Greengard's ChessNinja.com

Ren Xing versus Ji Qi

| Permalink | 24 comments

(That's my feeble and fittingly computer-assisted attempt at "Human versus Machine" in Chinese.) There was a big human-machine match this week, the top software programs taking on some of the world's best GMs. If you haven't heard about it and don't know the programs Cyclone and SevenSky or GMs Bu Fengbo and Wang Yang you probably don't play Xiangqi, or Chinese chess. A team of five GMs played five programs in the Copa Langchao at the Olympic Sports Center.

I learned to play Xiangqi (and Shogi) once but would need to brush up to play a game now. I remember there are obstacles on the board you have to move around and that the chariot was the strongest piece. Both games are much faster-paced than our European chess. Draws are rare and games are short. Imagine eliminating castling and most of the pawns. One thing is the same; the machines won the match 11-9.

The press conference for the November 25 Kramnik-Fritz match in Bonn took place in Essen the other day. Six games and Kramnik gets $500,000 for showing up and double that if he wins. Said Kramnik, who drew with Fritz in Bahrain 4-4 in 2002:

The machine is the clear favorite, but don't discount me just yet. I know some top players would be very nervous about playing the computer – they might even avoid this kind of match. That's understandable since a cut-and-dried defeat can affect your future game. Of course, this computing monster keeps getting better year by year, month by month, day by day: My opponent will be incredibly strong. But I think I can still beat it. Whenever I can fight, I'm extremely motivated. After all, I might be the last human [able] to defeat this machine. My team and I will be expending all our efforts to cut this so-called artificial intelligence down to size."

Sure, lots of players would turn down half a million bucks even if they thought they'd be crushed. Sign me up. Heck, I'd even settle for half that. I think I know enough anti-computer silliness to grovel to a draw or two. Poor Mickey Adams only scraped a half point against Hydra, but he made the mistake of trying to play normal chess and trying to win, at least at first. I also think Hydra is significantly stronger than anything on commercial hardware.

Not that Kramnik will have an easy time of it. Fritz has four years of programming and chips that are four or five times faster than the ones it used in 2002 in Bahrain. I was at that match running the website and Kramnik made it look easy at the start. But one poor decision and one blunder were enough to level the match. In Bonn he'll have a rest day after each game, an important detail. Playing classical chess against a machine is exhausting.

Perhaps the biggest news is that there's a big new sponsor for the event, the German energy trading giant RAG.


The German cocern should have used it's money to finance an additional quarter of the Topalov-Kramnik match raising the number of games to 18. Instead the chess public is preasented with this stupid non-event.

Or used it to sponsor all of the candidates matches all the way through to the final..

lets be fair here.

the match between fritz and Kramnik was announced before the Topavlo Kramnik Match so they could not raise the number of games from 12 to 18 at the time.

also at the time this was announced there was no idea that the candidates matches would have a problem. Not sure there is a problem with the candidates matches. I thought that Kirsan said it would all be taken care of immediately after the election. so it certainly must be all taken care of by now. I thought I remember reading in some of the threads here by posters, that Kirsan likes to make promises and never fulfills the promises.

by the way Mig. it was painful for me to read that Ali Nahant is so supportive of Kirsan now. Whatever happened to the new world chess federation. Mig, How about an update on what Kok has been doing since the election has ended.

I guess we can infer, from the tone of Mig's remarks, that Freidel is not hiring him to cover the Kramnik-Fritz match.

We ought to be careful before we start bashing chess sponsors for not putting their money into more worthwhile causes. They might take our advice and send it for humanitarian relief in Darfur.

Ah, thank goodness the koster troll is here to tell me what I think again. What is the tone of my remarks?

There will certainly be some negative statements coming, however. I went through the rules for the match yesterday.

I wasn't being serious about their sponsoring the candidates matches, Frank. Just working in a mention of them, RIP.

You can't expect Ali and other organizers and FIDE officials to abruptly boycott FIDE because Ilyumzhinov is still in power. You don't want to hurt your local players just to spite Kirsan. It would be nice to see them try to carry out a progressive agenda outside of FIDE, however. There are several pots bubbling regarding new organizations and such but I can't run the few existing details yet. I'll ask around and see how much people are comfortable releasing. It won't be about entirely replacing FIDE.

Some of those rules will certainly trigger debate.

"If the position is evaluated by the tablebase as a draw, the Operator will inform Mr. Kramnik of that fact in the presence of the Arbiter. This will constitute an offer of a draw."

Huh?? On the other hand I seem to recall Hydra-Adams used this as well.

Has the human been allowed to adjourn games in earlier matches? Here it seems Kramnik can do so after 6 hours or 56 moves.

Mig thanks for letting us know that at least some ideas are still floating around fide. that makes me feel better.

right now my attention has returned to home and the USCF mess with sam sloan. I guess there will be a vote to decide that situation on Sunday.

I would love to see USCF put up near the top of its priorities the finding of professional companies to sponsor chess. that will require a very professional approach and very professional personnel of impeccable reputation. Obviously the board of directors must be a shinning example to everyone.

I believe if USCF hired a top-tier global search firms such as Egon Zehnder International, Russell Reynolds, Spencer Stuart, Heidrich & Struggles and Korn/Ferry to find a board of director, they would never present sam sloan for consideration. the uscf needs to have a board of directors that these firms would present for consideration. the uscf needs to understand the proper type of person that should be a member of the board. and exclude all others. Beatrice Marinello is another example of a loose cannon on the board.

I am not saying anything personal about sam or beatrice. I am simply saying that all members of the board need to behave in certain ways and these two people do not behave in the appropriate way and therefore must be removed from any board considerations.


what don't you like about that particular rule? I think it's a pretty normal step if you want to see who _plays_ better. For as long as the comp calculates, something can happen beyond it's search horizon that the human might see and work with (or, more likely, the other way around ;o) ). Once tablebases are reached though, only the human can err while the comp hasn't even the _chance_ to do so. That's a no-win/loose - situation, thus, unfair.

The fact that Kramnik can adjourn games is a significant advantage on paper but if you look back at Bahrain 2002 the longest game was 57 moves, the second longest game was 51 moves, the third longest game was 41 moves and all others were decided in the first time control. So I am not sure he will be able to take advantage of this privilege.

As a Kramnik fan I am very pessimistic about the outcome of this match.

Is there information about the hardware Fritz is going to use?

Albrecht, the engine has its advantages, the human has his/hers. It doesn't make sense to me to compensate for any of these advantages to make it more "fair".

I don't think Kramnik gets an automatic win in a won tablebase position though. That might have been even more absurd. (But he gets to know that it is theoretically won.) But even in a drawn position it is anti-chess to not have him prove that it is actually drawn. Anyway, it's not likely that any of this will actually be realized. Especially as only 5-piece tablebases are allowed.


did you read that chessbase report on that gelfand vs. jobava game in Dortmund? Do you really want to tell me that such an ending, where Kramnik would blunder the theoretical draw to fall into a mate-in-68 would have anything to do with what a man-vs-machine match is about or supposed to show??

oh, and my above point was, that tablebases are _not_ an advantage of the engine. Because the engine is no longer in use when it plays out that 255move win in a r+b vs. 2N ending. The engine by itself probably _would_ draw that...
Allowing the engine tablebases is like granting Kramnik access to "Basic Chess endings"; it has nothing to do with the strenght of the engine (which - at least in my understanding - would be determined by how accurately it evaluates positions), because the engine is no longer in use once a tb-position is reached.

"did you read that chessbase report on that gelfand vs. jobava game in Dortmund? Do you really want to tell me that such an ending, where Kramnik would blunder the theoretical draw to fall into a mate-in-68 would have anything to do with what a man-vs-machine match is about or supposed to show??"

Assuming your argument is valid, a pure man-vs-machine match would not give ANY of the parties access to tablebases. These rules give BOTH access. (Although not in the same extent. Would you seriously want to see Kramnik awarded a win for free as soon as he reaches a won tablebase position?)

I still disagree however. Subtle semantics aside, tablebases is one thing that does give Fritz some advantage to the human. It's not "unfair" in the slightest, it's simply its equivalent to Kramnik's memory. A pure calculation contest is not possible and that's not what chess is about anyway.

"Assuming your argument is valid, a pure man-vs-machine match would not give ANY of the parties access to tablebases."

Indeed. And that's what has been proposed repeadetly for a few year now, or at least after the Hydra-Match.

"t's not "unfair" in the slightest, it's simply its equivalent to Kramnik's memory."

Nope, it is not. It is equivalent to Gelfand pulling an extremely expanded version of "Basic Chess Endings" out of his pocket, placing it on the chessboard, getting up and saying to Jobava "Well, this is a theoretical mate. I leave the book here for reference in case you don't know how it is reached, but since I don't need to think anymore but only follow the moves from this book, I might as well try to yet catch a decent Dinner. You'll be able to figure it out for yourself."

You can disable tablebases in Fritz without impairing his calculations. You can not disable his search algorithms without impairing his calculation abilities. Neither can you "disable" the memory of a chess player without imparing his chess playing abilities.
Tablebases are an "add-on" - they are _NOT_ a part of the chess engine Fritz.

Kramnik will play 1 Nf3 and DeepFritz will trot out a supertablebase informing him that the position is drawn.

"It has nothing to do with the strength of the engine (which - at least in my understanding - would be determined by how accurately it evaluates positions), because the engine is no longer in use once a tb-position is reached."

It's all part of the computer program, so unless you want to narrowly define a chess program as one that does weighted scoring of a search tree, computer use of tablebases for endgames and openings seems fair to me. The computer is "evaluating" the position using its algorithm, which consists of looking the position up in a table. It only happens to be doing it perfectly instead of "guessing". It's just like the computer had hard-coded solutions to specific problems; there just happen to be a lot of them. The computer isn't "thinking" any more when it doesn't use tablebases then when it does. It is executing a fixed algorithm in both cases.

hope the deep fritz programmers do a god job of selecting its openings. No closed positions or endgames.. maybe Budapest Gambit (defence?) with black and Kings Gambit with white (no Petroff or Berlin) or maybe better still, play the Albin Counter Gambit.. contact Morozevich


what you say is true, but misses my point.
My objection is basically, that we do man vs. machine matches for a certain purpose: We want to know how human understanding fares against computer calculation.

I think we agree on that?

But then it simply doesn't make sense to match a human against something playing perfect, be it chess or some dart-throwing machine that always hit's bulls eye. There's nothing to be gained by such a match up because, naturally, the human must loose.
I guess, that's part of the reason why not even John Nunn has shown any interest in a "Man vs. Tablebase" match, which, given this argument:

"It is executing a fixed algorithm in both cases."

wouldn't seem implausible at first.

Different programs have different search algorithms and therefore reach different strenghts. And the programmers can take pride in making their search algorithms even stronger, implementing more "chess knowlege" etc. But how are you going to take pride in your special version of retrieving information out of tablebases???

You are right, from a technical point of view, nothing more... ;o)

What they could take away from the computers in these matches is their opening-books.

After all, the tablebases were created by computers, but the opening books come from more than a century of human play. Let Fritz start calculating from the opening position!

more seriously, the other night I watched the movie/documentary about the historic GK/Deep Blue match. I forget the name, but it shows that the circumstantial evidence is overwhelming that IBM intervened after game 1 and for rest of the match.

These kinds of things can be determined indirectly. there is no doubt in my mind in the locked back room a GM occasionaly told Deep Blue to abandon some lines. It was a serious form of cheating (that ultimately raised IBM stock value).


I believe that the computer logs released after the match established to the satisfaction of most folks that there was no tampering during the games.

The Deep Blue article in Wikipedia has lots of interesting links, if you're interested in learning more about the match.

Greg -

Thanks for the tip. But I cant help notice the Wikipedia article's links all seem to be IBM-centric. Maybe 'most folks' should see the documentary before making up their mind (it's from GK's POV but isn't dogmatic).


Maybe Wikipedia has a skeptical bias generally. See their "Roswell UFO" article.

Twitter Updates

    Follow me on Twitter



    About this Entry

    This page contains a single entry by Mig published on August 11, 2006 11:19 PM.

    Rising Stars vs Failing Eyesight was the previous entry in this blog.

    06 Montreal Empresa Internationals is the next entry in this blog.

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