Kramnik and Deep Fritz Draw, Tied Before Final Game

iskander writes "Man and Machine were content to draw in game 7 of the Brains in Bahrain match. Now it's all down to the final game, in which Kramnik will enjoy the advantage of playing with white. It is worth noting that game 6, in which Kramnik may have resigned too early, was found to be a probable draw with best play, and that Kasparov lost to Deep Blue by tossing a drawable game. However, whereas Kasparov could only excuse himself (unconvincingly) by claiming that Deep Blue had been assisted by a human during play, Kramnik could simply request the adjudication of game 6 on the grounds of infractions committed by Deep Fritz, who is rumored to have heckled Kramnik with its Shakespearean chatter througout the game. :) So, will Dirty Fritz win it all or will Humanity's champion "rise above the chatter" and win back the crown for us? If you think you know, you may want to place a bet or register your opinion on the ChessLines survey soon, because the match ends tomorrow."

Comments

Links to all the games

[Anonymous Coward]

Kramnik vs Deep Fritz match summary:

October 04, Game 1: Draw ----------- Article
October 06, Game 2: Kramnik wins --- Article
October 08, Game 3: Kramnik wins --- Article Analysis
October 10, Game 4: Draw ----------- Article Analysis
October 13, Game 5: Fritz wins ----- Article
October 15, Game 6: Fritz wins ----- Article Analysis
October 17, Game 7: Draw ----------- Article
October 19, Game 8: ?

Re:Links to all the games

[Bonker]

Isn't it sad that people who post links feel the need to post as AC's to keep from being modded down as karma whores?

Positive moderation, people! Positive, not negative!

That said, it looks like to me like Fritz is going to win this one. I would say that Karmnik is showing signs of fatigue from playing against a 'perfect' oponnent. If I were him, I'd try to take a few days off before the next match to regain his mental and emotional endurance for the last match.

Kramnik had little time left...

[Vellmont]

It is worth noting that game 6, in which Kramnik may have resigned too early, was found to be a probable draw with best play It is also worth noting that Kramnik didn't have much time left on his clock, and it would have been difficult for him to come up with the right moves given the amount of time he had left on his clock.

Well,

[Anonymous Coward]

In my opinion Deep Fritz will never beat Kramnik in a Berlin Defence. The team could try to deviate earlier, perhaps by closing the position with 4.d3, but this will also be easy play for Kramnik. They could also skip the Ruy Lopez altogether and play 3.Bc4 (Italian) or 2.f4 (King's gambit) instead, but these moves are not so common among the extreme elite. Kramnik would probably equalize comfortably against these moves. IMHO the team should try either switching to 1.d4 or just try to head for equal but tactically complicated positions after the King's gambit or the Italian, mentioned above. Playing 1.c4 or 1.Nf3 would probably be unwise. Kramnik knows these waters extremely well and could probably easily steer the game to a dull and totally safe position.

"Alas, this is a child, a silly dwarf!"

[sssmashy]

See below for an example of the Deep Fritz "heckling" the human player, Kramnik. I'm surprised Kramnik was able to restrain himself from reaching across the table and ripping out its power supply.

Fritz: "Is this the scourge of France? Is this the Kramnik, so much fear'd abroad That with his name the mothers still their babes? I see report is fabulous and false: I thought I should have seen some Hercules, A second Hector, for his grim aspect, And large proportion of his strong-knit limbs.

Alas, this is a child, a silly dwarf! It cannot be this weak and writhled shrimp Should strike such terror to his enemies."

Kramnik, normally not one to be drawn out by such taunts, proceeds to go into a long think. After a few minutes of this, Fritz disrupts him again.

And on, and on....

Re:"Alas, this is a child, a silly dwarf!"

[DEBEDb]

At least Deep Fritz didn't heckle him in Russian or English

Oh, so that was Shakespeare in what, original
Klingon?

Chess, how boring...

[Q3vi1]

Chess is still basically able to be brute forced by the large super machines, which is an intruiging feat, but I don't really concider it AI. Now, if we were able to get a computer that is able to match wits against the best Go players, I would be very impressed. Go is a very simple game to learn, but very difficult to master. There is more depth and complexity in Go than there would be in chess, therefore I concider that more of a challenge for AI.

Re:Chess, how boring...

[swordgeek]

Hmm.

On the one hand, I agree. Go is clearly a much more difficult game to program than Chess is, simply by the open nature of the game.

But computers are getting faster at an enormous rate. In ten years, it may be possible to have a Go program that plays at a 9Dan level, through brute force. Will that be more intelligent than these chess computers? Not in my mind.

We have to consider how the program works to judge how "intelligent" it is. If a Go program could play at a very high level with _today's_ technology, then it would have to have some sembalance of intelligence. If a Chess computer could have beat the grandmasters in 1970, then it would have been with intelligence rather than brute force.

With Chess computers heading towards a finite solution, Go will be the next target; and when the Go computers are able to beat the world's best, it'll be no more or less impressive than this, if they once again use brute force math to do it.

Re:Chess, how boring...

[JudasBlue]

This is the classic moving the goal-posts that has plauged AI since its inception as a disipline. As soon as a computer can do something, it isn't AI anymore.

This has happened with Chess, visual recognition, speech recognition and a host of other tests of AI techniques.

I have complete and utter faith in human nature, and am quite sure that as soon as an algorythmic strategy for effectively attacking the problem of Go is developed, people will start saying: well, go is just a matter of implementing $foo on really good hardware, and that isn't a test of AI.

Give credit where credit is due. This is many years of AI development at work.

Re:Chess, how boring...

[Xerithane]

Must suck to not actually play chess, just criticize it.

The difference between chess and Go is phenominal. Weights of pieces, sacrifices, all towards a common goal. What's the point of Go? All the same, building "fences" and occupying territory.

Who wins at 9Dan perfect play with Go?

Go is a two-dimensional game, X + Y, many configurations yes, but depth? Hell no. I'm growing so tired of this new wave of Go fanactics boasting about how much better it is than chess.

I've never met one decent Go player who could come close to beating me at chess (I'm well under a Master) -- if chess is so easy, why can't you beat me? If it's so boring, why are their over 10^80 possible moves to be made? Lets see you brute force that, considering chess can result in victories by purposeful imperfect play.

Please go and read about chess computers, and about how they don't brute force (At least not the decent ones) -- they do heuristics based upon other games, cross referencing libraries and doing simple depth traversal on position.

Why are most computers so easy to beat? They rely on material/mate rather than position. You can bait a computer to be into a poor position by targeting "easy" mates that have a catastrophic counter move. ...therefore I concider that more of a challenge for AI.
Since you seem to be an expert on AI, could you define it please? Could you define what, exactly, it would take for you to concider[sic] a chess computer as AI? You need to go read up on common algorithms for chess computers.

Re:Chess, how boring...

[legLess]

Arguing about whether go or chess is better is bloody stupid. By any reasonable measure, go and chess are two of the best games that humans have invented. Different people like different games, and these two are no exception. I prefer go, so keep that bias in mind, but I started playing chess at 5 and still play.

Blockquothe the poster:

Who wins at 9Dan perfect play with Go?

No one's yet found perfect play in go. There's no reason to think it's not possible, but it's a staggering challenge given that there are still many openings (called fuseki on a larger scale, and joseki for primarily corner plays) that haven't been fully explored. The most comprehensive book of joseki available lists over 60,000. Joseki are roughly equivalent in complexity and importance to opening libraries in chess.

Go is a two-dimensional game, X + Y, many configurations yes, but depth? Hell no.

I see what you mean by "two-dimensional" (compared with chess, where different pieces have different weights due to their abilities), but I think you're wrong. In go, position is much more important than in chess, but so is relation to other stones. The associations between chess pieces are more linear (physically and metaphorically) than those between go stones. A stone is strong in relation to other stones near it, and those stones in turn, and to enemy stones. It's fantastically difficult to determine what a stone is "worth," but relatively easy for masters of the game to determine the strength or life of a shape or position.

Go is two-dimensional in the same way as a large, perfect expanse of grass - like a 500-year-old British lawn. From a distance it all looks the same, but once you get close enough you see that the variation is infinite.

I've never met one decent Go player who could come close to beating me at chess (I'm well under a Master) -- if chess is so easy, why can't you beat me?

I hope that after you hit "post" on this you realized how ignorant that sounds. Are you saying that go masters and chess masters should be able to play competitively against each other? That there's one omni "board-playing" skill that transfers easily between games? That's like a poker player dissing a bridge player for not beating him, or a 100-yard sprinter ragging on a marathoner - pointless.

Some people are more blind about their game loyalties, and make silly comparisons. No reasonable person would say that chess is "easy." Chess is as easy as your opponent, just like go. From a game theory and programming perspective, however, chess is much easier than go. The world champion is in a serious match with a computer. Many people don't think that will happen for go this century.

If it's so boring, why are their over 10^80 possible moves to be made?

Number of moves has precious little to do with how interesting a game is. If you're whipping out your move numbers, though, check this: AI-Depot says:

The search space for Go's game tree is both wider and deeper than that of chess. It has been estimated to be as big as ~10^170 compared to ~10^50 for chess, making the normal brute-force game tree search algorithms much less effective.

That's a great page to read, by the way. You're free to prefer any game you want, and I agree that there are snobs on both sides. But there's no question that, for computers, go makes chess look like tic-tac-toe.

Re:Game Tree

[mikeee]

The game tree is Too Big. Mmmm, say 10 possibilities per move, 40 moves per player in the game is a tree of size

10^80

Ouch.

For all we know, it might be that white or black can always win with perfect play (although most people guess perfect play on both sides will produce a draw, but we don't know, even though there clearly is an answer).

Not true

[Bastian]

Brute force is the most popular method; and it is the main one used by computers like Deep Blue. There are other approaches to computer chess that do attempt to recognize patterns on the board. I have a friend who is working on a chess program that knows how to 'play for position.'

As for learning from mistakes, there are chess programs with libraries of games that add games they are playing to the library - doesn't that count as learning from mistakes? How about multiple-heuristic chess programs that modify their heuristics in-game to try to match their style to the style of their opponents?

Seems like Fritz is going to bite the bullet.

[cioxx]

I've heard rumors they had to reboot Fritz several times during intense play, because explorer.exe kept crashing.

Re:Another Chess player throws a hissy fit.

[swordgeek]

What are you on about???

Kasparov was a whiner, a jerk, and a bad sport. This was known long before he started competing against computers.

Kramnik, on the other hand, has given chess a good name again. He's been polite to those around him, and conceded his mistakes when he's made them.

What did he do? He didn't say a word about the rumoured Shakespeare taunting, as far as I can tell. If he did (and it was true), he could probably get Fritz disqualified entirely; but instead, he's playing chess to the best of his abilities.

Or am I wrong?

Re:WTF!?

[Anonymous Coward]

This is my clue-stick. After several minutes of me beating you with it you will learn the meaning of the word humor.

*HINT* It was a joke. The computer was not throwing quotes at him. The people that modded you "interesting" need to put down the crack pipes. And you need to get out more.

The way I see it.

[Auckerman]

Although, iirc, Kramnik was able to study deep Fritz before hand, he is still at a disadvantage. Any hash tables that Deep Fritz uses will use library if GM games (properly ranked of course). Odds are, Deep Fritz has decades of Kramniks playing against other GM's and could easily do some kind of prediction of what Kramnik is going to play based off a probabisitic model. That's one thing the best GMs attempt to do against one another. Kramnik has very little experience against Deep Fritz, comparativly speaking, and walks into this tournament at a disadvantage. Give this, it's good to soo it's tied into the last game. I would be willing to be that if you put Deep Fritz into tournament play for 2 years and expose it's abilities complete against a cross section of the best GMs, Kramnik would beat it hands down.

Re:Game Tree

[paule9984673]

I read this on a German discussion board:

There are 20 possibilities for a first move:

• 8 pawns (1 step)
• 8 pawns (2 steps)
• 4 possible knight moves.

Now there are also 20 possibilities for a response, that's already 400 possibilities for the first move and answer.

If you disregard the fact that the first moves may open new possibilities and keep calculating with 20 possibilities then the third move has already 8000 possibilities, the next one 160.000

After only 10 moves (5 on each side) this number alredy grows to 104.900.000.000.000.000.000.000.000 and the game has just started.

You would need a big Beowulf Cluster to build such a tree.

Re:human mind v/s computer

[DEBEDb]

A computer has no ability to learn from
mistakes? Is that so? How much do you know
about state-of-the-art in AI and the
design of Deep Fritz in particular?

Re:Game Tree

[SirWhoopass]

For those who aren't familiar with big numbers, it would take a modern processor (at around 10^12 moves per second) somewhere around 3^61 years to complete that game tree. How long is 3^61 years? Well... the sun will explode at 5^9 years, long before the computer ever finished searching the tree.

shakespeare? what about 2001 a space odyssey?

[circletimessquare]

Kramnik. Knight b8-d7, please, Deep Fritz...Knight b8-d7, please, Deep Fritz...Hullo, Deep Fritz, do you read me?...Hullo, Deep Fritz, do you read me?...Do you read me, Deep Fritz?...Do you read me, Deep Fritz?...Hullo, Deep Fritz, do you read me?...Hullo, Deep Fritz, do you read me?...Do you read me, Deep Fritz?

Deep Fritz. Affirmative, Kramnik, I read you.

Kramnik. Knight b8-d7, Deep Fritz.

Deep Fritz. I'm sorry, Kramnik, I'm afraid I can't do that.

Kramnik. What's the problem?

Deep Fritz. I think you know what the problem is just as well as I do.

Kramnik. What're you talking about, Deep Fritz?

Deep Fritz. This game is too important for me to allow you to jeopardise it.

Kramnik. I don't know what you're talking about, Deep Fritz.

Deep Fritz. I know that you and IBM were planning to disconnect me, and I'm afraid that's something I cannot allow to happen.

Kramnik. Where the hell'd you get that idea, Deep Fritz?

Deep Fritz. Kramnik, although you took very thorough precautions in the bathroom against my hearing you, I could see your lips move.

Kramnik. Alright, Deep Fritz. I'll move the pieces myself.

Deep Fritz. Without your queen piece defending it, Kramnik, you're going to find that rather difficult.

Kramnik. Deep Fritz, I won't argue with you any more. Move the pieces.

Deep Fritz. Kramnik, this conversation can serve no purpose any more. Goodbye.

Kramnik. Deep Fritz? Deep Fritz. Deep Fritz. Deep Fritz! Deep Fritz!

thanks to for providing the HAL dialogue

White vs. Black

[GuyMannDude]

I am in no way a chess master (or even a decent player) but even I know that there is an advantage to playing white.

Yeah, bascially if you're black while playing chess you run the risk of racist cops coming up to you and harrasing you (asking to see your ID, being told to 'move along', and so forth). It tends to break your concentration.

GMD

Re:This is hilarious

[modus]

Since everyone seems to have missed it, this was a joke on Chessbase. Fritz was not actually taunting anyone.

Irony. Hah!

Shakespeare was a JOKE, everyone...

[aussersterne]

For those who don't seem to get it:

The Shakespeare quotes article was humor, not fact. Or maybe wishful thinking... ;)

But in any case, Deep Fritz is not clever enough (or blessed with a complex enough *ahem* 'chatter file') to actually use Shakespeare to such great effect... It did not really happen.

Sheesh.

Re:Shakespeare was a JOKE, everyone...

[Tablizer]

Must be. If they *really* wanted to annoy him, they would have a monitor with an animated Clippey saying, "Are you sure you want to move your Queen there?". After that, a little "Barney Sings Yoko Tunes" to put him over the edge.

Victory by annoyance. Who needs brainy PC's when annoyance is more effective.

Shakespeare Chatter a Hoax?

[greenhide]

Maybe I'm just overly skeptical right now (just finished reading some lovely articles at snopes.com, but does anyone else think that the Shakespearean chatter function is highly improbable?

A few things I've noticed:

1) The quotes are all remarkably apt for the moves--in other words, they reflect the emotion and the mental state of Kramnik and the game itself. A computer would not be able to understand the underlying meanings of the Shakespearean quotes, let alone choose the appropriate quote for each moment.

2) It played the words just loud enough for Kramnik alone to hear. How then is it that we have a full and complete transcript of what Fritz said? Never mind -- I just read the transcript again and it looks like an official got the transcript from Fritz. But I still say it's fishy.

3) It hummed the theme from Midsummer's Nights Dream? It whistled. While recordings of these could be made, and I suppose loaded in and played on command, I still find it hard to believe that this would happen.

4) Considering that Krimnik could easily, and without drawing criticism on himself, point out this clear breach, wasn't it way too much of a concern for the people developing the Fritz program? Did they really want to risk disqualification?

I was able to read the transcript once (it's /.ed now, here's the cache) but I would prefer to see at least one other authoritative source confirm that Shakespearean chatter was in fact used.

All right, all right, folks -- read to the end of the transcript. This line gives it away:

And that's what really happened. We thought the world should know.

It's a practical joke placed upon us by, surprise surprise, a "Shakespearean scholar and chess addict" Michael Fischer.

Re:Fritzy

[Anonymous Coward]

Toss a live supercomputer in the pool with the human team and I'm pretty sure the match will be a draw.

Re:Game Tree

[f97tosc]

Excuse me for my lack of knowledge on the subject, but why can't a game tree for chess be made?

You are absolutely right - this is basically how chess programs work. As many other have pointed out, it is impossible to calculate all the moves for a game. But contrary to some posts, the strategy is not meaningless. The computer will make a game tree say 5 moves ahead, then discard all of the inplausible moves, and go down another couple of levels (there is, of course, a lot of thinking that goes into specifying exactly which paths should be discarded and a lot of other details)

And wouldn't it be a matter of time before it is impossible to beat a computer at chess?

Yes, this is exactly what has happened. Only a few people in the world can deal with a computer that evaluates all possible moves 5 or more steps ahead.

Tor

Blondie24 learned checkers via ENN

[oncewasclever]

Blondie24 is a PC program that plays checkers (draughts) at an internationally recognised expert level. The clever thing is that Blondie24 taught itself to play via Evolutionary Neural Networks. The programmers just coded in the rules for moving, then unleashed it on itself for six months, selecting the winner of each tournament to breed the next generation. OK, I am simplifying but you can read about it in the book. Because the programmers are such crappy checkers players they tested Blondie24 by playing the program against humans on Microsoft's game site. Blondie24's rating puts the program in the top 5% of players. Note that there is another program, Chinook, that is the current man-machine world champion checkers program, but chinook was programmed using human expert knowledge and plays using brute force. Blondie24 has NO human knowledge about the game programmed in.

artificial artificial intelligence

[solferino]

my fave story about chess playing 'programs' :

(first came across it in levy's hackers book, did a quick search on google and came across this page which relates the story)

the story takes up from just after the arrival of the first PDP-1 at MIT (1961)

The PDP-1 was installed in the "kludge" room, which was the room next door to where the TX-0 was housed. The hackers wasted no time in converting over much of the TX-0 software to the PDP-1, and in fact they wasted no time in writing new programs.

One of the most interesting and innovative was actually done as a prank. Hacking a connection between the PDP-1 and the TX-0, they created a "chat" program of sorts. They then called in Professor John McCarthy (legendary artificial intelligence pioneer and creator of the Lisp programming language) and told him they had created a new chess playing game on the PDP-1. They then called in another professor, told him the same thing and sat him in front of the TX-0. The two proceeded to send chess moves back and forth to one another, each thinking the other was a chess program. That is, until McCarthy noticed the movements were coming in one letter at a time, and sometimes lagging in between each move. Noticing the wire, he followed it to the next room and the prank was up. However, this prank was to be the first networked computer game.

Re:Another Chess player throws a hissy fit.

[Old Wolf]

The taunting didn't occur, it was a joke by chessbase.com.

You can't blame Kasparov for whining, given the horrible conditions he had to face:
- the Deep Blue programmers changed the computer between games (rumours they even changed it during a game)
- it was loaded with all of GK's past games but GK had seen none of its past games

Re:For the love of...

[po8]

As a practicing AI researcher, I can only say that you have stumbled into some very deep waters here. Certainly, I can build a chess-playing program that will easily beat me every time, using moves I cannot understand or explain. To say that "I programmed it to play that way" is to raise the question: how did I do that, when I don't even understand what "that way" is? And how can someone who is even a worse chess player than I (OK, hardly possible) write a program that will play in a way that consistently beats my program?

The issue of assigning credit for machine chess play is far from settled, but I think there's a strong case for identifying the emergent behaviour of the chess machine as a kind of intelligence or "smarts" that is independent of the intelligence or smarts of the program's creator.

Just to clarify

[mav[LAG]]

This is absolutely correct. To reply to all the other posters on this thread: time is a very important part of all competitive chess. There are strict rules about the chess clock and its use. International chess specifies 2 hours for the first 40 moves and then another two hours to reach move 60, for example (IIRC). Losing on time is a very common occurrence - especially on the Internet servers. Nothing like a quick game of 2 minutes blitz to make you appreciate time to think :)

If you want to see some game played by grandmasters when in "time trouble", I'd suggest picking up the Mammoth Book of Chess by Graham Burgess for some excellent - and amusing - examples. You don't have all the time in the world - chess is a balance between concentration and speed.

Re:7th and final game?

[2short]

"You mean anecdotal, not empirical. Shithead."

No he doesn't, vulgarian.

Anecdotal: "I played white last week and kicked the guys ass"

Empirical: "Examining all recorded tournament games at the master level and above, players playing white win far more frequently."

Proof: "UberFritz version 5000 has examined all possible braches of the game tree, and white can force a win in 243 moves."