik can steer the course of the game towards positions he likes. The whole point is to do exactly that, in the same way that one would try to get the positions one likes against any opponent! What's strange in gaining advance knowledge of the kinds of position that your opponent is weak in?

Kasparov was a pioneer in computer matches (as in so many things), but he is clearly wrong in trying to discredit the match or Kramnik's or DF's playing strength.

Video of the start of game 2 and about 4.5 minutes of the post game press conference (in English) is up at http://www.chessbase.de/. He addresses his blunder in the first question.

Vlad's English is pretty good, so you can fast-forward to about the 7:00 minute mark if you just want to hear his reactions.

How many Brits are weighing in on this? This sort of hysterical overreaction to what is really not a hard concept to understand (Human Being Makes Mistake) reminds me so much of the British (well, English) press, in connection with pretty much any sporting event. Every piece of good fortune is the most awe-inspiring portent of glories to come; every error is the most daming of confirmations of how awful things were all along.

Honestly, do you expect Kramnik to be crushed by this? What has he got to beat himself up about? He made the kind of awful oversight he'll probably never make again in his life. By definition this is a singular event. What is he supposed to learn from this? What thought processes is he supposed to correct? What deep flaw in his approach to chess has been revealed? (Note to self: don't make mistakes if at all possible!)

Kramnik's already shown in his match with Topalov how incredibly resilient he is mentally. I don't see why this freak slip should cause him to crumble. Get a grip, people!

Well said, theorist...

>Honestly, do you expect Kramnik to be crushed by this?>

We don't know, we will have to wait and see.
In the Elista match Kramnik was helped by Topa's blunder in game 10 to get back on track.
But here ? Fritz doesn't blunder.

I for one can see Fritz dully winning the next
game against a Kramnik disheartened by this emabrassing counter-performance.

Kramnik was too distracted thinking of an excuse to go to his rest room. (hence the cup pick up) I am sure if he had gone to the rest room before making his move, he would have done fine.

(please note: there is heavy sarcasm in my fingers as I type this)

Well, it just seems that Kramnik obsessed on getting the queens out of the board... to the point that he momentarily made it the only goal to achieve, tuning out other small details, like, you know, getting a better position or just... not getting mated.

Maybe against an human player, he wouldn't have obsessed so much about exchanging queens.

Chess book reviewer Taylor Kingston once wrote that the highest quality of human chess occurs in correspondence games, not OTBoard. But is that true?

I can see how it might be true. A human who has a couple days to work out the best move could produce excellent moves.
Yet the chess experts who gravitate toward playing correspondence chess tend not to be the world's very best chess players.

So which outweighs the other, the player's chess talent, or the extra time provided by correspondence?

Hydra whipped M.Adams over the board, I think 6-0.
But correspondence player Arno Nickel defeated Hydra 2.5-.5 (see amici link below, you remove spaces).

*** Does the fact Arno had a better result against Hydra (than did Adams) prove Taylor Kingston was right? ***

http://amici.iccf.com/
issues/
issue_05/
issue_05_nickel_hydra.html

Gene Milener
http://CastleLong.com/

According to Chessbase, Kramnik thought for a couple of minutes before playing 34...Qe3?? and he had over 30 minutes in his clock (and would have gotten an additional hour after reaching time control at move 40) .

Charley and others,

I am not discussing who is a knowledgeable chess player and who is not. This is beside the point. My point is that knowledgeable chess players have a profound appreciation of how deep and beautiful this game is. This appreciation is I believe associated with how deep certain strategical and tactical ideas are. Well I dont see anything deep about a one move blunder, hence the idea of a takeback rule in man vs machine matchs. Maybe we should settle for the chess calculator idea, I dont know. What I do know is that so far the score of DF vs Kramnik match is misleading, especially for the general public. The general public is likely under the impression this match will be just another win of a computer vs a top GM whereas a knowledgeable chess player knows as a fact Kramnik has shown much superior understanding of chess so far but is behind the score due to an accident. A takeback rule would reduce considerably the psycological pressure on the human player; therefore I would expect it to be rarely used. This is as it should be as I believe blunders are either related to time pressure or associated with some sort of psycological pressure or fatigue. If the idea of a man vs machine match is to measure chess understanding then I would support a takeback rule. If not then leave it as it is. Anyway so far Deep Fritz has not left a good impression on me. It is a patzer, very strong tactically of course, but still a patzer. I may change my assessment depeding on the outcome of the next 4 games but so far this is how I see things.

i don't know what's so crazy about kramnik having the prog in advance. he's still playing middlegames and on against one of the strongest softs available comercially. and he does seem to be playing better chess than fritz, even thou he blundered ridicoulously.
i believe kasparov is the best player ever, the one who understood all phases of the game in greater depth, and i also think he's a very sore loser who tends to overstimate his rivals once he's beated them and does the contrary when he was in the losing end (he speak marvels of karpov, but says he lost his last profesional game for he was unfocused, not because topalov was stronger at the time)

This is Karma for Kramnik because what he did during the match against Topalov. Without Fritz's helping Kramnik cannot play long fighting's game.

Kramnik surely had the program in advance last time, didn't he? At any rate it was said that in his preparation he'd discovered a weakness in its positional evaluation of certain positions which he was then able to steer for. I don't know how he did that if he didn't have the program.

This blog is a fascinating example of the power of journalism. You don't realise how dim the masses are until you actually see in action how easy it is for Mig to manipulate them.

How about some more facts? In previous matches, hsa the program been made available to the human before, or not? Did Adams have any insight into Hydra? (I suppose not, since as I understand it needed special hardware.) Or Kasparov into Deep Blue, Junior, or 3D Fritz - indeed were the latter two any more than the commercial programmes? Or, say, Hubner into Fritz, or Smirin into the various programmes?

There's not much point in people expressing an opinion about whether these rules are extraordinary without they know this sort of thing, as far as I can see.

Ever consider doing some reearch to answer your own questions, rdh? Everything you've asked is answered several times on the web somewhere. Why you should think that a journalist has the job to fill in whatever you are ignorant of? A journalist must assume a certain level of informedness about the subject and his readers, and those who don't know about the subject, and won't go look it up themselves, are justifiably lost. Rather than rant about what you perceive as Migs failures, why not just look it up yourself? Like, google, man.

It’s not so much filling in the gaps in my ignorance I would like (though it would be nice and would make this a better blog), as refraining from making comments which exploit the ignorance of the general public in the way Mig’s been doing here. Journalists wield a certain power which I prefer to see used responsibly.

Moreover I assume – perhaps wrongly – that having covered all those matches Mig knows the answers offhand, whereas I doubt they’re quite as easy to find by googling as you suggest, and as I found out when I tried looking at the official rules of this match, are not easy to interpret anyway.

Hell, if you know the info can be found by googling, presumably it’s because you’ve tried it yourself. And if you’ve tried it yourself, presumably you have this information and could share it yourself if you wanted to. Go on, share it, why not?

But you make a fair point. I’ll give it ten minutes at lunchtime.

Thanks to rdh for his (indirect) investigative journalism. Your insistance has forced Mig out of the bush, showing that the rules of the match are far less absurd or unbalanced (or indeed not at all absurd or unbalanced) than rumor would have it. One might like them or not, that is another matter. By the same token, one might like or not Kasparov's idea of having his (colossal) opening book by his side.

The only dim mass around here is you, rdh. You are constantly demanding others answer the same questions over and over again and ranting instead of reading. You clearly have no interest in the answers and only wish to prove some obscure point by repetition, of course adding childish insults to every post. Many people here feel the rules are unfairly tilted. You disagree. This doesn't mean they are idiots or that they don't understand the rules.

As previously stated, and/or made blatantly obvious through context, Kasparov and Adams had no prior access to the game engines used in their matches. The Junior guys even changed the engine during the match (and included both versions on the next CD release). As for Bahrain, this is what David Levy said in my interview with him before the 03 Kasparov-Junior match:

http://www.chessbase.com/columns/column.asp?pid=158

"As to the main differences in the rules between New York and Bahrain, these were necessitated by Kasparov's desire to avoid either side having an unfair advantage. For Bahrain the rules were, to be frank, rather human-friendly. The program code had to be frozen several months before the start, with no changes being made thereafter. And Kramnik was given a copy so he had the luxury of being able to practice against the exact same version of the program against which he played. He even had an opportunity to play against Deep Fritz on the same hardware used for the Bahrain match. All this allowed Kramnik to test the strengths and weaknesses of the match version of the program, a far cry from what Kasparov went through before he played Deep Blue in 1997."

Found it with Google.

The moment anyone says the rules are absurd he rattles on about how horrible Mig and this blog are, as if these things are connected. There have been links to the rules and discussion of them going on for MONTHS.

I will repeat: the rules are absurd. Knowing when you can take the computer out of book is a large and wholly unnecessary advantage. It doesn't balance the playing field at all; that would be giving Kramnik access to his own book. (Or limiting computer book access in some way.) And giving the exact engine is also absurd and unnecessary. That he also got the same concession in 2002 doesn't make it right. And yes, I criticized it then, too.

The problem is that we tend to measure to the results. So the draw in Bahrain made it a non-issue to most observers. And even with these advantages he's down a point in Bonn. But that doesn't mean they are the best way to go about it or fair just because he's not winning. When these things are close to invincible I could see giving such handicaps - after all the point could be discovering any weaknesses at all. But since Kasparov drew two matches without the engine it just seems unnecessary and weird.

I'd rather find ways to level the field without sacrificing the integrity of the man-machine experiment so much. Balloted openings (Nunn match style) for example, or both players having access to a limited book.

I just wish Kramnik had converted that first endgame win. That would have been a sensational victory and shown how humans can still flat-out outplay computers.

Mig, you are very smart. But the "dim masses" are not always as dim as you think.

You are now going back to another point (pre-match use of the final version, hence Kramnik's advantage over the Kasparov-machine matches). This is an interesting debate (on what type of advantage the human side has to get to balance the chances). Another suggestion that has been made is that the comp play without any opening book at all (also an interesting discussion). But all your shrewdness won't make us forget that rdh's discussion was about the exact degree of access to Deep Fritz's opening book given to Kramnik (both live and pre-match). The rules are not that easy to understand (at least for a layman) and rdh effectively made you ask your buddies at chessbase, with one of the results being that we now know the precise (limited) access Kramnik has to the comp's book.

So, again, thanks to rdh for his (indirect, i.e. through you the professional journalist) informations!

Ray Derivaz Hernandes

Umm, so, you're welcome? Ah, nobody thanked me, my bad. It's posts like these that make it so tempting to simply ignore the comments and put the lie to how I'm "forced" to do anything at all. And what I said had already been said several times in several threads. The only clarification was to confirm something someone else already said, which is that Kramnik doesn't have his own screen and that Fritz's moves go by too quickly for him to see its book choices. The rules say he has a screen, but they are sharing. Others figured this out by looking at the photos.

I cannot understand why some people are posting so agressively against Mig. He has very strong views about the conditions that Kramnik got for this match and to a slightly lesser extent for the match in Bahrain. Mig should be free to air his sometimes strong views without all these personal attacks. Similarly there was some complaint about Mig passing on stuff Kasparov had said which is crazy isnt every chessplayer interested in any chess comments Kasparov has made?? Having said all that I think Mig is wrong to regard the opening book and program access as -unnecesary and absurd- obviously they are necessary for Kramnik thats why he asked for them. As for the abusrdity I think the proof of the pudding is in the eating - so far I think the chess has bee extremely interesting in the first 2 games. If Kramnik had won both games relatively easily then there would be some validity in Mig's remarks. The event as an absurd preperation exercise of you like. A one sided event. But it has not been like that at all. In game 1 he missed a win but a very difficult win that few players might have found OTB at the time. In game 2 it looked like he had an edge but blundered badly and lost. Mig seems confident that these - unecessary and absurd- rules will ensure Kramnik wins some games. I am not so sure I think he is underestimating the physical and psychological pressure that Kramnik feels. It may be that with these type of advantages Kasparov would easily win computer matches we will probably never know but that possibility seems to be behind the rule criticisms. Meanwhile I am sure the future games will be interesting

Well, tjallen is a better googler than me; I can’t find anything about the rules for Hubner-Fritz or the Smirin match with three (?) programs.

As I suspected however the rules for the present match are the same as the last time (except the opening book point), or at least according to that other reliable journalist David Levy.

http://technology.guardian.co.uk/online/story/0,3605,817484,00.html

In other words the impression which Mig has been creating that having the engine in advance is an new and unprecedented demand of Kramnik’s is incorrect.

I can’t find the rules for Kasparov-X3D Fritz or whether the version he played was specially created or just the commercial one.

And as I also recalled Kasparov didn’t have Deep Blue in advance. As to Deep Junior, I can’t find the exact rules of that either – kudos to tjallen’s googling skill again – but here’s an extract from Gazza’s article in the Wall Street Journal of 16th February 2003

“Moreover, due to Deep Junior's public track record of nearly a decade, I can analyze hundreds of its games.”

He goes on – here’s the link –

http://www.opinionjournal.com/extra/?id=110003081

to say what a good thing that was.

Armed with this information I think we can see Kasparov’s claim that this match is a ‘joke’ for what it is. It could better have been rephrased as ‘I’m a bit worried Kramnik might succeed where I failed and be thought of as better than me in some respect and it’s important to get in there and rubbish him before that happens’.

What's with rdh's constant whinging and insulting behaviour? He constantly hijacks threads and starts a spitting war, with the only outcome being the clarity with which he is revealed to be a miserable little sod. Some of the choice tidbits he's let fall in the past include ranting at people for not spelling latin phrases correctly, and lampooning people for his own inability to locate and assimilate information freely available on the web. Mig must be surely tempted to exercise the Blogmaster's editorial prorogatove a little more freely..

Mig, what set this off was your claim that ‘Kramnik could just about pick the opening he wanted’.

So far as I can see, this is utter nonsense. The true position is that he can see whether or not the move he proposes to play is in the machine’s opening book or not. Naturally, the machine will have most well-known moves in its opening book. How that enables him to ‘pick the opening he wants’ to any greater extent than we all can whenever we play is beyond me.

Moreover, as you also pointed out, the book actually will deviate very little from the one in the commercially available program. Kasparov therefore enjoyed a very similar advantage anyway in his matches against Fritz and (I presume) Junior, although not against Deep Blue.

The rules may or may not be absurd or unfair. In my view the very idea of them being ‘unfair’ is ridiculous; it’s a match on certain terms which may be interesting or it may not and ‘fairness’ doesn’t come into it. But obviously you’re entitled to your view. It’s clear that previous competitors including your boss have felt that something ought to be done to even up the machine’s memory advantage in the opening stages, so the question is not whether Kramnik’s a big wuss for wanting that, as you rather imply, but whether they’ve chosen the best way to do it.

Actually, one point we do agree on is that what has been done about the opening book is rather absurd – you because you claim it’s a huge advantage which is typical of Kramnik’s weaselly attempts to win by any means and thus polish his image in front of the uninformed public at the expense of more deserving characters such as your master (I paraphrase a little), I because I can’t really see that it makes such a difference as to be worth opening up the event to disingenuous accusations like yours that I quoted above.

"Marc, Fritz has access DURING the game to some external help in the form of an opening book and tablebases. Kramnik is not given access to such external help DURING the game."

I continue to be amazed that people don't understand what an opening book is. It is simply the computer's stored memory of opening theory, along with its preference for playing/avoiding certain moves. Does anyone here NOT know that human players also 'remember' opening theory?

Obviously, the computer has the usual advantages: its memory is more capacious, and whatever is stored it will remember perfectly. But the basic idea is that both Kramnik and Fritz are sitting down at the board with very large amounts of memorized opening theory at their disposal. The idea that Fritz should figure it out from scratch, starting at move one, is just as silly as suggesting that Kramnik should do that (and clearly he does not).

Tablebases are a different story, because no human has memorized perfect play for every conceivable position of four and five pieces on the board, as Fritz has done. However, tablebases have not mattered in either of the games played to date.

"I dont see anything deep about a one move blunder, hence the idea of a takeback rule in man vs machine matchs.....The general public is likely under the impression this match will be just another win of a computer vs a top GM whereas a knowledgeable chess player knows as a fact Kramnik has shown much superior understanding of chess so far but is behind the score due to an accident."

The score of Kramnik-Topalov was also deceptive after two games. Aside from two blunders, having nothing to do with "deep and beautiful" chess knowledge, Topalov should have been up 1.5-0.5, instead of down 2-0.

Guess what: blunders count too.

"Anyway so far Deep Fritz has not left a good impression on me. It is a patzer, very strong tactically of course, but still a patzer."

Earlier versions of this "patzer" have drawn matches against both Kasparov and Kramnik. Any human who could do that would be considered one of the world's strongest players. I think you need to look up "patzer" in the dictionary.

Dear friends, it looks like some of you do not understand how SMP engines choose a move.
All multiprocessor engines are non-deterministic. This means that for the same position the same engine running on the same computer can choose different moves after spending the same amount of time or taking analysis up to the same ply count.
Even more. For exactly the same position (or the same move sequence) it can give slightly different evaluations.
Therefore, saying that Kramnik have a guarantee that Fritz will play an expected move is incorrect (except forced lines, of course), no matter how much time he spent on Fritz style evaluation.

"So far as I can see, this is utter nonsense. The true position is that he can see whether or not the move he proposes to play is in the machine’s opening book or not. Naturally, the machine will have most well-known moves in its opening book. How that enables him to ‘pick the opening he wants’ to any greater extent than we all can whenever we play is beyond me."

RDH, surely you would agree that, in a normal match, you do not have the benefit of knowing in advance whether the move you are about to play is in your opponent's preparation. The only dispute here is precisely how much of an advantage Kramnik derives from this knowledge. As this was a concession that Kramnik wanted, it's safe to assume that he considered it important---otherwise, why would he have asked for it?

"The rules may or may not be absurd or unfair. In my view the very idea of them being ‘unfair’ is ridiculous; it’s a match on certain terms which may be interesting or it may not and ‘fairness’ doesn’t come into it."

I agree with RDH on this point. The real question is not whether it is FAIR, but whether it is interesting from a sporting point of view.

Mig's comments on this thread are spot on, as they nearly invariably are.
Of course any person with an ounce of common sense can see that the advantage Kramnik has insisted upon re Fritz's opening book is absurd. But Kramnik has no morals or ethics as we already know. And this pathetic guy is losing anyway!

Thank you, Mig (and d), for your great, absolutely correct comments about rdh. You must be sorely tempted to ban wankers like him and Greg Koster who fill up your blog with so much useless rubbish, and hijack the threads.
Another favourite device of rdh is when cornered by a perfectly clear explanation is to claim 'I cannot understand what you are saying' implying what you have posted is at fault, when of course he understands perfectly well.
rdh claims he is a lawyer. Well, if true, he would be the worst arguing one I have ever come across. If I had to choose a lawyer, he absolutely would not be on my list.
rdh also claims to be a 2400 rated IM. From his posts, I find this very hard to believe. Perhaps rdh could give us his name so that we can verify this claim (and the one that he is a lawyer).

Of course I’d agree that it’s an advantage, Marc S: it can’t do any harm at least. I don’t see that it follows that Kramnik thought it was an important one. I imagine it was the product of a negotiation. It might merely have been the greatest concession which the manufacturers were prepared to make, which Kramnik’s team accepted even though they didn’t think it was important. But I would like to see it correctly characterised. Mig’s description of the possibilities it gave Kramnik was seriously misleading.

I wonder if someone can explain something else to me. I had assumed that the programmers ‘told’ the machine to play, say 1 d4, and to meet the Slav in this way or that way. But there’s a photo on Chessbase (http://www.chessbase.com/newsdetail.asp?newsid=3509, second one down) of the screen before the game. I can’t quite make out the headings, but I think they are N, %, Av, Perf, fact and then Prob(%). As I understand, N is the number of GM games in the machine’s book, % is how well the side to move scored with the move in question, Av I assume is the average rating of the opponents and Perf the rating performance of the players who made the move, considered collectively. I also assume that Prob indicates the probability of the machine playing the move.

So this would seem to suggest that the machine is in fact choosing its opening move randomly according to certain pre-set probabilities, and that I and most of us were wrong to assume that the machine will ‘avoid’ the Scotch, or whatever. Does anyone know if that’s right?

I also understand that while Kramnik may see this before the game (assuming the machine is White), the moment the machine moves what he will see is the same screen for his own moves. So it is misleading to say that he will see the machine’s ‘evaluation’, in the sense of whether a move is good or bad, and certainly he will not see any ‘opening analysis’, as some have suggested. He will see only which possible moves of his the computer has GM games for, the performance of the move in those games (which will only be the ones the programmers have chosen to put in) and what the probabilities would be of it playing each move, were it Black (which again is a number put in by the programmers, I assume, or at least he doesn’t know it isn’t).

Yes, yes, yes, Chris B. Easier to abuse me than answer the point. It's a short one.

Is it or is it not true that Kramnik can 'basically pick the opening', and if so how?

I think you will find that the words you and Mig are looking for are, 'It is not possible. That statement was simply wrong and I retract it.'

Mig, by the way, already knows who I am. But, to adapt a point made against me, if you want to know, get your google out and search for English IMs rated 2400+. That will narrow it down. Then google each of them and see which are lawyers. It won’t take you long; you’ll be able to unmask me in no time.

Vlad Kosulin:

"All multiprocessor engines are non-deterministic. This means that for the same position the same engine running on the same computer can choose different moves after spending the same amount of time or taking analysis up to the same ply count."

That's an interesting point. But I think the subtle point here is that we are talking about openings and endgames as dictated by books/tables, not middlegame evals. Now, in openings, the decision of a move would be, most intuitively, a lookup with a seed, since while still in the book, the engine need not crunch the position with its evaluation algorithm. And tablebase position has been reached during and endgame, it is *completely* deterministic from that point on, because the position has been "solved" already. On any modern processor, this should be a near-zero time operation. Which is why I was trying to suggest earlier that the human's clock could be augmented with the amount of time actually taken by a tablebase unaware engine to converge to the tablebase line. Well, whatever. :)
So, reiterating my point, openings and endgames are pre-evaluated and cached away in books, so the non-determinism factor is pretty much non-existent.

bs,
I was objecting the statement that giving Kramnik access to the engine in advance is not fair and lets him fully predict it's response to his moves.

To see the opening book and to have the engine before the match to test what the comp will do in a given position "out of the book" is a lethal combination as I will explain bellow.

It is not without reason that Kramnik demanded these two and it is for the same reason that he got excellent anti-computer postions (which he however failed to materialize) in both games so far and he got this while bashing out opening moves and left comp behind on time (unusual for such M-M matches).

We are going to see more of this for the rest of the match.

It works rather easily :

1. pick up a line (or a sideline even better to avoid novelties) and start look for favourable(involving postional judgement) deviations before theory ends.

2.Play out these deviations at home with the engine, of Fritz, to select [accept or reject] function of the result postion ( dull or not dull enough/sharp).

3. play the line on the board during the match- you can do it fast when you see that the comp still has your moves in his opening book until you get to the moment to usefully deviate and then the comp will just move as you expect ( as you know from home analysis) and you will get in a a good for you position

This has been the "engine part", but why do you still have to see Fritz's expected moves from you during the match, the "see the book"-part ?

Well, you don't know in the months preceeding the match what exactly Fritz's opening book will be.

Fritz may not have theory line, or have it but not deep enough to reach the deviation point, that you choose to just play out until the moment comes to deviate.
You still have to look at the screen to make sure that you are still in Fritz theory and it will go as you expect.

What if he doesn't have your line ? well, he may have some other so back to point 1. you have to pick up few such lines analyze to increase your chances to get what you as you go deeper.

With team of well paid GMs experts in given openings and the engine you can prepare rather well. Kramnik has done just that.

Meanwhile, from what I hear, Kramnik has hardly been glancing at the screen. He seems to be very well prepared notwithstanding the rules.

Interesting middlegame positions have resulted. The real chess play is here, and it is Kramnik who looks better.

Unfortunately, like a human, Kramnik falters and loses. A day later it still baffles me.

I understand blind spots and realize that Kramnik was pursuing a plan he had visualized, checked and rechecked. But I can't understand why, with Fritz's Queen and Knight in dangerously close proximity to Kramnik's king, he did not feel compelled to pause for a moment before playing the ill fated ...Qe3.

>Meanwhile, from what I hear, Kramnik has hardly been glancing at the screen. He seems to be very well prepared notwithstanding the rules.>

He doesn't have to do it much, not anywhere close to "starring", If my understanding is correct he does it only to make sure that Fritz expects the line in which he has prepared a later deviations.
He does it to be sure that he will not force the comp out of the book before the moment of his "novelty". After this novelty the comp will switch to engine and play what Kramnik has already analyzed with the engine at home.

Thanks to rdh for his efforts to clarify things.

And thanks to Mig: although his loyalties often tilt his posts, he doesn't censor opposing statements.

"So this would seem to suggest that the machine is in fact choosing its opening move randomly according to certain pre-set probabilities, and that I and most of us were wrong to assume that the machine will ‘avoid’ the Scotch, or whatever. Does anyone know if that’s right?"

As I understand it, the retail Fritz comes with an opening book with move probabilities comparable to real-life GM practice. For this match, the Fritz team has altered those probabilities to choose lines in which they believe the engine will have favorable chances against Kramnik. For instance, I'm sure they are avoiding any lines where queesn are traded early. They could also very well have set the probability of 1.e4 to zero. The rules say that the Fritz team can change move probabilities during the match, so it's very clearly anticipated that Fritz won't be playing with the same probabilities that the retail version uses.

"Dear friends, it looks like some of you do not understand how SMP engines choose a move.
All multiprocessor engines are non-deterministic. This means that for the same position the same engine running on the same computer can choose different moves after spending the same amount of time or taking analysis up to the same ply count."

This isn't quite the full story. When the engine is still in-book, its moves are determined by probability weightings. Once it is out-of-book, there are a variety of reasons why it might not play the identical game each time. But if Fritz evaluates a particular move as superior, it is guaranteed to play that move, unless a deeper search refutes it.

So, in his preparation, Kramnik can see how Fritz is evaluating certain lines, and he can steer towards positions where he knows the engine tends to get the valuation wrong.

So, Ovidiu, how is that different from normal opening preparation? Say Kramnik wants the line that occurred in game 2. He can’t make the machine play 1 d4 or 3 e4, obviously. But say it does. We have already established that he has no way of knowing what its reply to 3…b5 will be: all he will know from the screen is that it has one. He can look at the commercial version of Fritz if he wants (just as Kasparov could have done, unless anyone knows different) and gamble that the opening book will be the same, and in that case he knows he’ll get up to either 9 Nf3 or 9 Ne2. And once 9 Nf3 hits the screen, he can tell that the computer has no prescribed reply to his proposed 9….a5.

So it’s easy really. All you need (with Black, anyway) is a prepared defence to the Ruy, Scotch, 1 c4. 1 Nf3 and the QGA with 3 Nf3 or 3 e4 with 9 Nf3 or 9 Ne2 (we’ll take a chance on the other deviations) and in each case it’s just a matter of finding a good sound novelty which takes the machine out of its opening book which you haven’t seen. That shouldn’t be too much trouble: after it’s not as if hundreds of GMs have been scouring the same positions for just such novelties for many years.

Then you play hundreds of games with Fritz from that position (that should get round the randomness problem), and how can you lose? Especially given he’s had a whole six weeks since Elista finished.

I don’t know why they bothered to pay Kramnik really – hell, you or I could do it.

(it seems by the way that Kramnik doesn’t get to see ‘Fritz’s expected moves’ as you suggest, but hey, who needs facts when we can have opinions instead, right?)

I agree with Bill M; credit to Mig for not censoring. It's not personal: Mig's talking bollocks with an agenda in mind just as I and most other posters are doing.

Marc S; yes, that was my understanding. But the photo in that case is curious. Maybe before the game they just display for advertising purposes what a commercial user would see, and the photo caption is misleading.

True as well about the machine only choosing at random if there’s not one clearly best move. Kramnik can indeed see in advance how the machine is evaluating certain positions (supposing its book has ended by that time, which he won’t know but can perhaps assume). I don’t see that there’s all that much in dispute. The point of giving the engine in advance is precisely to give the human a better chance of finding and exploiting positions where the machine’s evaluation is wrong. The question – one of taste, essentially – is whether that will give a more interesting contest than watching humans ritually slaughtered in complications and/or open computer positions like Adams was or Kramnik was a couple of times last time. And of course whether it will assist the manufacturers more in improving their product (the answer to that is clear).

Maybe better use of the time spent here would have made a GM of RDH.

rdh wrote

>I don’t know why they bothered to pay Kramnik really – hell, you or I could do it>

not really of course, Kramnik is the champ and Chessbase's boy, just Topa is (was) FIDE's

To do the preparation as I suggested to always get the opening postion that you want is still work.
You have to study the lines you want to play against with the engine - and with the engine.

In essence Kramnik accepted to play this match and advertise Fritz and Chesbase granted that the games will be played out starting with positions where he has chances.

it is match with "opening odds", something that Kramnik has always stressed as necessary for him if he is to stand a chance against comps.
Read Mig's interview with Kramnik after Bahrain and who would want a painful and humiliating experince if he is certain of it ?

It is reasonable what he asked. Only that it would make the match appear as phoney, as a farce, if you take as the standard matches as Kasparov-Junior or Adams-Hydra.

As simple as that but you seem to have hard time coming in terms with the facts.

"Mig's comments on this thread are spot on, as they nearly invariably are."

"Thank you, Mig (and d), for your great, absolutely correct comments about rdh. You must be sorely tempted to ban wankers like him and Greg Koster who fill up your blog with so much useless rubbish, and hijack the threads."

I am no longer surprised that the most logical people on here capable of individual and deep thinking and whose posts I find the most interesting (and backed up with citations and facts) are consistantly called trolls and troublemakers by those who demonstrate fuzzy, dull thinking and/or sycophantic allegience. It is sad that this minority of contributers who really add value to a discussion are always made outcasts by the large set of mediocre, fat-bell-curve, least-common-denominator majority, but looking at how the world in general behaves, again I am no longer surprised. It is in that low-brow, fat majority's interests to keep thinking simple and conformity wide-spread.

What Mig realizes either consiously or unconsiously is that rdh, koster, and a few others add to this site as unpaid contributers that attract intelligent readers looking for their type of comments, even if they disagree with him. Mig would damage himself by banning them.

In fact, he should be appreciative of those willing to take the time to make intelligent posts, even if they contradict his own. That is why I think he knows of their contributions only unconsiously, because he instead does his best to belittle them.

If these intelligent posters were petty and wanted 'revenge', they can do so by simply not participating, leaving the comments littered by the Chris Bs, the ovidius, the tommys, and other intellectual lightweights who love to ramble pointlessly, and thus letting the 'lurkers' soon learn that spending time reading the comments they once enjoyed is now an exercise in ass-kissing compliance and unoriginal thoughts.

If all the 'good' posters (including rdh, koster, Marc S, Theorist, one of the Russian guys (Russian Bear I think, the one who translates russian press) got fed up by their treatment and migrated to, say for example, Susan Polgar's blog, it would have a deeper impact in readership on both sites than you would think. I think the only thing preventing that is currently there is no secondary choice worthy (yet), as some may find S Polgar's self-promotion even more revolting than here.

I don't imagine any informed person thinks that this match is on the same terms as Adams-Hydra. Obviously it’s useful to have the engine beforehand.

But what I don’t follow is your notion that seeing the book on the day is so vital. Take game two as an example: how exactly do you say that Kramnik was assisted in getting the position he wanted (let’s assume that was the one after move nine) by seeing the opening book on the day? We agree of course that he couldn’t have known for sure that 9…a5 would take the machine out of its book. He could have guessed from the commercial book just as Kasparov could have guessed in 2003 (?) against Junior from its commercial book, but no more than that.

This is the worst blunder for a champion in many years.

Seems to me there are two types of blunders, though:

1. A move is missed entirely. (Kramnik never saw the checkmate)
2. You see the blunder, but then forget it after analyzing several lines, instead choosing to go with a simpler move, forgetting why you dismissed it earlier.

It would be interesting to hear what was in Vlad's head. This can not be good for his confidence after he so impressively displayed his skill in Elista.

well said Stern. mr. koster, rdh and others are very good contributors on this blog. rdh has simply asked for clarifications about the match rules and you see from the above posts, how the friends of our blog meister insult him unjustly.

According to Kramnik, he had calculated from the Qa7 move, the resulting position, and did not realize that Qh7 checkmate was there! He said he does not know why he missed it either.

"It would be interesting to hear what was in Vlad's head. This can not be good for his confidence after he so impressively displayed his skill in Elista."

If you're going to blunder, this is the way to do it. Kramnik knows that only once in a million tries will he miss a mate in one. It's not as if the computer found some fundamental weakness. Obviously he's disappointed, but he's just going to keep on playing his game.

>He could have guessed from the commercial book just as Kasparov could have guessed in 2003 (?) against Junior from its commercial book, but no more than that.>

No man, you have not understood what I wrote.

It is a bit tricky and stresses that BOTH
"engine" and " seeing the book " are necessary for Kramnik's "magic cards always get the Ace whatever card you draw at random from the pack" trick to work.
(Not that I believe that the chessbase guys dont realize how this "magic" works, it is all agreement with Kramnik beforehand, a condition for
the match to begin with)

Precisely you start with the commercial book (at the limit you start with a general, universal, book that contains main lines, ECO )

Then you assume that this commercial book won't be identical with that of the match but slightly altered. It may contain more but what bothers you is that it may contain less.

Then you have to prepare FEW lines from this book.
You do this studying them by looking for deviations from the line at each move of yours and looking how the engine reacts to each until
you find one where your deviation makes the engine to, happens to be so, go for a bad (computerwise) postion or, equivalently, will go for a postion where postional understanding is what matters (rigid pawn structures, simplified by massive exchanges, and so on).

Then you play this out on the board in the match.
But the comp still has to see your moves as theory and not switch to engine before you get to the point you hit with your deviation if the good thing is to happen. It will very likely, but certainly ? No.

Why ? because of the "slightly altered book".

What if it happens that its opening book
unlike the commercial one at some point does no longer contain the line ? If you play it you will force the comp to switch to engine before the moment you want to. What you do next? well that's
why you have to prepare FEW lines.

To give an rather exaggerate example :

If the comp did not have 3..b5 in his opening book then Kramik would have played something else also studied 3..e5 so as to keep the comp in its book and to deviate at move 8h from the main line of 3..e5 with a "novelty" (deviation) also found at
home for this alternative line.

Deviation which, implicitly, would have been
something for sure not in Fritz commercial book
thus Fritz would have switched to engine and then go forward ( and unwittingly downward) by playing what Kramnik would have known that it was ultimately bad for comp (let's say leading to a
simple postion) since it tested it at home using the engine provided by CB.

Yes, I understand precisely what you mean. It’s no problem, all you need is to find in every possible existing theoretical line a deviation from opening theory which is either objectively good or which you know the computer will misevaluate, and then hope the computer doesn’t get to spring a novelty on you first.

My point was though that Gazza could have done the same without seeing the book on the day. You say: no because Kramnik can ensure that he keeps Fritz in book until he drops his bomb by switching if necessary to some other line where he also has a novelty prepared.

I don’t think you have any idea of how difficult it is to come to a match with a novelty ready in every line which the computer will react poorly to. It’s much much harder than what Topalov did in Elista.

I have a very easy explanation:
during game 2 ChessBase programmers in violation of the match rules modified the engine: they added an explanation of what one million dollars is. After that Deep Fritz woke up and immediately gave Vlad checkmate in one! ;-)

stern, ever heard the phrase "physician heal thyself"? Its very relevant to you..

Mig wrote of Kasparov:

"He considers the whole match a joke because of Kramnik getting the engine in advance"

It should be pointed out here that in '97 Kasparov complained bitterly when the IBM team refused to furnish him with scores of all pre-match test games played by Deep Blue. Kasparov even suggested that he be allowed to use a ThinkPad to double-check lines during match games, also rejected.

At the time I thought it was grossly unfair of the Deep Blue team; today I remain in the the "Kasparov-was-railroaded" camp. Perhaps Kramnik learned from his observation of Kasparov's experience and was able to strengthen his hand this time. But either way, should Kasparov be pointing fingers like Mr Short-Term Memory? He must truly dislike Kramnik to consider the match a joke for that reason.

Marc,

I takeback my patzer statement :-). It was meant as an irony. Hydra crushed Adams convincingly but DF has so far been unable to pose any problems to Kramnik. Just for the record I am of course against takeback in human vs human match and am proposing it just for man vs machine. In human vs human BOTH players suffer from fatigue and psycological pressure (so the playing field is level) but in man vs machine only one does. Therefore a takeback rule reduces the influence of such factors in the outcome of the game, thus giving more weight to pure chess understanding.

Well, he is an IM.

(That was to gg)

I don't know sometimes why Gazza doesn't go the whole hog and write a book called 'My Miserable Snivelling Successors'.

rdh
>I don’t think you have any idea of how difficult it is to come to a match with a novelty ready in every line which the computer will react poorly to. It’s much much harder than what Topalov did in Elista.>

We are not talking here about GM like novelties, about win-lose novelties in Marshall-attack or Najdorf strokes.

You look here normal moves as 7.Qd3 (bang! Fritz out of the book) in the first game. Your aims and judgment (good/bad) for your move referes to the character of the position, pro-human or pro comp, therefore the choices are way more and the evaluation of the postion is trivial for a GM :
(closed or open, simple or many pieces, Qs off or on ?...etc..

Practically you have to work with the engine provided the lines from the book that you intend to play and do so move by move up to 10-12th move.

You are mistaken if you think about these "novelties" as being the same thing as what is a serious novelty in GM OTB-play.

It takes a lot of work and a team to do it for you but it is not something as earth-shaking as it is robotic work in testing over again what the comp will go for, keep the evidence of the work done, kill the transpotions, etc. until you get a number of ideas for each game your are going to play.

Up to now it worked as magic for Kramnik judging the kind of positions he got out of the opening.
Even Adams would drawn Hydra out of such kind of "bad for comps" positions.
.

>Yes, I understand precisely what you mean. ...and then hope the computer doesn’t get to spring a novelty on you first.>\

well, not sure that you really do from what you write

The issue of being hit first by a comp novelty is not something that seeing Fritz opening book during the game can avoid. To avoid this you will have to play sidelines and hope that it was in the main lines that novelties were prepared by the Fritz's team.

What seeing the screen with Fritz book avoids is being able to switch between few ideas (deviations) that you have prepared for Fritz in case of that you hit a difference with the commercial book.

A short glance at the screen can tell you where you stand, no need to stare at it.

rdh, another way to have it, seeing the screen with opening book enables you to play if one then another of few novelties that you have prepared...but does not help you avoid those who have been prepared for you Fritt'z team.

Of course with this advantages of seeing the book
likely it will be you who is going to be the first to strike. You are right in this sense.

And Kramnik has been the first to do so up to now.

Changed your tune since you posted this a couple of days ago, ovidiu?!

"one should read before writing: it works better

I have just read the chessbase.com article which has that :

> The exact speed of the computer and the modification to the openings book are the two unknown factors for Kramnik in this match.>

I guess that this means that he can NOT actually prepare against it. He can not study it to look for style and "flaws" ( for instance search where he could deviate and afterwards the only sensible response, for anyone not only Fritz, would lead to simplifications )."

We shall have to disagree about the ease or otherwise of trying to find novelties computers deal with ineffectively. I would guess from what you say that I've spent more time trying to do it than you, but that's never a very useful comment between internet strangers.

After 7 Qd3 in the first game the machine went back into its book later on, so Kramnik can hardly have studied that very effectively, unless you think his bomb was later on. It would be interesting, to be sure, to know whether he knew that the machine was not likely in such positions to go ...Bxf3. That wouldn't surprise me; be a much more practical use of his time than your rather apocalyptic suggestions.

Actually that old thread was very instructive to read, from the point of view of seeing the effect of Mig's propaganda. I'd forgotten how many posters there were who had sadly got the impression from Mig's introduction that Kramnik could actually see the machine's analysis in some way.

radh,
bad grammar , corrected :

enables you to always play something prepared at home, if not one then another of the few novelties that you have prepared at home as you used the engine plus the commercial book ; and not the book actually used during the game (granted that the differences are not great)

rdh, if you are to qoute do it in the context.

At that time me and the others were barely starting to understand what was going on. We didn't even knew what exactly was Kramnik given and what the probability factor was referring to. I thought that it was altering the evaluation function and in such case using the engine to prepare openings against itself would have been useless.

You quote by taking things out of their context. This a common fallacy and a "low" way to carry an argumentation.

> point of view of seeing the effect of Mig's propaganda>

again you pre-judge a situation, calling something "propaganda" already, implicitly, assumes that the info is false.

we have started to debate it after Mig pointed that it was something wrong with it and we have gone long way since then...albeit not in the direction you would have wanted

"I continue to be amazed that people don't understand what an opening book is. It is simply the computer's stored memory of opening theory, along with its preference for playing/avoiding certain moves. Does anyone here NOT know that human players also 'remember' opening theory?"

Marc, I can only repeat that you missed my main point, and I certainly don't need your patronizing on what an opening book is, thank you.

I suggest you read my posts beyond the first three lines and try to understand what I'm saying, instead of what is more comfortable for you to believe I'm saying.

Alternatively, ask yourself why Mig suggested as a possible solution that Kramnik is given access to his own opening book.

Actually, it's neither here nor there, but you might be amused to know, rdh, that I found out your identity a few weeks ago through -- of course -- Google.

"Dear friends, it looks like some of you do not understand how SMP engines choose a move.
All multiprocessor engines are non-deterministic. This means that for the same position the same engine running on the same computer can choose different moves after spending the same amount of time or taking analysis up to the same ply count.
Even more. For exactly the same position (or the same move sequence) it can give slightly different evaluations."
-Posted by: Vlad Kosulin at November 28, 2006 8:58

Are you sure this is an inherent trait of an engine running on a SMP machine? For certain, any engine (single or multi-processor) will quite often play different moves when confronted with the same position in a series of games. But this is due to the randomness built into the engine, not anything to do with the computational architecture, right? This randomness seems necessary in commercial engines (designed to teach) since it would be pretty boring if it always played the same game, and also seemingly necessary in this match to prevent Kramnik from being able to tightly steer the games to a desirable position. Vlad, how is it that evaluations can vary when analysis is taken up to the same ply count?

Sorry, cadlag, but you wrote: "IMHO, it's unfair that Fritz gets human assistance in the first place (in the form of an exogenous opening book). If it is really that unfair that Kramnik can get a peek into that book, then switch the damn book off and we'll see what's what."

This betrays at least three fundamental misunderstandings.

First, you think there is something wrong with the computer's built-in memory, although it doesn't seem to bother you that Kramnik *also* has chess theory memorized.

Second, you think it's a problem that the computer had human assistance in the creation of its chess knowledge, even though quite obviously Kramnik has also had such assistance. To believe otherwise, you'd have to ignore the fact that Kramnik has seconds, and you'd have to believe that he learned all he knows about chess without help from anyone else, which is absurd.

Third, you think switching "the damn book" off is plausible, apparently forgetting that the human player would still have an extensive retained memory of opening theory --- the very thing you're complaining the machine is benefiting from.

Hey, I should put a rule where I can take back moves, 2 per game

Well said, Marc Shepherd!

The bottom line is quite simple, actually: given that this match is (SUPPOSEDLY) meant to determine whether the best humans can still play better than computers, the only way to OBJECTIVELY determine that is by sitting Kramnik and Fritz and letting them play to the best of their ability without outside interference during the game.

The two players should be able to train/prepare in any way they see fit, without having to disclose their methods or strategy.

THEN, you have an even match.

However, the average chess player is willing to cheat, buy or force his way into the winner's circle without regard for fairness or truth.

Crippling computers (by getting their opening book, eliminating the opening book, allowing take-backs or any of the other pathetic ideas thrown around) is akin to fitting a car with square tires in order to win the race!

For better or worse, computers are what they are and they have strenghts we don't have. If they have reached a playing level that the best humans can't match, so be it. Crippling the computer or tilting the playing conditions in favor of the human player is truly sad and unnecessary.

There is no way Kramnik can claim an honest win in this match (assuming he can pull a comeback).

Marc, keep building absurd theories about what you believe I think all you like, they're just a product of your wild imagination. I've already explained myself in other posts. So, I'll let you believe I'm mad enough to think Kramnik's mind was created ab nihilo and that he can switch it off at will. I understand now why you deliberately and repeatedly missed my point: you obviously feel safer in a dream world in which everybody else is completely crazy.

Look guys, no need to be subtle, this is how to do:
- make Rybka/Toga/whatever play 1000000 openings against Fritz, until Fritz gets out of book (~5-10 moves of computation, 3 min per move = 50 games per day, 100 days, 20 computers)
- get the evaluation of final positions
- filter the out the good lines (high probability of finishing with good position, and not too tactical)
- while you are at it, make Rybka play Fritz on the selected positions until the end, to see if they are really promising - without working too much
- come at the board
- play the moves most likely to yield good results
- (at the same time, look on screen to mentally note what part of the opening tree seem to have been changed, so that you avoid it next time)
- profit! (*)

(*) Offer may not be used in conjunction with any other discount, promotion, toiletgate or getting mated in one.

Kramnik was only uncounciously fair, he knew he had all the rules advantages and he did the right thing, gave the computer a point.

"I understand now why you deliberately and repeatedly missed my point: you obviously feel safer in a dream world in which everybody else is completely crazy."

There's always a big problem with resorting to the rhetorical tactic of saying that your opponent is deluded or crazy. If you sound deluded or crazy yourself, the tactic seems less like clever argument and more like psychological projection.

"cadlag", you're projecting.

getting apart from blunder psycological point.
1 Question, would it be fair to allow Kramnik to see and Use Nalimov Tablebase?, remember that this table is the product of thousands of computer Hours

All this hoopla about computers vs humans is nonsense. Chess computers are now stronger than the best human players, period. Why care about all the tweeking to make it more fair.

I really enjoy the Summer Olympics, but I haven't seen any dragsters or funny cars entered in the 100 or 200m's. Also, there isn't a champion weightlifter around who can defeat your standard forklift. My point being, let humans continue to face humans otb and let computers compete against one another to see who has supremecy.

Besides, I believe that the vast majority of us still enjoy watching a game between two world class players than a computer vs human, or even computer vs computer. In the meantime, I will continue to use my computer to hopefully improve or at least maintain my current level of play.

Right, Ernest, I'm projecting and I'm crazy. Just ignore my posts then, will you? Same goes for Marc. I never asked for his wild interpretations in the first place.

Enjoyable thread.
Stern, don't worry what d (who laughably called rdh a beginner last month) or any other poster submits to (in their misguided attempts) bolster Mig.
Mig is perfectably able and willing to defend his positions and he does so admirably (whether you agree with him or not).
Every community has their sychophants.
Ignore them.
Im betting Mig does.

sycophant <<----- for rdh the spelling nazi.
:)

Chesstraveler, the problem is that computers are rapidly exhausting the game of chess by finding out all the "chess-truth". This is the difference from the other examples with the weight lifting etc.
This has to do with general chess-info processing including opening databases etc. not just playing s/w like Fritz.
At the time of Fischer one could find innovation in 10th move and win a strong opponent ! Now a kid with a strong memory can play till 30th move in theory (regardless of the outcome).

Naturally, it's a matter of if one is happy with all these or not. I don't like the whole story. Others find it fascinating. It's OK. However, remember that chess s/w companies want us to talk all the time about their products and then buy them..

My humble advice (sorry if I sound like a teacher) : DON'T use your stupid computer to learn you chess. It's more stupid than one can imagine. Play with humans instead. Make your errors and then discuss with the other player about it.

PS I am not a good chess player but chess has taught me 2-3 good things computers cannot..

George,

I agree with what you're saying. If fact, I had thought about stating something along the lines of it won't matter 10-20 years from now anyhow, but didn't want to sound too pessimistic.

I prefer playing otb and do so regularly at various tournaments. Due to the fact that I currently live some distance form any clubs, I use Fritz to play 2-3 games per week [G/90} to stay sharp and check my openings and thats about it. No internet chess, I already have Fritz if you know what I mean.

youngvalter wrote:

"Here's a neat coincidence. Kramnik's blunder in this game has one thing in common with his blunder against Deep Fritz in Game 5 of the 2002 match: both blunders happened when Kramnik offered a queen trade. I wonder if Kramnik's anti-computer strategy has affected him to the point where he relaxes as soon as the queens are off the board."

Makes sense to me!

He tries to trade the queens when he plays against
any strong player.
Kramnik must think that playing the middle game in chess is a sort of misfortune or error.

"Just ignore my posts then, will you?"

Sure thing, hon.

I guess the sponsors are really happy - the blunder sure got everyones attention. I guess Kramnik can hold up his cheque for 500,000 euros [maybe 1 million] and say - how do you like them apples?

http://standartnews.com/en/article.php?d=2006-11-29&article=1923

Theorist, I'm not just amused; I'm honoured!

Beautifully spelled, Mondo. It's an interesting word, sycophant; I seem to remember the derivation has something quaint about it. It looks like 'fig-speaking'. Can that be right? I suppose it could be, if you count figs as a paradigm of sweetness; somthing like 'sugar-speaking. No doubt Google can tell me.

So has anyone seen this 'protocol' of which Danailov speaks?

>I guess the sponsors are really happy - the blunder sure got everyones attention.>

sure and has shown that Kramnik too can play exciting chess...

but now Chessbase how to solve the inverse problem and make Kramnik win one, because 3-3 would be the perfect result for advertisment

but, as in math, the inverse operation is a always much more difficult than the direct one

From the Danailov interview:

"Now, it is easy to speak from the distance of time, but there I was in fear for my life. They came to me and said: "Be careful! You will have problems!"

Danailov is right. He did speak. And he does have problems.

> They came to me and said: "Be careful! You will have problems!">

does Danailov name these "they" in the interview ?

Of course not, Ovidiu.

Pronouns are very telling. My law firm has a rule: we don't take on clients who refer too frequently to their enemies as 'they' in the initial interview. If you can't tell from the context who 'they' is, experience has taught us that it is likely that the speaker is truly an unsatisfactory individual, rather than merely one with a loose grasp of grammar, as one might have thought.

Fascinating observation, rdh.

Let's hope it never becomes conventional wisdom... because then the paranoids and assorted other "unsatisfactory individuals" will catch on and start replacing "they" with "the Jews." That works just as well to denote a nameless enemy, and in the entire world excepting (parts of) the USA, carries no social stigma.

Paranoids - good Lord, you misunderstand me. To a lawyer, 'unsatisfactory individual' means someone who may not pay our fees when they fall due.

rdh,

lol!

I was puzzled too and was going to ask how could a law firm's client be an "unsatisfactory individual" if he paid his bills.

Touche, rdh! And I'm in your debt, for giving me an excuse to go out and rent "Devil's Advocate" again...one of the most entertaining films I've ever seen. Had I watched it recently, I might have been better equipped to discern the meaning of "unsatisfactory individual."

GM Banjamin talking with Mig on chess fm last night about the rule changes [advance access to the final program, seeing opening book] you know the ones that Mig said were "unnecesary and absurd" and Kasparov apparently believes render the match "a joke"

GM Benjamin: "maybe they've hit something pretty good because this is kind of an even match" Mig reply "hmm yeah"

I agree with GM Benjamin and Mig - those rules are pretty good

Thanks for bringing it up, Andy, but you cut my actual reply, the part that comes after my noise of encouragement for Joel to continue and explain his opinion. It certainly wasn't agreement! My reply was that his statement is a form of deciding these things based on the score and not what is most interesting, most fair, etc. As I posted above a few days ago. Kramnik dominated play in the first two games but was down by a point, so saying the match was even is tricky. That is, just because Kramnik clearly isn't crushing Fritz doesn't mean I think letting him see when he can take Fritz out of book is a good thing.

As Joel pointed out, and what is more interesting, is that there are two definitions of fairness. One is an attempt to define what is ethical or appropriate, which is obviously subjective and is also mostly what we talk about here. The other is trying to keep the sporting balance even, giving the human a fighting chance even if the rules don't sound, and here in quotes, "fair." So while Kramnik seeing Fritz's opening book doesn't sound fair as in fair play, it may turn out to be fair in keeping the score or the play even. It was an interesting discussion. I again suggested just giving Kramnik his own opening book.

To further illustrate the point in a slightly silly fashion, it might balance the score if comps gave up a pawn. But this would look embarrassingly unbalanced and unfair in the first sense even if it produced even matches on the board and scoreboard. So we need to balance fairness and evenness.

For example, I think giving the engine and letting Kramnik see the book are too far against fairness no matter their effect on evenness. The question is whether or not there are "fair" things that can be done to keep the fight even. Or is it simply time to give the engine and just about any other advantages just because the damn things are so strong it won't matter that much despite our pseudo-ethical preoccupations. The score of this match is close to indicating that, which is really the point Joel was making. If we have to start getting pawn odds it's clearly not worth it!

Thanks for the detailed thoughts Mig. I didnt intentionally cut your reply but on reading your response I remember the fairness/eveness issue coming up but I didn't get the point at the time. [too busy noting Joel's comment]I guess in the end I thought words like absurd and joke were too strong and the access given to Kramnik is better described as questionable.

Anyway I think you have raised the key point - where to draw the line - taking back a move or giving a pawn odds well then words like absurd and even joke might be right. One thing Mig this access that Kramnik had was definitely necessary for him to be competitive as I am not convinced that Kramnik dominated the first 2 games he certainly was in control and in game 3 he lost the initiative. Even with the advantages he had I think it would have been a big achievement for him to have won the first game and drawn the rest. I think computer programs are just too strong now although I would like to see them play without their opening books to see whether they would be competitive at top levels. It would give programmers incentive to build in further strategic concepts to avoid repeat strategic errors in certain openings. After this match I am not sure I see the incentive in Fritz , 11,12 etc or future man machine matches. Its the next logical step to design a chess program that can win the world championship without any opening book - just using principles to generate the opening. I am not sure what the commercial incentive would be but I think it might give insights into chess. By the way I am sure it's my ignorance but I did not know what Joel stated that Kasparov had asked for Deep Blue to play without an opening book.

I think I know why Kramink blundered.

When playing against computer you are not supposed to challenge it in tactics. Sometimes it does not make much sense to check tactical variations because computer has probably checked them anyway and you are supposed to do strategic fight.

So Kramnik most likely didn't spend much time on short term tactics but was concentrating on long term plans - in other words, he was playing for a win.

Kramnik should have protested!

Matthias was clearly playing with the help of a computer!

(Where are those managers when you need them...)

;-)