FIDE WCC R6-2: Slow day in Tripoli

Links Pairings Tree World championship news History of World Chess Championships List of all players (with personal cards) Picture gallery Live coverage

Results of Semifinals

Report on semifinals game two

It was a short day at the office for the four semifinalists. After 17 moves Mickey Adams decided not to push any harder in a typically turgid line of the Bb5 anti-Sveshnikov Sicilian. After ten moves Black had pawns on c6, d6, e6, f6, and g6 and White had a knight on h4. These positions are so ugly that it's no wonder the players couldn't bear to look at it for longer than 17 moves. Adams will defend his one point lead with the black pieces tomorrow.

The only time Radjabov had a must-win with white he blew Dominguez off the board with white in the quarterfinals. It seems unlikely he'll be able to do the same to Adams, who is in excellent form. The young Azerbaijani can't wait for the final game to try and win with black, so we should see some action on Saturday.

Action is what we didn't see much of over on the other board with Veselin Topalov and Rustam Kasimdzhanov. In their first game they played 15 moves that had all been played before various times before agreeing to a draw. Today they made it all the way to move 25, only 17 of which had been played before!

This one followed the modern sensibility of offering a draw as soon as you're not better with white and accepting just about any draw offer with black. Only two pawns had been exchanged. Time to bring back the chicken factor?!

This diagram shows the final position after the repetition. (..Ne5-c6 and Rd1-d4) White has no good way to protect the c4 pawn after ..Ne5. If he plays Nd2 (instead of Rd4), Black finally achieves the dream break of his entire strategy, ..d5! Still, seeing draws with a board full of pieces like this is always a shame.

Garry Kasparov chimed in today after the round finished and wondered if Topalov had forgotten what it was like to face tough resistance after cruising so far so easily. Kasimdzhanov is likely to put him under some pressure tomorrow with white, so we'll see if the Bulgarian's form has been dented at all by three days in a row without a hard fight.

Although he has now been slowed by two draws against Kasimdzhanov, many have been wondering if Veselin Topalov was on some sort of record pace with his tremendous 9.5/10 score. Most of his opposition hasn't been of the highest level, so how does his performance stack up against other great runs?

We corresponded with chess statistician Jeff Sonas on the matter and discussed different versions of TPR, or tournament performance rating. We add a few other considerations specific to the knock-out format, which while not changing the statistics certainly influences how they might have come about.

From Jeff Sonas

As you may know, there is no single "official" way to calculate a tournament performance rating (TPR). I have tried various approaches over the years, using FIDE ratings, Chessmetrics ratings, etc. The most "official" way to do it, would be to calculate the theoretical FIDE rating a person would need, such that their actual %-score matched their expected %-score (based on all the FIDE ratings and using the Elo expectancy table). You can't just take the average rating of the opponents because the low rating of Abulhul would drag everything way down. With his FIDE rating of 2737, Topalov would be expected to score 99% against Abulhul's 2076 rating, but even if Abulhul's rating had been 2176 instead, Topalov should still score 98%. The difference between 2076 and 2176 is miniscule in terms of Topalov's expected score, but it would make the difference between an average opponent rating of 2509 vs. 2529.

Because you have to calculate the expected scores on an opponent-by-opponent basis (not just taking the average opponent rating), you have to work backwards, trying out various theoretical FIDE ratings until you find one that leads to the expected score. If Topalov's FIDE rating had been 2948, then a 9/10 score against those opponents is what you would expect. Thus I would say that his "official" tournament performance rating was 2948 at that point. After the draw against Kasimdzhanov, it's down to 2897..

Anatoly Karpov at Linares in 1994, on the other hand, scored 11/13 against a host of opponents with an average rating of 2680. Eight of those thirteen opponents had a rating above 2680; whereas none of Topalov's opponents have been rated above 2660 yet. Unless Topalov comes up with a better score than 11/13, there's really no comparison. By this methodology, Karpov's TPR at Linares 1994 would be 2977. If Topalov manages to defeat Kasimdzhanov in the next two games, for a total score of 11.5/13, then his official TPR at this tournament would be 2942. If he does that, and then scores 3.5/4 against Adams to win the championship, his overall score would be 15/17, for an overall TPR of 2977, matching Karpov exactly. If Topalov manages to win his next six games, including a 4-0 sweep of Adams in the final, then his total score of 15.5/17 would be a TPR of 3026.

Note that under this methodology, you can't calculate a TPR if somebody has a 100% score or a 0% score. Thus we can't evaluate Bobby Fischer's two straight 6-0 sweeps in the Candidates tournaments.

If your head hasn't exploded yet, we'll add that you shouldn't compare a KO TPR with those from other tournament systems, especially a round robin like Linares. When you lose the first game of a KO, as all of Topalov's opponents did until Kasimdzhanov, a loss is the same as a draw in the second game. You throw the kitchen sink, do anything you can, and usually end up losing.

It would be more in the spirit to calculate a range. What would Topalov's TPR be if all of his second game wins after a first game win had instead been drawn. That would be the low, with the high being his actual TPR. Of course it is a tremendous performance no matter how you add it up. Sonas points out that if you include just those game one wins his TPR is 2992!

• Replay and download all the semifinal games
• View all round five games online and download PGN
• View all round four games online and download PGN
• All the games from round three in PGN
• All the games from the first two rounds in zipped PGN

Previous reports

• BBC report: Gaddafi's strategic chess move
• FIDE report on round 3-1 with pictures and videos
• Interview with Zhang Zhong
• ChessBase pre-match report
• ChessBase round one game on report
• ChessBase round one report
• ChessBase round two game one report
• ChessBase round two report
• ChessBase round three game one report
• ChessBase round three game two report
• ChessBase round four game one report
• ChessBase round four game two report
• ChessBase round five game one report
• ChessBase round five game two report
• ChessBase round six game one report
• GM impressions by Sergei Tiviakov
• Postcards from Tripoli by Paul Hoffman
• Odds: Who will be the next FIDE world champion?
• Revised odds for FIDE championship
• Odds after round three

General information

The FIDE site, which is being hosted by Libya Telecom And Technology, contains the schedule, list of players, results tree, games, reports, pictures and videos. The start page is http://wcc2004.fide.com and http://wcclibya2004.com.

Live coverage

The live game transmission from Tripoli, apparently of all games, requires you to have Java Virtual Machine installed on your PC. This program is distributed free of charge by Sun Microsystems (and is useful for many other applications). Visit the Java check page to see if you have everything required for the live coverage and install Java if you don't. To follow the games click on "Live coverage" in the link list above. There is a "View" button behind each pairing of players. Many of the key games will also be covered and discussed on the Playchess.com server.

Schedule

Note that local time in Tripoli is the same as in Central Europe. The start of the games is generally at 14:30h, which is GMT + 2 and translates to 13:30 London, 8:30 a.m. New York, 16:30 Moscow, 18:00 New Delhi, 20:30 Hong Kong, 21:30 Tokyo, 22:30 Melbourne, and 03:00 a.m. (on the next day) in the French Polynesia-Marquesas Islands of Taiohae.