Million Dollar Man vs. Machine Chess Challenge

Statement by BGN Match Director Raymond Keene and Rules Commissioner Eric Schiller regarding the "10 Questions". Dated: 9th September 2001.

The rules for the Brains in Bahrain event were developed by Raymond Keene over the years and as a reaction to the unfair conditions under which Garry Kasparov had laboured in his 1997 match against IBM's Deep Blue. Because there is a lot of money at stake, it was important to create as level a playing field as possible for this historic event. The details were worked out by representatives of the Fritz Team and Kramnik Team in coordination with Eric Schiller, our Rules Commissioner. The Bahrain rules were an attempt to level the playing field. We may not be perfect, and are still making adjustments, but these were our honest conclusions for a genuine test between man and machine.

An article on the Club Kasparov [the precise name is kasparovchess.com MC] website posed ten questions on the match conditions. We are happy to provide answers to these, and any other questions which may arise. Sometimes questions from the public or press may address an area which has escaped our notice, in which case we can adjust the rules accordingly, but the following questions were already dealt with long before they were proposed. Specifically, we should point out that weeks ago the question of opening book modifications were solved with exact terms and wording. So rest assured that we are on top of the issues and are happy to reply to the questions raised by Club Kasparov.

1. Is it true that Vladimir Kramnik and his team will receive (or have already received) a copy of the Deep Fritz program that he will play against in Bahrain?

Yes, he has received a copy which was dispatched in the first week of August to his second, Grandmaster Miguel Illescas, in Barcelona. The final version will be shipped to him this week, in accordance with relevant agreements.

2. Is this exactly the version that Kramnik will face, or can ChessBase modify it before the match? If ChessBase must submit the final version, as rumour has it, what will be done about any errors discovered after the program is submitted?

The program must respond to a suite of test positions proved by the Arbiter in a manner consistent with pre-match testing. Any discrepancies would have to be explained to the satisfaction of the Arbiter and Kramnik Team.

3. We heard that the program's opening book can be modified. What percentage of the book is it possible to change in so little time?

Brain Games have agreed with the world champion that the ChessBase team will be allowed to add no more than ten ply to the openings book between games.. In addition, it can change the weighting of lines so that another variation, which is already in the pre-match book, is chosen. But no new analysis (outside the 10 ply per game) can be added. That is an insignificant percentage.

4. What is to prevent Kramnik and all his GM assistants from finding literally hundreds of games in which Fritz can be beaten, and Kramnik repeating eight of these games in the match in Bahrain?

Because the Fritz Team can choose a different line of play in each game, this might work once, but is hardly worth the effort and will offer no advantage in the other 7 games.

5. The eight-processor machine Deep Fritz will be running on in Bahrain is probably faster than anything Kramnik will train with, but can't a slower machine imitate a faster one just by giving it more time to think on each move?

Why assume that World Champion Kramnik is testing on outdated equipment? He can get hold of the same kind of machine Fritz will use, if he chooses to do so. On the other hand, he knows that this is just a matter of speed. Different machines should come up with the same move, but it may take much longer on a weaker machine. The Kramnik team just has to give the machine more time per move in testing. It isn't for us to tell them how to optimize their preparation! Kramnik will win a MILLION DOLLARS if he succeeds. Investing in a little hardware hardly seems to high a price to pay. Even if he loses, his prize is very substantial.

6. Is it true that there are a number of amateur chess players, rated around 2100, who would be able to beat all top programs under similar conditions?

Since the computer team can change opening weighting, it would require superhuman preparation. Not going to happen in the next few years!

7. Is ChessBase allowed to modify the openings or any program parameters between the games in order to avoid a repetition of games that Fritz has lost in previous rounds?

Of course! But just the opening book, and it must choose another line that already exists in the opening book as delivered before the match, can't add new lines. After each game 10 additional ply of analysis may be added.

8. Will ChessBase take a team of programmers to Bahrain if they are not allowed to do anything during the three weeks they are there?

Of course the tech team is needed in case of some non-program related bug and in any case such teams are generally present in other individual competitions. Consider the comparable sport of race car driving. There are rules governing adjustments they can make to the cars, and the rules are enforced. It is the same here. Chessbase deserves the publicity it will reap, and will certainly want the programmers to be interviewed on major media.

9. Does ChessBase believe that Fritz has a ghost of a chance of winning, or even drawing, the match under the current conditions?

The Match Director and Arbiter aren't supposed to offer predictions. However, both are on record as expecting a very close contest.

10. Why did ChessBase agree to play a match under such conditions? Is ChessBase satisfied to take the worldwide publicity, even if the games are essentially pre-prepared by the opponent and their program is massacred?

We believe we have made it clear that this scenario is impossible. It is as relevant as asking how we will act if a meteorite crashes onto the playing board. Any businessperson knows the value of worldwide media coverage and publicity not just for Fritz, but for the entire product line. Fritz is demonstrably the best program around. They have proven it time after time. Fritz's performance will no doubt delight and impress chess fans whether it wins or loses. Let's face it, you can't buy a Vladimir Kramnik for under 50 pounds! Fritz has trounced professional players all over the world. Most use it for training. Regardless of the result of the match, Fritz will be shown to the buying public and we expect that all chess fans will want to run out and buy a copy, even if it loses to the World Champion. However, why do you think that Kramnik can defeat Fritz convincingly, when no other Grandmaster has? In any case, ChessBase are going to benefit greatly from the match, and deservedly so.

This is the first contest between man and machine which is designed to be fair. We are all doing our best, knowing that in the future there will no doubt be some aspects of our experience on which we can improve. Discussion of these issues is interesting to the chess community and public at large, and during the event there will be ample time for discussions and in-depth interviews, so that each item can be explained. Right now, as surely you can appreciate, our goal is to simply make sure all the rules are in place and let both teams concentrate on their chess skills.

We look forward to an exciting match that will pave the way for regularly scheduled contests between the best human player and best chess program.

Journalists with questions pertaining to the match should contact: John Fioravanti [john@marstonwebb.com] or Hector Proud [hector@ideageneration.co.uk] or John Henderson [jbhthescots@blueyonder.co.uk]

For further information on the Man vs. Machine: http://www.brainsinbahrain.com/