Man vs. Machine Challenge:

The True History of the Kramnik vs. Fritz Match

by International Arbiter
Eric Schiller

October 2002

Man vs. Machine Challenge: The True History of the Kramnik vs. Fritz Match

by Eric Schiller (author@chessworks.com)

As Vladimir Kramnik does battle with Fritz in the battle of Bahrain, scheduled for an October 2002 start, many will no doubt be interested in how this match came about, and how the participants were chosen. The Human representative, Kramnik, was automatically invited by virtue of his holding the traditional World Champion title, which he earned by defeating Kasparov in their 2000 match, where I participated as arbiter.

As soon as the BGN World Championship match between Kasparov and Kramnik had concluded, Ray Keene asked me to begin working on the Man vs. Machine Million Dollar Challenge (as it was then called, at my suggestion). It wasn't until the end of March, however, that the organization of the qualification and final match got down to details. The qualifying event turned out to be quite a controversial affair. I felt to withdraw as arbiter for reasons which will be presented in Part One of this document. That did not affect my status as arbiter of the final, and I developed a full set of rules for the match, in cooperation with both Team Fritz and Team Kramnik. That will be covered in Part Two. However, events of September 11, 2001 forced a postponement of the match. During the interval, BGN sold the rights for the match to the Einstein TV group, who hired Malcolm Pein to take on the role previously occupied by Ray Keene. I was again invited as arbiter, but my participation was vetoed by Team Fritz, as will be discussed in Part Three.

Part One deals with the controversial qualifying tournament held in Cadaques, Spain,. to choose a challenger for Kramnik. Because many questions have been raised about this event, I am going to refrain from presenting any prejudicial commentary, though I may do so at some time in the future. Instead, I place in the public record a large set of internal emails. I should point out that I never received any compensation for my work, and am under no obligation to keep this material from the public. It is certainly in the public interest to make this material available, and however one might judge the actions of each individual, I think that all the correspondents behaved very well. Indeed, even after I withdrew from the qualifier I remained an integral part of the team working on the big final match, so obviously we felt confident that we could all work together. It is not my intention to disparage any individuals involved. Each has made very significant contributions to chess and our disagreements over matters of policy and philosophy should not be taken as personal animosity.

Part Two contains email documentation and all the drafts and final agreed form of the rules for the planned Bahrain match in 2001. These are NOT the rules under which the match is being played. My rules were replaced when I was. However, it is a fascinating set of exchanges and contains discussions of many tricky questions. Anyone interested in human vs. computer mind sports will learn a lot!

Part Three is simply a matter of putting the facts of my non-participation (including lack of compensation for any of the 20012002 work!) on the record. This may not be of particular interest to most folks, but I want to complete the tale.

Because this document is intended as a record, I have made it available in secured PDF format. Although I retain copyright to the overall document, you may use as much of the text as you deem appropriate in discussions of the event. If scholarship or reporting requirements demand an exact copy of some original email with all headers and so on, please contact me by email at author@chessworks.com and I'll see what I can do.

Part One: How the qualifying event for the challenge was arranged

I have presented all of the emails I deem to be relevant, and have omitted emails which contain only discussion of travel arrangements, legal issues involving a previous Braingames CEO, and other unrelated matters. I have not edited any of the emails except to run them through a spell-checker, and I have not modified Ray Keene's typical single case letters, except to correct names. I have deleted quoted messages identical to the ones presented. I also deleted personal information such as mailing addresses and telephone numbers. I believe this presentation will allow readers to come to their own conclusions and perhaps avoid swallowing any of the rumors currently circulating on the Internet.

The emails were sent from different parts of the world, and therefore the timestamps aren't a good clue to the order in which they were written, but that was all I had to work with. So some may be out of order. However, I have placed the emails into sections by date, so if you read a full day's worth everything should be clear.

In this part, documents from March and April 2001 are presented, with a short followup from May.

Cast of Characters:

Bertil Eklund: Swedish computer chess specialist

Byron Jacobs: Keene's assistant, handling technical decisions on the events.

Enrique Irazoqui: Organizer and arbiter of Fritz vs. Junior qualifier. Replaced Schiller as arbiter for Bahrain.

Eric Schiller: Arbiter of Kasparov vs. Kramnik, appointed by Keene with agreement of Chessbase and Kramnik to be arbiter of match in Bahrain. Also appointed to be arbiter of Fritz vs. Junior.

Frederic Friedel: ChessBase, distributor of Fritz and Junior but not programs which were refused participation in qualifier.

Ossie Wiener: Represented Shredder in negotiations to include that program in qualifier

Raymond Keene: Braingames.net Games Director

Vincent Diepeveen: Programmer of Diep, a program which was not allowed to compete

22 March 2001

Sent by Raymond Keene

thanks to Enrique for a prompt response!! We must clearly add a prohibition against the computer winner competing in another man v machine match before Bahrain starts-Byron pls note to put this in the revised contract! Further points- 1 I would like to invite Enrique to be one of the two arbiters in Bahrain for the main match- Schiller will be the other-there will be a modest fee plus covering of all your expenses and a stay in a fabulous hotel with multiple swimming pools restaurants and amazing beaches 2 I want to visit Cadaqués one weekend during the qualifier-what is the recommended hotel (i.e. best-most picturesque most beautiful) and which is the best way to get there from London? I seem to recall the Salvador Dali museum is there-is it open to the public ? 3 the reason for stopping end April is that Kramnik wants to see fifty games against grandmasters by the winning program well before Bahrain starts- which I think is fair-I wanted to give enough time for this to happen-of course we can slip into may by a few days maybe even one week but time is a bit short 4 however-I think it important we don't start the qualifier until April 16-we must get the main contract and press release issued first-this is

scheduled for April 15.the notice on the BGN website states nothing more than our policy to hold such a match-I don't regard that as a leak! 5 also I wd like to invite Eric Schiller-one of our arbiters from wcc London 2000-also officiating in Bahrain- over to be in Cadaqués and he cannot make it before April 16-I want him to be involved in writing the final set of rules with Enrique! 6 what is this tournament Shredder just won ipccc Paderborn or something-pls inform me!! 7 is there any problem calling our event in Cadaqués the BGN world computer championship? I hope not!! To summarize-lets start on April 16-it's the day after the big press con and I think it will make a big splash-if the event goes over the end of April I don't mind so long as we still have time to provide Kramnik with his specimen games. A speedy response would be appreciated thanks ray

Sent by Raymond Keene

we want the event to be officially known as the brain games world computer championship! I trust there is no problem with this. Byron Jacobs will officially represent brain games in all questions. it is imperative all entrants sign the brain games contract before the start of the event. I will forward this to you soon and I wd hope that Enrique will be able to get everyone to sign- we have agreed with Freddie Friedel that there will be an entry fee of $us 2500 per computer to go towards the organizational costs-finally I understand that Shredder-deep blue deep Junior and Fritz are the four invitees. there should be no press announcements until I have signed the agreements with both Kramnik and Bahrain both of which I anticipate early next week. so long as the event finishes by end April we have no time trouble!! best wishes ray Keene brain games games director

Sent by Raymond Keene

I just emailed you a bunch of forwarded messages from me and Enrique the local organizer-looks like we can start April 16 which is good for you-make sure you pick up the forwarded messages don't screen them out as potential cyber threats!! you will need to write the rule book for Bahrain while you are in Spain! you will get all the contracts to help with background! can you stay in touch with Byron and Enrique to find out where to stay and how to get there! Ray

Sent by Raymond Keene

I have no particular problem with any of this- while some of it I regard as extremely positive news e.g. the hotel situation! BGN is planning to install the Staunton trophy in Simpson's on April 11, and faute de mieux we also could have made that the press announcement day for man v machine. if April 15 is generally acceptable in Bahrain I think that wd be better! we can start the computer qualifier in Spain on April 16 which then becomes ideal timing. many thanks to Justin for a lucid exposition-I hope to sign the Kramnik contract over the weekend!! Ray

Sent by Eric Schiller

I went through them all. Timetable is very tight. What time does the first round start? I need time to get from Barcelona, which is probably the closest airport. I'll check flight schedules. I'll do that ASAP. Doing the rulebook for Bahrain in Spain is no problem. I presume rules for this qualifier are the same used as Cadaqués tournament, but I need to see a copy. I'll be in touch with Byron and Enrique re arrangements. You specify four programs, but if the winner of the current computer event isn't among them, that might be strange. I haven't checked results recently, will do so soon. Preparing a website for covering the games should be a high priority. That's what the spectators will be watching. Eric

23 March 2001

Sent by Raymond Keene

we are working on all of the above-can you tell me flight costs-does not matter if you are not there at the start-important thing is that you and Enrique who I have invited both to be the two arbiters for the Bahrain event should emerge at the end of Cadaqués with a perfect rule book for Bahrain we can discuss details later after you've seen the contracts-as I said liaise with Byron and Enrique re logistics and well speak soon once the actual contract with Vlad is signed which I hope is imminent! Ray

Sent by Eric Schiller

Glad to see you don't need me at the start, flights are difficult (and VERY expensive, lowest quote I got so far from Dallas is over $3000!). I worked on them for an hour last night, nothing satisfactory. If I don't have to be there on the 16th, it is a bit easier, I can leave 15th from Dallas (Maybe, all flights early, I must be in Dallas until end of final round)). I'll try to sort something out today and give you details. What will I be doing other than the rule book, will I be working as arbiter too? More later. Eric

Sent by Eric Schiller

I am trying to make travel arrangements, but they aren't easy. I must be in Dallas through Sunday (April 15th) afternoon until the last round of the Dallas invitational ends. I do not know when that is, but the latest I can leave Dallas and get to Barcelona next day seems to be about 4:00 PM. I don't think that will work, especially since I don't know where I will be in Dallas and how far it is from the airport. All arrangements for the Dallas trip are being made by Stanford University (I will be coaching the team). They were planning on flying me back to SF with the team on Sunday night. From his latest message, it seems I don't need to be there on the 16th. Can you confirm this? Looks like I may have to overnight in Dallas and fly on the 16th, which would get me in on the 17th. I also have no idea how to get to Cadaques. The guides I read that it is a few hours. Will a car be sent to Barcelona? The ending date of the event is not clear to me. Should I be looking for an open ticket or do you have some final date? When I get the details, should I send them to you for ticketing? Please send me as much information as you can ASAP. In addition to the travel arrangements I must also alter my teaching schedule and find substitutes for the schools. Is the format of the event set? The participants? I'm still pretty much in the dark here. Finally, is there anything you want me to do re website technology before I arrive in Spain?

Sent by Raymond Keene

Byron will answer your questions-he will put you directly in touch with Enrique who is the local organizer- Byron-pls put Enrique directly in touch with Eric-thanks- ray

25 March 2001

Sent by Enrique Irazoqui

Hi Eric,

Greetings! Ray Keene suggested I get in touch with you regarding the upcoming computer qualifier in Cadaques. I have no information at all, so far, so could you send me all details that are complete? Dates are not clear yet, but so far it seems it will take place between April 16th and May 7th. Three programs will participate and play 20 game matches all against all at 40 moves in 90 minutes on 2 dual Pentium-933 machines. I must arrange flights, which is turning out to be a bit tricky as I am in Dallas on the 15th and must stay at least until 3 PM. I can arrive on the 16th sometime in the afternoon (Barcelona). What is the best way to get to Cadaques? I assume that Barcelona is the nearest airport, but let me know if there is something closer.

Barcelona is the closest international airport. Once in there you have 2 options: take a train to Figueres and I'll pick you up or else take a taxi, as I usually do. The taxi will cost about $150.

When does the event end? I must advise clients and students here and reschedule.

Ray indicated end-of-April/early May, but no specifics. As I said, we are not sure yet, but in any case the tournament should be finished before May 7th.

What is the level of formality of the event (should I bring suits or something less formal?)

Zero formality, please! :) I have never seen a necktie or a suit in Cadaqués. I always go around with blue jeans.

I understand that my major task is to complete all the rules and regulations for the Kramnik match, but assume that I'll be working in several capacities (as usual with Ray), including arbiting, press and internet coverage. However, I await Ray's (and your) guidance.

We'll talk about all this here. I understand that you and I will be the arbiters, but I don't know anything specific about it.

Is the local language primarily Spanish or Catalan?

Either, we are all bilingual. By the way, I never spoke so much English as in Cadaqués, not even during my 14 years in the States. You won't have a problem.

I always try to learn a bit before I travel. My Spanish is not very strong (English, Russian, German, Cambodian and a few others are my main languages) but I can understand a bit.

Don't worry, really. I speak Spanish, Catalan, French and Italian, and I mumble some English too. We could open a translation agency. :)

Is there broadband access in Cadaques?

Nope. If you are lucky, you will be able to connect at 56k, more often at 45k. It's driving me nuts.

I will need to access some updates on development projects that usually come in the form of 25 Meg downloads.

Impossible. If there is no broadband, I'll need an address where the developers can send me CDs. My address: <DELETED>

I have a laptop with USB and Ethernet capabilities and will have all my web tools with me should anyone need them. If you need any background on me, www.chessworks.com/Schiller/Schiller.htm should have everything you need. My main chess site is www.chesscity.com, which I operate for Cardoza Publishing. I'm also setting up WorldFamilyChess.org in anticipation of the World Families Championships in Hawaii next year. I look forward to seeing you next month! Eric Schiller International Arbiter

Please let me know the date of your arrival so I can make reservations. By the way, the 16th of April is still Easter holiday in here and hotels will be packed. Arriving the 17th is safer. I am very much looking forward to meeting you soon. Kind regards, Enrique Irazoqui

Sent by Eric Schiller

Below are the responses I got from Enrique. It seems that the event may not start on the 16th, and he advises that the 16th could be difficult re Hotel. Of course I can't book a return flight until the dates are set. How are the travel arrangements to be made? Should I simply pass on the flight options and let Byron arrange tickets? Or do you want me to price them and have you pay for them? If I have to go to work immediately on arrival the overseas portion (at least to the event) should be premium economy if at all possible, otherwise I'll be a mess for a day or two. Return flights and intra-Europe no problem. So far, the computer has been spitting out fares so high that it wouldn't make much of a difference. Haven't heard from Byron yet. I still don't understand why so many games, such a long time, for a mere qualifier. Wouldn't 10 games do as well as 20 for our purposes? These long tournaments are hard to publicize. The ongoing Odyssey tournament (which you presumably know about) is not even a blip on the radar. The final match is only 8 games, so what's with the marathon? And if there are only 4 teams (Gambit-Tiger is going to be rightfully miffed for being left out!), why not Semi-final and final match, 8 games ? Much more dramatic and easier to publicize. And only 16 games/days. Just my .02 euros... Eric {COPY OF PREVIOUS MESSAGE}

Sent by Eric Schiller

!@$%!$%! Outlook was just supposed to digitally sign things. Here is what I sent: Of course I am also using MS Office 2000 Outlook, which must be why the HTML doesn't work between us, though it does everywhere else! I realize that the in-crowd of computer people recognize that short matches are less reliable, but having a 20-game qualifier for an 8-game final is still illogical, IMHO. I see the long match as a potentially dangerous precedent for BGN. I would have suggested a higher entry fee to Ray had I known the qualifier would be longer than the final! I just don't see anyone but a computer freak following the action for 20 days. But in any case, why just one game per day? But it is BGN's call, and once the call is made I just get to work. Travel plans are starting to come together. I may return to SF from Dallas and then go to Barcelona via London, but in that case will arrive the 17th, which makes the hotel situation easier, right? I'll need a reservation from then until the end (whenever that turns out to be).

Sent by Enrique Irazoqui

>> I realize that the in-crowd of computer people recognize that short matches are less reliable, but having a 20-game qualifier for an 8-game final is still illogical, IMHO.

In computer chess, short matches are like Russian roulette. Are you familiar with the SSDF list? They play thousands of games to make sure, and some people still argue that it is not enough. :)

I see the long match as a potentially dangerous precedent for BGN.

Not necessarily, provided that we keep differentiating computer chess from human chess.

I would have suggested a higher entry fee to Ray had I known the qualifier would be longer than the final! :)

Good idea in any case. I just don't see anyone but a computer freak following the action for 20 days. You are probably right, but among other things it is a matter of choosing the very best program to play Kramnik. Computer chess needs many games.

But in any case, why just one game per day?

I expect to get 4 games per day. Three programs will participate, playing each other 20 game matches, a total of 60 games in 15 days, hopefully. Games will be played 24 hours a day automatically.

But it is BGN's call, and once the call is made I just get to work. Travel plans are starting to come together. I may return to SF from Dallas and then go to Barcelona via London, but in that case will arrive the 17th, which makes the hotel situation easier, right? I'll need a reservation from then until the end (whenever that turns out to be).

The Rocamar hotel is more beautiful, has a nice beach and a decent restaurant, but it is a bit out of town, about 15 minutes walk. The other option is the Playa Sol hotel, just in town. For a few days I would recommend the Rocamar, but for a longer period of time the Playa Sol may be more convenient. You could rent a car, but it is not the best way to move around. I find mopeds ideal here. Let me know. Enrique

Sent by Eric Schiller

I don't think a tournament is a rating list. Whatever is needed for accurate rating is irrelevant to qualifying, and eliminating luck has never been part of sport. As long as there is a level playing field, even short matches work fine. Granted, FIDE's 2-game knockout formula is silly, but a qualifier with 6-game matches should be fine. Of course going 24 hours per day makes for more games in a short period of time, but it seems to me that going more than two weeks is just way outside the media attention span. The legitimacy of the qualifier is already under question because only a select few were invited to participate, and many strong engines were not. In any event, in my experience the programmers all find something to complain about anyway. Those complaints hold whether the match is 5 games or 50. Kasparov might well have defeated Kramnik in 24 games. What are the web broadcast arrangements? If the games are going 24 hours, is an arbiter present all the time? Are there any written rules and regulations for this tournament? If so, please send me a copy. As for the hotel, I suppose that depends much on the weather. A 15 minute walk doesn't bother me, if the weather is nice enough to make use of the beach. I've heard there is some good snorkeling around there somewhere. I spend a lot of

time in the waters of Hawaii. But if the weather is too cold to swim, then the beach holds less attraction. I'm not much into bicycles, mopeds, scooters. But I don't want to rent a car, too much bother. Where is the tournament played? I have good links to Cadaques web sites and can probably locate everything online, then decide. Eric

Sent by Enrique Irazoqui

>> I don't think a tournament is a rating list. Whatever is needed for accurate rating is irrelevant to qualifying, and eliminating luck has never been part of sport. As long as there is a level playing field, even short matches work fine. Granted, FIDE's 2-game knockout formula is silly, but a qualifier with 6-game matches should be fine.

In human chess, yes. In computer chess I disagree, unless there is a large number of candidates, which is not the case. 6 game matches among 3 programs is not better than FIDE's formula. Ideally, we should have played pre-qualifiers and qualifiers, something like the Interzonals and Candidate tournaments of the sixties. But time was short and the only way out was picking the very best programs and do the best possible job.

Of course going 24 hours per day makes for more games in a short period of time, but it seems to me that going more than two weeks is just way outside the media attention span.

It should last just 2 weeks.

The legitimacy of the qualifier is already under question because only a select few were invited to participate, and many strong engines were not.

Which ones? Considering that this tournament will be played on dual machines and that in the man-machine event the program will run on a multi-processor machine, there are no candidates other than Deep Fritz, Deep Shredder and Deep Junior. All the others are as far from these 3 as Kasparov from a 2500-2600 player.

In any event, in my experience the programmers all find something to complain about anyway.

Yep.

Those complaints hold whether the match is 5 games or 50. Kasparov might well have defeated Kramnik in 24 games.

Yes, but you can't play a world championship in 4 blitz games. It must be believable.

What are the web broadcast arrangements?

No idea.

If the games are going 24 hours, is an arbiter present all the time?

I hope not! You and I are the arbiters, and as far as I know this will be played home, in a small room. Worse: I sold my house 4 months ago and provisionally I live in a small flat. This can create some problems of logistic.

Are there any written rules and regulations for this tournament? If so, please send me a copy.

None whatsoever, aside from:

- identical dual Pentium-933 machines with 768MB RAM

- 3-4-5 men tablebases

- 20 game matches

- 40 moves in 90 minutes

As for the hotel, I suppose that depends much on the weather. A 15 minute walk doesn't bother me, if the weather is nice enough to make use of the beach.

Weather here is usually fine, but the sea is still too cold in April. There are a few tourists that dare to swim in Spring...

{irrelevant questions on snorkeling and hotel omitted}.

Enrique

Sent by Eric Schiller

Ray, I'm confused by a lot of this. 1. I thought you were NOT going to go to dual-processor ready machines. This limits the entries to just a couple of machines, not necessarily the best. It is hardly a qualifier in the true sense, and excludes 99% of the programs. Bad for PR, piss off most developers. I think this has even less legitimacy than hand-picking the challengers. Why not just use the computer rating list? I am especially concerned that the assumption is that only 3 programs support multi-processors. I know of at least one other, a developer who has been nagging me about how to enter. Problem is, I don't want it to play because it would create potential PR problems since I am using that engine (in weaker form) in a commercial product. What efforts were made to reach potential players? 2. No broadband, no website local. How on earth is anyone going to see this? Phone dial-up from Spain to England, 24 hours per day, to feed moves? I don't get it. 3. The event will be held in Enrique's home? What will the media make of that? 4. Overall, I think you will have a hard time selling this as a legit qualifier. The Odyssey tournament looks much more impressive. 5. I understand the rush, but have real reservations about how this is going to make BGN look. There will be accusations of deals with Friedel to insure Fritz plays World Championship. I hope you have given some consideration to the PR aspects, because at the moment, this whole thing looks very bad from many points of view. With established events like Paderborn and Odyssey, plus a rating list, this "qualifier" looks more like a stunt to justify one particular engine. And use of dual-processor reduces the relevance of the event to normal people, who don't have such things. You may well wind up with a Junior vs. Fritz match, both sold by the same company. The controversy won't be good in this case, especially if Gambit Tiger finishes ahead of the winner of our event in the other major tournaments. Eric

Sent by Raymond Keene

Byron-can you look into Eric's points in the email he sent tonight-we have not issued the invites yet so everything is still possible-also has Eric looked at the strange USA people who constantly send me emails-who are they??? do they have a real machine? are deep Junior and Fritz both distributed by the same company-I want it to appear fair and open not a stitch up- but lets face it though-it doesn't matter where we play it _Enrique's house or not-this is a website event not a spectator event-what are the validity of Eric's points about website coverage-I thought Friedel had all sorts of super website ideas! pls sort this amongst yourselves since I am not a computer expert and I am also about to go to china-all I want is to sign the Bahrainis on April 15 and know that a credible qualifier is going to start next day!! ray

Sent by Eric Schiller

I'll keep in touch with Byron, but yes, Junior and Fritz are distributed by Friedel's company, in almost identical packaging. Here are a few points you should consider. I agree, a website event can be anywhere, but when people get to writing about the qualifier, and they will certainly look strangely at en event which is essentially a match between two programs with one shared commercial sponsor who also seems to be involved in the website presentation to the public and press, held in an out of the way place in someone's house. The last point may be irrelevant, as you say, but it adds to an ugly picture. Now if the event really is open to say, 8 programs, running on a single processor or dual (placing those without support for dual processors at a terrible disadvantage), and includes the top programs as invitees (even if they decline), then I see the problem as minor. I don't know how much money is being put into this, but it does seem overkill to have two arbiters on site the whole time. I'll need a week to work out all the details for Bahrain, no more. Of course I'm willing to go as soon as possible and stay to the end, but wonder if that might not be a waste of money. I'd rather a comfortable journey and pleasant working conditions. Lacking broadband access in Spain means doing everything by phone-modem, which is very limiting. If I were to handle part of the webcast duties, it would be a whole lot easier from here. To be more specific, I suggest offering invitations to the top 12 programs on the official Swedish list, appended below. Note that they are NOT in order (you need to look at the ratings), this is from the ongoing Odyssey tournament website http://www.rebel.nl/odyssey.htm. Many will decline, the expense plus $2500 EF being too much for them. Not our problem. As they decline, you could go lower in the list. Enrique tells me 4 games per day is feasible. Here is a reasonable format: 2 groups of 6 teams, each RR section

with 5 opponents x 4 games = 20 games = 5 days. Top two from each section go to finals: 4 players: 3 opponents x 10 games = 30 games = 8 days. Total days = 13 (one day break before finals?), final day has just two games so media deadlines can be met. Total overall 50 games for finalists. Tie break could be overall score prelims+final. I can modify the plan for any non-prime number of entrants, I think. I don't like Swiss, would prefer to see the kind of RR which is envisaged for the cycle. Also, this way the pairings are known, and critical games in the prelims can be hyped. Anyway, that's a preliminary thought. All I need to know right now is what is BGN already committed to. Sounds like there is enough flexibility to allow some modifications to aid the promotion of the event. 24-hour website coverage requires a lot of commentators. Since the games are on the web, you don't need a 24-hour team in Spain. I think 1 commentator per session (4 per day) can handle everything. But I wonder if the resources are there to provide satisfactory web coverage? Is the budget enough to handle it? Is the promotion of BGN website a major goal? If so it needs a lot of attention in the coming month. If the answers to any of these is "no", then serious consideration should be given to handing off the actual game coverage to ICC, where commentators can be provided from anywhere in the world. ICC channels can be used for different languages (Spanish would seem to be a must!). This solves a whole lot of logistical headaches. With under a month to go, and no website facilities up and tested, as far as I know, the public face of the qualifier needs serious attention. From a PR angle, I can see spinning Cadaques as a fountain of creativity, with the world watching to see if the computers can be as creative as Dali. Maybe a link to the Dali Museum would be a nice touch, people could view great art while waiting for moves. Remember that chess is a sport. So I urge you to get from each competitor (or threaten to create ourselves) a sporty logo which can be on screen with the program. Makes it easy to visually identify the competitors, provides relaxation from the text. Haven't forgotten that it is also a business. Please take note that the ad industry has just approved new (larger) standard sizes for banners. You should be soliciting ads from computer companies, computer book and software publishers, and the usual chess crowd. Send me the rates (pretending that I really believe you have a rate sheet ready) and I'll pass them on to Cardoza. Summing up (before the rating list): There is a whole lot to do, and as usual not much time to do it in. I suggest simplification wherever possible, but creating an event which will have credibility. The losers may grumble anyway, but not getting invited is going to create an angry army who will not help our cause when it comes to publicizing the final. PARTICIPANTS 26 PLAYERS

1. SHREDDER 5 2570 —— D

2. GAMBIT-TIGER 1.0 2635 —— GUA

3. REBEL-TIGER 13 2598 —— GUA

4. HIARCS 7.01 2541 —— GB

5. SHREDDER4 CHESSBITS 2641 —— D

6. CHESSMASTER 8000 2520 —— NL

7. VIRTUAL-CHESS2 2530 —— F

8. FRITZ 6B 2629 —— NL

9. JUNIOR 6A 2589 —— IS

10. REBEL-CENTURY3 2582 —— NL

11. GANDALF 432G 2560 —— DK

12. ZCHESS 2.2 2400 —— F

13. ZARKOV 4.5R 2521 —— USA

14. CHESS SYSTEM TAL 2.03 2408 —— GB

15. WCHESS 2000 2509 —— USA

16. LITTLE-GOLIATH 2000V3 2480 —— D

17. PATZER 311B 2550 —— D

18. COMET B27 2510 —— D

19. CRAFTY 18.1 2500 —— USA

20. NIMZO 8 2555 —— A

21. GENIUS 6.5 CZUB-STYLE 2561 —— GB

22. SOCRATES X 2420 —— USA

23. MCHESS 8 2450 —— USA

24. YACE 0.23 2500 —— D

25. EUGEN 7.92 2470 —— E

26. GROMIT 3.7.4 2480 —— D

Sent by Raymond Keene

all good points-what I want is solutions not problems so can you discuss all this with Byron and Enrique and come up with answers-we do have a budget and the more progs that enter at 2500 bucks a shot the larger the budget! thanks ray

27 March 2001

Sent by Eric Schiller

Attached is a word document with suggestions, ranking lists etc. Let me know which way you want to go on the issues discussed.

Here is the Attachment:

Computer qualifier for Man vs. Machine Championship

Eric Schiller, Tuesday, March 27, 2001

This document contains a few comments and suggestions on the upcoming qualifier in Cadaques. Since time is short, I've tried to prioritize them, but since it isn't clear to me exactly what the parameters are I'm assuming the following:

1. The event must take place during the last two weeks of April.

2. No formal criteria for invitation exist yet.

3. The event is designed to choose a challenger for Kramnik in October

4. The goal of the qualifier is the selection of a credible opponent

5. The secondary goal of the qualifier is to promote the main event

6. Another goal of the qualifier is the promotion of the BGN website

7. The entry fee is $2500. The prize is participation in the main event

To establish a credible challenger, we must have a representative number of the top programs. Given that the human cycle qualifier, as presently contemplated, has 24 participants, I think that 12 programs would be ideal. However, the entry fee probably will limit the number to something less. I've already suggested a format for 12 programs, and will do so for 6 below.

The essential question which must be addressed immediately is whether a single or dual processor platform should be used. If there are very few dual processor programs, the additional processor would give them a huge advantage. However, it will also produce the strongest chess. I prefer, without strong feeling, a single-processor competition open to more machines. In the final, it could compete on the strongest compatible hardware available. Either system will work. Enrique is probably best qualified to make the decision, though I think we mustn't over-value the importance of the computer chess crowd, who will not be satisfied by anything we could possible do re the qualifier at this late date.

Once that decision is made, developers of the top programs should be invited. They'll only have a few days to make a decision, and come up with the entry fee, but that's better than shutting them out.

For credibility purposes, I think the SSDF computer rating list is the best guide for invitations. Invite everyone, but accept only those applications (at first) from the top programs, "waitlist" others to fill the spots that are declined. See section below on ranking and locating programs.

As a timetable, if invitations can be sent out by March 31, they could have a week to accept or decline, after which we simply implement the appropriate format. I can work out all variations in advance. There will be bitching about short notice, but the alternative was no invite at all, and there will be more time in the next cycle. The world will view this as fair enough under the circumstances.

Obviously to send out invitations the dates must be fixed. The event should take no more than two weeks under any circumstances, so go for dates that wrap around Dos Hermanas, so that the Spanish press, at least, can deal with both. I think that given some Easter weekend travel problems, that the first games should take place on the 18^th (the day before D.H. starts) and end on April 30 or May 1. That's 13-14 playing days.

Ranking and locating programs

Contact information: http://www.rebel.nl/authors.htm gives you 26 of the top programs. There are a few others to consider, but it is a great start.

Here are the top 50 programs from the March SSDF list. Notice that most of them are versions of the same few basic ones. The top 12 non-repeating are:

Fritz, Junior, Chess Tiger, Nimzo, Gandalf, Hiarcs, SOS, Rebel, Goliath, crafty, Chessmaster, Mchess, Shredder.

I think that is the primary list you should consider. But I would also invite Chess Genius, Chess System Tal, Comet, Gromit, Patzer and the rest of the Odyssey field. You can do so without guaranteeing a spot. The rating list is for games played on machines that are hardly state of the art, so those lower down or missing may not be that much weaker than the top programs. And dual-processor programs are not included.

The applicants will want to know the rules, so these should be determined well before this week. I'll defer to Enrique on all rules issues. But they should be discussed. The main issue for the moment is the dual-processor question.

  Rating + - Games Won Average opposition

1 Deep Fritz 128MB K6-2 450 MHz 2650 34 -32 470 66% 2537

2 Fritz 6.0 128MB K6-2 450 MHz 2626 24 -24 897 66% 2512

3 Junior 6.0 128MB K6-2 450 MHz 2594 22 -21 1109 64% 2490

4 Chess Tiger 12.0 DOS 128MB K6-2 450 MHz 2578 27 -27 691 62% 2492

5 Fritz 5.32 128MB K6-2 450 MHz 2547 26 -26 741 59% 2485

5 Nimzo 7.32 128MB K6-2 450 MHz 2547 24 -24 857 59% 2485

7 Nimzo 8.0 128MB K6-2 450 MHz 2539 30 -30 546 58% 2486

8 Gandalf 4.32f 128MB K6-2 450 MHz 2529 29 -29 584 52% 2518

9 Junior 5.0 128MB K6-2 450 MHz 2528 26 -25 750 57% 2476

10 Hiarcs 7.01 128MB K6-2 450 MHz 2526 37 -37 361 48% 2539

11 Hiarcs 7.32 128MB K6-2 450 MHz 2525 27 -27 679 56% 2481

12 SOS 128MB K6-2 450 MHz 2524 23 -23 925 53% 2501

13 Rebel Century 3.0 128MB K6-2 450 MHz 2514 31 -31 504 50% 2516

14 Goliath Light 128MB K6-2 450 MHz 2496 30 -30 546 46% 2527

15 Crafty 17.07/CB 128MB K6-2 450 MHz 2487 24 -24 857 47% 2505

16 Nimzo 99 128MB K6-2 450 MHz 2486 26 -26 731 51% 2481

17 Fritz 5.32 64MB P200 MMX 2477 19 -19 1338 56% 2437

18 Chessmaster 6000 64MB P200 MMX 2473 61 -53 184 76% 2277

18 Hiarcs 7.32 64MB P200 MMX 2473 24 -24 844 59% 2407

20 MChess Pro 8.0 128MB K6-2 450 MHz 2471 32 -32 474 45% 2507

21 Fritz 5.0 PB29% 67MB P200 MMX 2459 23 -22 1005 66% 2342

22 Hiarcs 7.0 64MB P200 MMX 2458 21 -21 1112 55% 2419

23 Nimzo 99 64MB P200 MMX 2446 23 -23 885 51% 2438

24 Junior 5.0 64MB P200 MMX 2433 20 -20 1185 49% 2440

25 Nimzo 98 58MB P200 MMX 2425 21 -21 1126 56% 2379

26 Rebel 9.0 47MB P200 MMX 2420 24 -23 900 61% 2341

27 Hiarcs 6.0 49MB P200 MMX 2417 24 -24 829 56% 2372

28 Rebel 8.0 51MB P200 MMX 2406 23 -23 887 50% 2406

29 MChess Pro 6.0 41MB P200 MMX 2405 24 -24 831 52% 2392

30 Shredder 2.0 58MB P200 MMX 2400 20 -20 1190 46% 2426

31 MChess Pro 7.1 46MB P200 MMX 2393 22 -22 1042 53% 2370

32 Genius 5.0 DOS 46MB P200 MMX 2390 21 -21 1137 51% 2383

33 MChess Pro 8.0 64MB P200 MMX 2389 27 -27 681 53% 2366

34 Chess Tiger 11.8 Pentium 90 MHz 2382 43 -43 261 50% 2383

35 Gandalf 3.0 64MB P200 MMX 2364 41 -40 307 59% 2296

36 Kallisto II 64MB P200 MMX 2342 35 -35 403 52% 2327

37 Rebel 9.0 Pentium 90 MHz 2334 23 -23 890 47% 2355

38 Hiarcs 6.0 Pentium 90 MHz 2332 18 -18 1437 51% 2328

39 Genius 5.0 DOS Pentium 90 MHz 2328 18 -18 1558 47% 2347

40 Nimzo 3.5 Pentium 90 MHz 2292 22 -22 998 46% 2322

41 Junior 4.0 Pentium 90 MHz 2287 22 -22 1035 42% 2341

42 Shredder 1.0 Pentium 90 MHz 2282 59 -58 145 53% 2262

43 R30 v. 2.5 2274 41 -38 343 69% 2135

44 Meph Genius 68 030 33 MHz 2198 45 -44 248 55% 2161

45 Berlin Pro 68 020 24 MHz 2125 24 -24 850 58% 2071

45 Meph RISC 2 1 MB 2125 62 -66 125 39% 2205

47 Mephisto Montreux ARM 14 MHz 512K 2099 29 -28 689 73% 1930

48 Atlanta SH7000 20 MHz 2090 29 -28 647 69% 1949

49 Sapphire II 2012 35 -33 444 63% 1916

50 Milano Pro SH7000 20 MHz 1974 33 -32 469 61% 1895

6 Team qualifier

Each team plays 6 games against each other team. After those 30 games (8 days), the top two machines play a 16 game match (4 days).

{END OF ATTACHMENT}

28 March 2001

Sent by Byron Jacobs

Eric, cc Ray Thanks for the various emails either sent to me recently or forwarded on to me by Ray. I have read through all this material now and my thoughts are as follows: I think there is one overriding factor which we need to consider before anything else: do we invite machines which are capable of running on a multi-processor platform, or do we open the event up to those which operate only on single processors? My view is that we should restrict the event to those machines which run on multi- processors, as one of these will certainly give Kramnik the best match. It is possible we could open the tournament up and have a single processor event and have the winner running, `on the best available hardware'. However, if a program is currently written only to run on single processors, surely it would be at the very least a major (and presumably impossible) task, for the software to be rewritten to run on a multi-processor platform in a very short time period. Remember that we have to get Kramnik a load of games played by the machine against grandmasters, and so the final version of the program has to be ready within maybe one month of the conclusion of the qualifier. I am no expert but it sounds like a tall order to me for a piece of software to be rewritten to run on a different platform in such a short time span. If we have a number of machines playing, it is quite possible that, due to the slightly random nature of the event, one of the weaker ones will win. This program will then be obliterated by Kramnik which will make the whole event a shambles. I do not think the sponsors will be thrilled to put up 2 million dollars to see Kramnik win 7-1 and press interest to wane after about two days. I do not see that there is any great need to invite machines that will obviously not be able to perform against Kramnik. My understanding is that there are only three programs that fit the bill: Shredder, Fritz and Junior. Shredder has already declined and so we should have a match between Fritz and Junior. Okay, all the other programmers will whine and say that we are cozying up with Friedel, but they can rewrite their programs for our next qualifier. Also, Kramnik was pretty much handpicked (as clearly the strongest challenger) to play Kasparov last year and very few people had much of a problem with that. We need to have a credible challenger and if there are only two then so be it - let them play a match. If you want to hold an eliminator to find a car to compete in a formula one race you would invite the best cars that run on a formula one track. You would not have an eliminator which features cars which specialize in going uphill through mud. There are other issues we need to discuss around the event, such as website coverage, but these are, I think minor, compared with the need to decide on the invitees. In summary, I think we should decide now to invite only multi-processor capable programs. If there are others that can compete with Fritz and Junior then all well and good - let's invite them. If not, we will have a match between the two.

Sent by Eric Schiller

What I don't understand is the assumption that there are only 3 multi-processor programs. Diep and Crafty both have versions, and I don't know about Crafty, but Diep was not invited (probably would decline, because of fee). I don't know which others have dual- processor systems, and don't know that they will be asked. I'd be justifiably furious if I were a developer of such a program. As I said, I don't care about the processor issue much Dual-processor is fine. But if it is just a match between two properties of the same company, closed to everyone else, then frankly I'll pass on this, don't want to be involved. Eric

Sent by Raymond Keene

nobody has been invited yet-we are still refining the invitational process. the point is to get it right-which program does Eric actually think wd win if all were invited-what is the company that markets both Fritz and deep Junior? Ray

Sent by Eric Schiller

Ray, it is Chessbase, as I've mentioned a few times. You are talking (at the moment) about handing the invitation to the final to a commercial Chessbase property with no one else involved. That I can't be part of. The smell would be a permanent part of my reputation. Going with dual-processor is fine, but there must be some open invitation to anyone with a dual-processor program or (I think) anyone else who wants to pony up $2500. Even though notice is

short, and most will decline, you've got to get someone else in the mix. Even if the event has Junior and Fritz demolishing the single-processor opposition, that is better than a match between two ChessBase properties. My own antipathy toward these two programs is well known, but then I only hate the programs I use daily. I have no problem with one of them winning, but nothing will erase the stink if Friedel is the only possible benefactor. Now that the processor decision is made, the next priority is clearly invitations. I suggest that the event be open to the top engines willing to pay the fee. Get at least a field of 6. If Diep plays, then I can't arbit those games because a version of that engine is going into my new program. But I doubt he'd pay the fees. Also, at this short notice, not many can take two weeks to go to Spain, or really get ready for the event. I don't see any chance of a deluge. Eric

Sent by Raymond Keene

Eric has some good points here

ray

Sent by Eric Schiller

As things aren't sorted out yet, I told Dallas just to ship me back here on Sunday night, the 15th. They couldn't wait any longer, have 14-day advance purchase to worry about. If it is necessary to fly me from Dallas, I'll just toss the unused return flight DFW-SFO. Keep in mind that the advance purchase of any SFO-BCN fare will expire in a few days. But then again, if it is just the two-machine match there is no reason for me to go for such a long time. Probably could do everything from here, or maybe a quick trip to London and Cadaques. Of course formal rules are needed for this event, and Enrique doesn't have any. Determinations have to be made concerning use of opening books and tablebases. Also some security so that arbiters can check to see that there is no interference. The easiest way, I think, is to have a copy of the program running on an identical machine in the custody of the arbiter. Eric

Sent by Eric Schiller

The following links provide some useful background with discussion of multi-processor chess or other useful stuff. http://www.inficad.com/~ecollins/winb-mod.htm http://www.xs4all.nl/~verhelst/chess/publications.html http://www.lkessler.com/cclinks.shtml http://www.dcs.qmw.ac.uk/~icca/journal.htm http://home3.swipnet.se/~w-36794/ssdf/ Eric

Sent by Vincent Diepeveen

Hello Raymond, I understand there is a computer-machine event. My program DIEP is a parallel program, running on n processors, and a very strong program. Tactical at the moment probably strongest in the world with loads of chess knowledge, which is the basic difference between Fritz and my program. Hard work has been done at it. I plan to get world champion (in computer chess) real soon with it. This won't be a major problem as this is a new generation of program: the same advantages as Fritz/deep blue had combined with more knowledge. The reason is easy. I am 2284 FIDE rated myself and work fulltime at DIEP nowadays. At AEGON 1997 my program was confused a lot with Deep Blue. Though names look very similar, it is not deep blue. it is better, but commercially seen that is very cool as the average watcher will think a player plays against deep blue :) I would like to know how to join your man-machine event! Friendly greetings, Vincent

29 March 2001

Sent by Eric Schiller

Diep is the programmer of the engine used in my forthcoming program (Cardoza Chess). However, we use a modified version of the program, not full strength. Nevertheless, should Diep play, Enrique should probably handle any disputes to avoid the appearance of conflict of interest. It is a *very* strong program. Hope the tube strike isn't too inconvenient today. If we go with the London press thing on the 18th, then fly me on Virgin or some other non-US carrier, as we are heading into a dicey period with one strike already and lots of labor troubles. Eric

Sent by Eric Schiller

If the following are now decided:

Event begins 4/20

Event will take place in Cadaques

Dual-Processor Pentium platform

Entry fee $2500

Prize: guarantee of loser's share in big match

Then I think the most efficient way of soliciting entries is: a) a post in rec.games.chess.computer with details and entry form (or pointer to web page) b) announcement in TWIC IN TIME FOR MONDAY's ISSUE!!! c) direct contact with Odyssey tournament participants (perhaps they can provide mailing list?) with deadline of April 6. Entries can be reviewed on the weekend, we can adjust format as necessary, and should be ready to go. Enrique can no doubt provide details of what the programmers must supply. I suggest that as with Odyssey tournament we provide a web page describing the participants and giving contact information. After all, with an EF of $2500 the losers should at least get some publicity out of it. We also need this stuff for the press. Next up: determining internet coverage methodology and promotion. I have a few creative ideas for PR, but am concentrating fully on getting the thing set up first. Byron, do you want to be in the loop on rules or should I just develop these with Enrique?

Sent by Enrique Irazoqui

Posting in a computer-chess forum is the kiss of death, the quickest way to attract ferocious criticism no matter what we end up doing.

RGCC in particular developed in such a nasty way that we had to create CCC in order to escape from there. Then CCC became more of the same.

I am sure Frederic Friedel will confirm. By the way, I took the liberty to include him in the CC line.

b) announcement in TWIC IN TIME FOR MONDAY's ISSUE!!!

c) direct contact with Odyssey tournament participants (perhaps they can provide mailing list?)

The Odyssey tournament is conducted by a notorious psychopath, perfectly discredited in the field of computer chess. Ask the SSDF about him and about his tourney.

We must decide on our own who will be invited to participate in the qualifiers. We know which programs have a chance, which programs are worthy entries.

Enrique

Sent by Eric Schiller

Agreed, but Friedel, who has commercial interests, must not be involved in any decision making and I strongly suggest that he not be copied on our internal discussions. A match between ChessBase products exclusively is unacceptable. Period. All programmers bitch about each other. They are all equally wrong, and I doubt any (except a few) are worse than the others. There is too much money at stake (over half a million dollars to the winner of the qualifier, as

that is losers share of final) to allow input from a commercial developer. Eric

Sent by Eric Schiller

So far, just comments from random bozos who got address somehow (except Diepeveen, I supplied him with the contact and he wrote a simple (for him) request for information since he has a dual-processor program. The way to reply to these is with a form letter. Byron can do that in Microsoft Outlook. But the form letter must point to an up-to-date website. The alternative is a mailing list, but not practical for us given time constraints. What I discern from today's messages is that we are going with dual-processor, open only to programs which run on dual-processor. I'm not sure I share Byron's concern that a single-processor program could "accidentally" qualify. Surely we have enough games planned (at least 50) to prevent that from happening. But I'm not an expert, and this issue is not critical to me. As long as the event is not a Chessbase Ltd. Production I'm OK. I get the impression from Enrique that Friedel is involved, and he has sent some copies of internal correspondence to him. I objected strongly, cc. to Byron. We can't have a high-end chess commercial company making decisions. Sure, I'm writing a program now, but it is deliberately dumbed down for the average Joe and the engine I use could not possible win (it is a weakened version of Diepeveen's engine). I would prefer to think that the decision making here is by you, with Byron, Enrique and me feeding in ideas. Eric

Sent by Raymond Keene

I have had an extensive dialogue with Ossie Weiner of Shredder-shay Bushinsky of deep Junior and Friedel himself. I have spent less physical time talking to Friedel than the other two but welcomed input from all three. none of them is involved in decision making but of course I listened to all their views. I got nowhere with Weiner but I am hoping that both Friedel's and Kasparov websites will push he qualifier heavily-by the way it will be known as the BGN world computer championship ray

Sent by Eric Schiller

Having websites helps with promotion at this late date, but word gets round the internet quickly, and as long as Crowther covers it, the news will percolate through the system. When considering the promotional value of the website, the formula is: Number of viewers minus Viewers who would know about it from some other source It isn't important that most people don't know about the event yet, in Internet time we are in no rush. The news can get out quickly enough, and what is more important is the quality of presentation BGN can achieve. While we don't want to see any weak programs, having one non-commercial combatant (say, crafty) would allow a Cinderella-effect, with lots of people rooting. Would help viewing numbers a bit. Each program has a fan-base, and the more programs, the more viewers. Eric

Sent by Raymond Keene

I am assured by Friedel that ChessBase only distribute other programs- apart from Shredder- these other programs are not ChessBase programs per se! it is like saying we cannot have just certain programs because they are all distributed by the USCF. the programs and manufacturers in each case are quite distinct and in any case Friedel thinks deep Junior will not renew its chess base distribution contract. there is also a fallacy about the prize fund floating around- the truth is as follows: if the machine wins the final v Vlad it gets $400k-if it draws it gets $200k and if it loses it gets nothing and Vlad walks off with t he 1 million dollars intact-no tax in Bahrain! indeed we may rewrite the contract so that funds won by the program go to a chess foundation for young players i.e. charitable purposes. the real prize for the computer program team is prestige publicity and a wonderful all expenses paid vacation in Bahrain. ray

Sent by Eric Schiller

I don't see any problem with the machine keeping the cash, especially if it is the lower amount. This means the sponsor will be rooting for the machine :-) I have no problem with CB distributing the programs, just not ALL the programs. It is more than just distribution, it is, I'm sure, the main distribution. All copies of Junior I have seen in USA are packaged by CB, just like Fritz. Similar packages. Same slogans. The relationship isn't just that of a seller of various goods. Eric -

Sent by Enrique Irazoqui

>>A match between ChessBase products exclusively is unacceptable. Period.

I also believe that an all Chessbase festival is not the ideal scenario. On the other hand it will be difficult to avoid, given the position of quasi-monopoly that they hold in the computer-chess business. There are 2 sides in this, the strict consideration about which programs must participate and the public perception of fairness beyond reproach in everything we are doing. I think it would be greatly valuable on both accounts, and mostly on the second, to get the support of the highly respected SSDF. The best known SSDF member in the computer-chess forums is Bertil Eklund, a man with a justified reputation of competence and personal integrity. If you all agree, I could explain to him what we are planning and include him in our email group as consultant. Else, Raymond or Byron could write to him. … The SSDF as such is never involved in chess events other than their own rating list, but having Mr. Eklund as consultant would be a de facto endorsement. He would also take care of public arguments on CC forums. Of course, we shouldn't mention either point to him. Regards, Enrique

Sent by Eric Schiller

Seems a good idea to me. I think we just need a few competitors. If Fritz and Junior compete in a small field, no eyebrows are raised. Even if it is just 4 teams. Eric

Sent by Enrique Irazoqui

Even if it is just 4 teams. But they are all distributed by Chessbase! That's the damn problem. Enrique

Sent by Enrique Irazoqui

Dear Mr. Eklund,

Brain Games Network who, as you are no doubt aware, organized the Kasparov v Kramnik World Championship match in London last year, are now planning a Man v Machine event to take place in Bahrain later this year. In this match Vladimir Kramnik will play against the strongest chess-playing software program available. This match will be over eight games and will be played at classical time limits. In order to select the program to play the match, we plan to hold an eliminator tournament in which the best programs available will play an all-play-all event. The games in this eliminator must be played at something approaching classical time limits, typically lasting around 4-5 hours, in order that the successful program is the best at slow play. We have yet to decide which programs will participate. We are well aware of your reputation as a man of personal integrity and of great knowledge in the field of computer chess. As such, we are wondering if you would be interested in being involved in this tournament as a consultant. Please let us know if this possibility interests you and, if so, we can open discussions about how to proceed. Kind regards, Enrique Irazoqui

Sent by Eric Schiller

There are several programs (Crafty and Diep come to mind) that are not distributed by CB. What about Gambit-Tiger, Rebel, MChess etc. Do they have dual=processor versions. I know Diep does, believe Crafty does. Socrates? Surely there are at least two programs that are not from the CB stable! It doesn't matter whether CB sells the program, the question is do they distribute it and package it in their usual CB/Fritz packaging. Eric

30 March,2001

Sent by Enrique Irazoqui

Hi Eric,

>> There are several programs (Crafty and Diep come to mind) that are not distributed by CB.

Crafty is distributed for free as a native CB engine in every CB package. Diep is much too weak. It never managed to do better than middle of the road amateur.

What about Gambit-Tiger

Good program, no SMP, also distributed by CB from April on. , Rebel, No SMP, not that strong anymore. MChess Marty released the last version 3 years ago. He retired. Besides, Mchess is weak by today's standards and not SMP ready. Same for Genius. etc.

Do they have dual=processor versions. I know Diep does, believe Crafty does. Socrates?

Socrates has been dead for ages.

Surely there are at least two programs that are not from the CB stable!

:) Which ones that are minimally strong and SMP ready? You forgot SOS. It is SMP and strong for an amateur. Still, over 100 points weaker than Fritz, Shredder and Junior.

It doesn't matter whether CB sells the program, the question is do they distribute it and package it in their usual CB/Fritz packaging.

They do. The only exception is Diep, but Diep is much too weak. Oh yes, Chessmaster too, but once again it is not SMP and it can't even autoplay. I hope I can convince Stefan to enter Shredder. Wish me luck, because that's our best chance.

Eric, I don't like a CB-only event any better than you do, but if worse comes to worst a DF- DJ match is better than a nobody's tournament. Imagine a Linares with Kasparov, Kramnik, John Smith and Pedro Perez. It seems that the new Deep Junior is not CB yet and maybe it will never be. That would also be a possible way out. Enrique

Sent by Eric Schiller

It doesn't matter if one or two weaker programs play, as long as they have the entry fee and run on the dual-processor. In any qualifier someone finishes last. If the engines you mention are too weak, so what? It means either Fritz or Junior will win, but that's likely anyway. If we can make the claim that any dual-processor program had a chance to compete, it clears all the PR problems. I don't think the dual-processor Diep is weak at all, he's been working on it almost full- time. A weakened version of Diep is the engine I'm working with for my new mass-market program. (I have no commercial interest in Diep and the Cardoza product is packaged explicitly as a weakened version. I'm not interested in high-end stuff, we are making a program that is fun to play, not a serious analyst.) You seem concerned that one of the weakies might somehow win, and undercut the credibility of the final, but I think that is an extremely remote possibility and in any case would be no different than any other sport. I don't think the event should be "rigged" to exclude upsets. If a "weak" program can pull off an upset in 50+ games, it simply isn't that weak. Our goal is a credible challenger. As long as two strong programs like Fritz and Junior play, our event is credible, no matter who wins. I realize the computer chess crowd will moan no matter what we do, and frankly they are not even a blip on the radar map. Our concern is what a CNN journalist will think, not what a loony programmer says (and Enrique, you should see some of the loonies who have been emailing Ray!). The whole BGN setup was designed so that all championships are OPEN to all via qualification. Given time constraints and other criteria, we have to limit the first event to dual-processor chess programs. Outside of that, I think no criteria other than the entry fee need apply. Certainly we should not exclude weak programs, because that violates the very principle of the BGN enterprise. I think this is a very important point, but defer entirely to Ray and Byron, who are closer to BGN central. Assuming only 4-6 teams, there is time for enough games to insure no freak accidents take place. Fritz, Junior, Shredder, SOS, Diep, Crafty would be a credible field. The first three would be the probably competitors for top honors, but the last three would keep it interesting. If the result of the event is that the world thinks that CB distributes all the best engines in the

world, so be it. But it is a position that must be earned. Eric

Sent by Enrique Irazoqui

As agreed before, I include Frederic Friedel in the CC line. Eric, I strongly disagree with you about which programs should participate in the qualifier. To begin with, whether or not Chessbase "must earn" whatever position is irrelevant to us. As I said before, a Linares or a Candidates with Kasparov, Kramnik, John Smith and Jose Perez makes no sense. It would be the laughing stock of the chess world. With enough time ahead of us, maybe for the man-machine event of 2003, we may want to play pre- qualifiers (something like the Interzonals) and a Candidates. Right now we are running too short of time for that. We only have time for the Candidates. The only programs that are not John Smiths are the Deep versions of Fritz, Shredder and Junior. You say that Diep is not weak. Please, show tournament results. I can give you some that are rather pathetic. For instance, in the last Paderborn of a month ago where Diep on a Dual didn't manage to do better than other amateurs on single processor machines. Diep never ever did well in any serious tournament. Aside from Diep, a classic John Smith of your list, all the other programs you mention work as Chessbase native engines, with the exception of Shredder. Then: 1 - Nothing is solved by adding Crafty and SOS, CB engines and weak, between 130 and 170 points under Fritz. 2 - The field is so much weaker that loses credibility: example of Linares with John Smith. 3 - Adding random entries means fewer games per match, greater randomness in the end result. 4 - Shredder is the missing participant that matters. I can try to convince Stefan. It will be difficult but it may work. Byron? I strongly believe we must stick to our original plan of limiting the participation to Deep Fritz, Deep Shredder and Deep Junior as the only worthy SMP programs in existence. I am sure that Bertil Eklund, who I hope will join us as consultant, will agree wholeheartedly. We are getting short of time and we should start moving fast. Enrique

Sent by Eric Schiller

I still disagree. You mention 100-170 points weaker. That is not John Smith, that's Nigel Short, Boris Gelfand level. Now let us consider the plan for the human qualifier, which, since its inception, has planned for open qualifiers. Anything might happen, and given that the top players are invited without qualifying there is a high probability that qualifiers will be over 150 points lower than the top rated invitee (Kasparov, at say 2800). A 2600 player would not be considered a joke. In any sporting qualifier you have the chance for "Cinderella" results. The NCAA basketball tournament shows this every year. The public is used to a "weak" team reaching the final four. In fact, there is more public interest when a dark-horse does better than expected. The only justification for eliminating weaker (but over 2400) programs is logistical, but that doesn't apply to creating a 4 or 6 team field. Of course, if you scrap the idea that this is a qualifier, and treat it strictly as a World Championship, then you can hand-pick the opponent, just as BGN did. But that is not the concept here. If there is a radical change from qualifier to championship, then it is crucial that this not be identified in any way as a qualification, and BGN would have to extend assurances that a true open qualifier would be part of the next cycle. If the BGN human qualifier were in place, that would be an easier sell, but there is still great skepticism about the BGN cycle, understandably, since no public announcements have been made. Even in the case this is a hand-picked championship, I still feel that having one company owning exclusive right to participate will be viewed as totally invalidating the event in the eyes of the public. Since credibility is such a high concern, it is counter-productive. If the rules simply state that only dual-processor programs can compete, then every such program in the world could play and I doubt that even 10 entries could be found. It is better to let other programs compete and do badly if they are weak. Fritz, Junior and Shredder would of course benefit as distinguishing themselves as in a class without peer (such as Kasparov, Kramnik and Anand enjoy now). And if Chessbase distributes all the top programs, all credit to them! But I think that your comments that all other engines are too weak to participate is simply an opinion which BGN would have to go to great trouble and expense to defend empirically. Public opinion is going to go against any hand-picked event where participation was limited to two or three engines. Four participants seems to small to me, but possibly can be justified, whether or not Shredder is among them. Six would be ideal. I can't see any serious, justifiable criticism if BGN, in its very first event, extended an invitation to all dual-processor machines. Of course, I'm looking primarily, almost

exclusively, from the PR angle (which is what Ray expects me to do). Credibility is the primary goal. I don't care at all what the in-crowd of the computer chess world thinks of the quality of the weakest games. I care what the world thinks of the quality of the best games, but more importantly, of the fairness of the selection procedure. There is enough head-to-head evidence among Junior, Shredder and Fritz that we could probably just skip this event and select one on the basis of some objective criteria. I don't think the world would view this as any less fair than only inviting the three and seeing what happens in one event. If the event is not in any sense open to the world, it just isn't part of the BGN scheme of open qualifier —> match championship. If it is a championship, and not a qualifier, then use the same scheme as in the human championship and just select a challenger. A Anand-Kramnik match was never contemplated, due to time constraints. The exact same time constraints exist now. Eric

Sent by Eric Schiller

Did I mention ferret? It won many strong events. Definitely parallel. Eric

Sent by Eric Schiller

Here are the best engines on ICC, playing at standard (> 15 minutes) time controls.

1. evilcomp = Deep Fritz

2. Netsurfer = Deep Fritz

3. Crawly = Fritz

4. HIARCS

5. ShredderX

6. Val Vipp = Deep Fritz

7. Silicon = Tiger

8. ChessBeta = Tiger

9. Ferret

10. Interceptor = Junior

11. SpitFire = Tiger

12. Kooloo = Tiger

31 March, 2002

Sent by Enrique Irazoqui

This list is truly great. It shows once again which programs deserve to be consider as contenders for the match Kramnik-machine. You will see that Hiarcs and Tiger on one processor machine do better than Ferret on a Quad (4 processors) Xeon. SMP ready programs that do well here are, as expected, Deep Fritz, Shredder and Junior. We must draw the line somewhere, and sheer strength is our only possible criteria. The inclusion of relatively weak programs like Crafty, SOS, Ferret, Diep, will be perceived as absurd, for the simple reason that each one of them on a Quad machine is less strong than Tiger and Hiarcs on a single processor machine. Aside from this absurdity, the inclusion of such programs in a relatively short tournament increases randomness and uncertainty to such a degree that we risk falling in a scenario in which Kramnik will end up playing a meaningless program. This scenario would disrupt the whole event, and thus we must avoid it at all cost. Face it, Eric. There are only 3 worthy contenders, Deep Fritz, Deep Shredder and Deep Junior, as confirmed by the SSDF list, Cadaqués tournaments and even this ICC list you mention. It is as simple as that and there is no way around it. Best scenario: these 3 engines play each other over 20 game matches, all against all. Second best, if Shredder does not participate: a "final" between Deep Fritz and Deep Junior. Both of them won Cadaqués, topped the SSDF list, won World Championships. Either one is beyond reproach and free of risk. I would like to hear what Bertil has to say about this issue. Enrique

Sent by Enrique Irazoqui

I agree with all your points, Bertil. I will try to convince Stefan, but you know Ossi well enough and how difficult it will be for him to release his jaws and let Stefan go. Wish me luck... Enrique

Enrique Irazoqui skrev: This list is truly great. It shows once again which programs deserve to be consider as contenders for the match Kramnik-machine. You will see that Hiarcs and Tiger on one processor machine do better than Ferret on a Quad (4 processors) Xeon. SMP ready programs that do well here are, as expected, Deep Fritz, Shredder and Junior. We must draw the line somewhere, and sheer strength is our only possible criteria. The inclusion of relatively weak programs like Crafty, SOS, Ferret, Diep, will be perceived as absurd, for the simple reason that each one of them on a Quad machine is less strong than Tiger and Hiarcs on a single processor machine. Aside from this absurdity, the inclusion of such programs in a relatively short tournament increases randomness and uncertainty to such a degree that we risk falling in a scenario in which Kramnik will end up playing a meaningless program. This scenario would disrupt the whole event, and thus we must avoid it at all cost. Face it, Eric. There are only 3 worthy contenders, Deep Fritz, Deep Shredder and Deep Junior, as confirmed by the SSDF list, Cadaqués tournaments and even this ICC list you mention. It is as simple as that and there is no way around it. Best scenario: these 3 engines play each other over 20 game matches, all against all. Second best, if Shredder does not participate: a "final" between Deep Fritz and Deep Junior. Both of them won Cadaqués, topped the SSDF list, won World Championships. Either one is beyond reproach and free of risk. I would like to hear what Bertil has to say about this issue. Enrique

Bertil: It is in fact true that if this event had been played on today's fastest single-machines it had been a must to invite Tiger (Gambit). The new Tiger with tablebase-support could very well be the strongest of today's programs. In a short match it could also have a good chance to win against the dual-programs. Mr Theron, the author of Tiger doesn't seems to have any intensions to make Tiger a SMP-program, at least in the near future. Therefore it seems clear to me that the only realistic contenders are Deep-Fritz, Deep-Shredder and Deep- Junior. The only problem is to convince Millenium or probably a better way is to convince Stefan Meyer Kahlen to participate in the event. Shredder5 isn't officially included in the SSDF-list but it isn't far away from the top-program Deep-Fritz. I think you can try to convince Stefan with the argument he use himself, he seems to believe that Shredder is relatively better in long-time-control games and on faster hardware, another point is that Shredder slightly dislikes the AMDk6-2 450 processor used by SSDF. With "today's" hardware it could probably do a bit better. I know that Shredder5 (after several hundred games) is clearly weaker then the competition in blitz-games, but one of the strongest in tournament-control games. I am absolutely sure that Stefan wants to participate in this event and my suggestion is that someone with a good relationship to him try to convince him. I know that Enrique and Stefan often corresponds by email, so I guess he is the right person. Bertil

Sent by Enrique Irazoqui

Absolutely. It must be well understood and signed in the contract that if Shredder participates in the Qualifiers it will play Kramnik. First step, convince Stefan. Second, Stefan convinces Ossi. Third, both sign the contract. All are required conditions. Otherwise Shredder is obviously out. Enrique

Sent by Eric Schiller

I think this is a perspective from the small world of computer chess. No one in the world expects a competition to be without weak participants. From the Olympics to every major sport. I find the emphasis on strength strange. It is only the strength of the winner that counts. I find every person in the computer chess world has a pile of strong opinions which can be reduced to : I like this engine, all others are crap. There is no substance at all to this argument. No reason to exclude weak competitors from an open event. Exclusion just creates more (justified) criticism. The difference between the strongest and weakest engine is far less than that between Kasparov and many of the players he faces in qualifying competition. In 1984, Smyslov was much weaker, comparatively, than any of the programs you list. In any case, there is no empirical evidence, since the programs are constantly changing and evolving. The event would have more credibility with Hiarcs and Tiger in it. Seems to me someone is afraid that one of the lesser programs might just expose weaknesses in the commercial giants. This would be a great

result in terms of PR. Everyone loves an underdog. In light of the below I feel more strongly that single-processor programs should be allowed to compete, but can live with the decision for dual. (Of course, why not quad? Wouldn't the experimental programs running on quad beat out the big commercials who don't). Not even a fraction of a percent of the people who watch the games are capable of understanding the subtle differences that distinguish 2800 from 2700 or 2600. The only people who care are computer chess insiders, who are of no relevance to BGN's business plans. The event should remain open or a single program should be hand-picked. No other path is worth contemplating. Eric

Sent by Eric Schiller

Good point Ray. Look, you will have a lot of grief if BGN just arbitrarily chooses a few programs (more trouble than if you choose the challenger directly!). This is why I am so adamant that either you have to let everyone in (why the hell not, there is no decent argument against a free for all?) or create exacting criteria. Among the criteria, the company with distribution rights must be invited. I have heard no decent or credible argument why all dual-processor programs shouldn't be invited. And if a top single-processor program like Tiger wants to play, why not? Enrique's attitude is that of a typical computer chess insider. No one on earth can rationally discuss comparative strength of programs. The only test is to let them battle it out. This attitude is like saying that Judit Polgar shouldn't play a top event because she is a girl and isn't good enough. I doubt you could take more than 30% against any of the programs Enrique is dismissing as "weak". I know I can't. Suzanne was an idiot but on one issue she was right: the need to keep the cycle open so that some random bozo can get to the final stages before being cremated. If a computer enters and scores zero points out of 50, no one but computer geeks would raise an eyebrow. The perspective you are getting just is not valid in the PR or business world. It is our job to make whatever event we have credible. Basic PR. If some programs bomb out, we smile and say that is the price of open qualifiers. Next time we will have the opportunity for a more rigorous qualifying procedure. Meanwhile, either open the door or just designate Fritz or Junior for Bahrain and be done with it. Looks like this one has to be your call. I don't see any chance Enrique and I will agree. We are in different worlds. Set your own goals and pick the solution that most advances them. Eric

Sent by Eric Schiller

Thanks for the input! I think I've made most of my arguments in exchanges before you came in, but just so you know where I stand: I believe that AT LEAST all dual processor programs should be invited. It is of no concern to me how strong the programs are, as long as some top programs are there to legitimize the result. My views are based on the goals of BGN in public relations and promotional terms. The human cycle will be open to all. For me, the worst possible scenario is one in which ChessBase GmbH has exclusive distribution of ALL the competitors, while programs with bona fide credentials like Tiger and non-commercial efforts like Crafty are excluded. I believe in such a case that BGN would be better off simply choosing the challenger, as in the case of Kasparov vs. Kramnik. To exclude programs because someone thinks they are too weak is just not a defensible image in the media. I have very little contact with the computer chess community, in spite of working on Kasparov's Gambit, Chessmaster 5000 and now Cardoza Chess. I know the personalities tend to be extreme, and have a policy of not putting much faith into various claims of comparative strength. I haven't looked at rec.games.chess.computer in a couple of years. The only results I trust are head-to-head on equivalent hardware. To me, the ideal event has between 4 and 12 programs participating. I doubt that we could find anywhere near a dozen (after all the entry fee is $2500), but would me most comfortable with 6 competitors. 4 would do, if it included at least 1 which is not exclusively distributed by ChessBase. Enrique obviously feels strongly the other way. It will be up to Ray Keene and BGN to make the final call. Eric Schiller International Arbiter (no cc on this, it is just to bring you up to date with where things stand in the discussion)

Sent by Eric Schiller

It isn't easy to see what you are agreeing to in the last few emails. Maybe worth a complete sentence from time to time. I'm still uneasy about Friedel being copied on internal correspondence. He's going to hate me more than ever now, though I think that either of my proposals benefits him more than the others (assuming that one of his "properties" wins). Eric

2 April, 2001

Sent by Eric Schiller

It's Monday, when I traditionally try to sort out what is going on in the world. Not very successfully. Can you help me by updating me on the following?

1. Are we set for an April 20 start, with some function in London? You might want to get my ticket purchased this week to save some money on the advance purchase?

2. Bahrain dates confirmed for Oct. 1-14? (Schools are asking me to work out a schedule for next year)

3. Any word from Argentina on the Candidates? What are they supposed to be supplying and do you need a fallback option?

4. Internet qualifier can be on ICC. They are handling the Dos Hermanas Qualifier quite well. I just finished playing one yesterday, finished 12th (ahead of many GMs and IMs), defeating players rated way up there (performance rating on ICC over 2750, I think).

<deletion of material regarding non-related business proposition>

And for you: 1. Latest "news" on your website is Jan. 29. I've mentioned this enough times to Byron that I don't want to push. But it really is sad. 2. Web pages still show human open qualifier starting in "Early 2001", Ron King match this month! "Cobweb" sites are usually seen as a sign that a business is failing. Any potential visiting your site would be alarmed. I think you really need to crack the whip here. Your website is all the world sees of BGN, and it is pathetic. If you search for the Kasparov-Kramnik match, more hits show up at Malcolm's site. I tried a generic World Championship search yesterday, wound up directly at Malcolm's Henderson page!! I deliberately say nothing about the computer qualifier. You have all the issues, all my positions, and I await the decisions from above. I'm happy to do anything you like, unless it is just Fritz vs. Junior, in which case you don't need me there anyway. Eric

3 April 2001

Sent by Byron

Dear Eric, We have now taken a final decision that we will invite four programs to play in the eliminator. These will be Fritz, Junior, Shredder and IBM. Obviously IBM will not play and probably Shredder will not play either. We still have some hopes to gain the participation of Shredder but most likely it will be just Fritz and Junior. All reputable sources indicate that these are the best programs and I do not believe the event will lack credibility as an eliminator. For the next cycle we will invite the top programs and also have a further event where the best `amateur' programs can compete for a place in the final eliminator. However, for this cycle we do not have sufficient time for this. Of course many people will complain, but there is not a lot we can do about this. I know you have reservations about a Junior - Fritz match. However, we would still like you to be involved in the event and also the Man v Machine match in Bahrain, specifically as one of the arbiters. However, if you feel that you do not want to be involved in a Junior v Fritz eliminator we will respect this decision. The event in Cadaqués will probably start on the 16 April and we anticipate it will run for two weeks. Please let me know if you want to be involved. Thanks, Byron

Sent by Eric Schiller

I cannot take part in an event limited to Fritz and Junior, so if the others decline I will have to pass on this. It is unfortunate, because I had already committed to the time period and cancelled all of my engagements here. So it is going to cost me a lot of money to give up the match in Spain, not just the lost fee but a lot of school teaching which cannot be restored. But I feel strongly that this event will come under justifiable criticism, and I carry around enough baggage as it is. I can't be part of such an event. If Shredder agrees I can consider the event so please keep me informed. I have no problem with the final, and am willing to work on that match subject to suitable arrangements. Eric Schiller

Sent by Eric Schiller

Since it seems I will not be participating in Cadaques (unless Shredder changes its mind and a fourth team is found, both unlikely) I need to know whether you still want me to do the rules for Bahrain, and if so who I should be working with. It can be done by email and some phone conferences. But we'll need to work out some other financial arrangements. I'm sorry it didn't work out. I made a lot of difficult adjustments to free up the time in late April, and have driven everyone here crazy. It is too late for the schools to change their schedules, so this is going to cost me serious money, just after Tax Time. Still, I'll cope. But lets get the Bahrain stuff sorted out contractually, since now it seems I must pass on the World Youth because of the date conflicts. Eric

Sent by Raymond Keene

once we know about Shredder for sure we will make a plan b ray

Sent by Eric Schiller

OK. Since I've already cancelled all my teaching, I'll just wait and see. Shredder-Junior- Fritz is far from perfect but is acceptable. Still, would be better with a fourth, even a weakie. But if Shredder says no please let me know ASAP so I can try to reschedule a few things and cut my losses. I have an organizational meeting in a few hours with one of my new schools. I'll tell them I am unavailable this month but will keep the door open for sessions last two weeks of April just in case. Eric

8 April 2001

Sent by Eric Schiller

OK. So it looks like I won't be going, though I guess that's still up in the air. Just keep me informed. I've let my students know that at this point it is unlikely that I will go to Spain, but haven't scheduled anything (though I just pushed one chess mom out the door as she tried to grab both weekends, told her I'd let her know by Thursday). On the other hand, I have paid my taxes and have kept the time period clear, in case there is a need for me there and there is another participant. Short notice is of course the rule on a Ray Keene production! On that assumption, should I be planning to do the Bahrain rules with Enrique by email/fax? I see no problems with that. But if I am not on the Cadaques budget, we should work something out to build the rules into a letter of agreement for Bahrain. Meanwhile, any movement in Buenos Aires re Candidates? If it is not firm, what are they offering so I can see if California or Hawaii has interest. I have good venues in both places.

20 April 2001

From Byron Jacobs

Dear Eric,

Ray forwarded me a recent email from you and asked me to respond.......

The Machine v Machine event, with Junior playing Fritz, will start in Cadaques on Tuesday next week. We will be doing the publicity for this (mainly on the web) today, Friday, and on Monday.

The dates for the Kramnik v Machine match are still as you indicated in your recent email. I presume we will need you there, but you should liaise with Ray on this one.

I do not know anything about the Candidates qualifier - again refer to Ray.

Best regards, Byron

PS: Sorry I have not been in touch, but you indicated you did not want to be involved in the Machine v Machine event.

Sent by Eric Schiller

Thanks. No, I don't want to be involved in the Fritz-Junior match (under any circumstances), but of course I am interested in following it. Ray had wanted me in Cadaques primarily, he said, to work with Enrique on match regulations for the October match. That of course can be done comfortably at a distance, However, he indicated he wanted the rules and regs done by the end of the Cadaques event, so presumably work should begin soon. Can you send me whatever rules exist and are implemented for the F-J match? Eric