Chess: Kramnik vs Deep Fritz (again) is billing it as the last chance for mankind to defend themselves against the computer.

Kramnik hasn’t faced Deep Fritz since 2002 (a match I watched intently.) I’ve looked through the media available and can’t find any direct answers, though they give specs based on a 2 gHz notebook.

I don’t know if Kramnik can draw it this time again, we’ll see.

What do you guys think?

– IG

Oh, and some info links:
Official Site -
Chessbase Article -

Kramnik certainly will have chances, particularly given some of the rules for the match regarding adjournments, tablebases, etc. I’d expect Kramnik to at least get a win or two in, but it’s hard to predict. The comps are still vulnerable to certain types of games, although of course they continue to improve.

I’ll predict a Kramnik win by +2 -1 =3. Not that I have any particular insight. :slight_smile:

I think Kramnik has negotiated a pretty good deal, all the way from the money through having Fritz’s openings book to having a copy of the program beforehand.

However I am pessimistic about a human’s chances against a computer these days. Michael Adams got crushed by Hydra.

I think Kramnik could win one game if he finds a mistake in Fritz’s openings book, otherwise I predict Fritz will be unbeaten and score about 75%. :eek:

Could you (or someone) explain the notation above? I don’t know much (well, I don’t know anything, really) about chess tournaments, so I’m not sure what the + and - signs represent.

It just means 2 wins, 1 loss, 3 draws. It’s one of the shorthand ways to describe a match or tournament score. So my prediction for the overall score was that K will win 2 games and lose 1, with the rest drawn.

Somewhat similarly a player might be described as +2 for an event, which means that they won 2 more games than they lost.

You know… I wonder if the experience for Kramnik is anything like the experience us regular chess players get when we face a chess program.

I can beat lower levels, but when I crank it high enough, I get spanked all the time.

It’s pretty scary.
The top players can prepare for each other by:

  • studying every game their opponent has played
  • knowing their opponent’s opening repetoire (there might be one new opening with each colour in a match, but it’s too much work to assimilate more)
  • knowing that time-pressure causes people to play poorly
  • knowing that a setback in one game can affect that player next time
  • knowing that a history of bad results (e.g. Shirov v Kasparov) can unsettle a human

None of this applies to a computer, plus computers don’t get tired or make occasional mistakes.

Susan Polgar (2500ish rated player) reckons on her site that Kramnik will play boring lines and score 3-3.

I note from the rules that when the position reaches a known draw (from endgame tablebases), that Fritz must offer a draw. That seems like quite a concession, since Fritz plays perfectly from there on and Kramnik may still have to find some ‘only moves’.

I can’t see what happens to the prize fund if the match is drawn. (There’s $500,000 for the winner.)

Well keep in mind that Kramnik has had a copy of the program as well as it’s opening playbook etc. So he has had a good opportunity to study his opponent.

– IG

Am I wrong in thinking that it will soon be impossible to have a fair chess match between human opponents?

The assumption is that chess programs will continue to improve, as will undetectable communication technology. If this is so, human opponents will be able to receive real-time coaching that can’t be detected or prevented.

I hope I’m wrong here.

In a sci-fi world, sure it will be impossible to stop a cheater. But it won’t be simple. For the world championships they had a few weeks ago they had noise generators to interfere with wireless communication and such, cameras everywhere, etc…

– IG

The much more scary thing is that, inevitably, the “problem” of chess will be solved by a computer. Either it is a forced mate by White, or it is a forced draw.

Just needs someone willing to devote enough computational power to the problem. :frowning:

The amount of possible moves and the amount of possible positions on the board, make this a problem which would require an absurd amount of processing power in today’s world. There isn’t any financial gain to be earned by solving this problem, so we’re still a ways off of doing that… if it is even possible.

– IG

Game 1 ended in a draw. Kramnik played his stoic and safe game, avoiding several opportunities to complicate the position and accepting the draw on move 47.

It’s just a start but if Kramnik plays like this, I expect to see another draw for Mankind.

– IG

Somewhat surprisingly, it’s actually a forced mate by Black! I have a truly marvelous proof of this proposition which this “reply” window is too small to contain.

The thing to realize is that chess is an exponentially-large problem. Even if we assume that Moore’s Law continues unabated, the computers a millenium hence still won’t be able to solve chess via brute force. The only way chess will ever be solved is if some genius finds a clever but simple non-brute-force calculation (analogous to the solution of Nim), which seems incredibly unlikely, or if some entirely new sort of computer is developed. I suspect that a sufficiently-complex quantum computer could do it, but like most tasks, no quantum algorithm has yet been found, so this is just speculation.

Well, computers are twice as fast as they were in 1973, so who knows?

I refer you to Wolfram’s Math World for further reading about solving chess:

– IG

There was a great short story I once heard about that I’ve never actually read (I rewrote it for a storywriting contest, but didn’t win): Alekhine, on his deathbed, friendless and alone but still World Champion, tells of an adventure he had at St Petersburg in 1914. His preparations for the morrow’s game against Nimzowitch are interrupted by a knocking on his hotel room door. It is an elderly Russian peasant insisting that he has solved chess, and found that it is a forced mate in 12 for White. After failing to get the maniac to go away, Alekhine sets up the board and pieces and challenges the peasant to prove his assertion. Twelve moves later the future World Champion is staring at the board in disbelief. “Do that again!” he demands. Several more losses ensue, with Alekhine mated time after time whatever defence he tries. He goes to Lasker’s room along the corridor and summons him to the board. Lasker too is unable to find a line that keeps Black from being mated in twelve. The present and next-but-one World Champions’ eyes meet over the board…

“And then we killed him, of course,” whispers the dying Alekhine.