Fritz has been the world's leading program for more than ten years now. It is the analysis partner of virtually all the top players in the world, and it has successfully played against many of them over the years: Veselin Topalov (No. 1 in the world rankings), Viswanathan Anand (2), Peter Svidler (4), Alexander Morozevich (5), Peter Leko (8), Michael Adams (9), Alexej Shirov (14), Judit Polgar (16), Ruslan Ponomariov (20), Sergey Karjakin, Loek Van Wely, Anatoly Karpov, Viktor Korchnoi and many others. Fritz has played two mega-matches: once in 2002 against world champion Vladimir Kramnik in Bahrain, with a 4:4 result, and once in 2003 in New York against Garry Kasparov, a match that ended in a 2:2 draw.
With all these matches and tournaments against top human players the Fritz programmers have gained a tremendous amount of valuable information on the way humans play and understand chess. And they have been diligently implementing this knowledge into the program to make it play more rational chess, from a human point of view. Top players have for some time now realised that Fritz was the chess engine with the largest amount of chess knowledge and with the most human playing style.
This is not just good for games you play against the computer. It is also important when you are analysing with the computer, where the program will come up with a reasonable and plausible strategy even when there are no tactical motifs in sight. This does not only hold true for the position on the board, but also occurs in countless positions in the search, where proper strategic judgements need to be made, even when a position is devoid of concrete tactics.
In ten days (on November 25th) the latest version of this remarkable program will face world champion Vladimir Kramnik in a six-game duel. The venue is the renowned Art and Exhibition Hall in Bonn, Germany, where the match will share billings with the big Guggenheim exhibition that is being held in the same location. The schedule is as follows:
| Game 1: | Saturday | 25.11.2006 | 16:00 h |
| Game 2: | Monday | 27.11.2006 | 16:00 h |
| Game 3: | Wednesday | 29.11.2006 | 16:00 h |
| Game 4: | Friday | 01.12.2006 | 16:00 h |
| Game 5: | Sunday | 03.12.2006 | 16:00 h |
| Game 6: | Tuesday | 05.12.2006 | 16:00 h |
Vladimir Kramnik is currently preparing for the match with a team of chess and computer chess experts. The world champion has a copy of the program and realizes that Fritz 10 has made tremendous strides in playing strength, and that strategic tricks worked on chess programs in the past no longer are applicable to the new version. Kramnik now sees Fritz as the favourite in the match in Bonn.
So Fritz, one of the most knowledgeable programs around, has gained even more chess understanding in version 10. But apart from stronger play and a healthier, human playing style, how does the program make use of its new abilities? Well, there is a feature that taps directly into the program's ability to search for and execute plans, based on its chess knowledge.
Before we come to the revolutionary new function of "show analysis" let us take a look at an area where it is really useful – when you are watching top-level chess games on the Playchess server.
One of the pleasures of Fritz 10 is the sharper, crisper graphics, which allows us to follow multiple games on the server. Above is a case in practice: watching all five games of the Tal Memorial in Moscow. With the kibitzer engine running you can simply click on any of the boards to concentrate its attention to that position (the notation switches as well). Naturally all the boards are refreshed whenever new moves come in.
What, we hear you ask, is with the arrows and green squares in the diagram on the top left? That is the new function of Fritz 10 we have been leading up to. With the engine running you can see the plans that are being considered, graphically on the chessboard. The display of the critical moves and squares is generated in the search process of the engine.
Here's an example. The orange arrows show white plans the engine is currently examining, the blue arrows are the most important black plans. Green squares are the critical ones in each of the plans. The above position is from Leko-Morozevich, Tal Memorial (6).
The ability of Fritz 10 to graphically display plans is interesting because many of those plans do not appear in the main line displayed by the engine. This is mainly because many important motifs, e.g. mating threats, have been refuted in the main line. But they are critical elements of the position. The main line is just a very narrow and restricted window into the full contents of the position on the board.
It is instructive to watch how plans dynamically change and new plans appear as the engine goes deeper into the position [above: Aronian-Carlsen, round six]. Slowly you begin to see the most important elements of the position and a feel for how plans are developed.
Above is an example from Shirov-Ponomariov, round six. Another new feature of Fritz 10 is the main line of the move the engine is currently looking at, displayed below the main line of the best move found so far. You can watch while each "try" is refuted in as the engine goes deeper.
In the above example White has 40 legal moves, and Fritz is looking at the 20th option. It is searching at a depth of 18 ply full width, with some lines going as deep as 51 ply (a "ply" is a half move, one by black or by white). Its evaluation of the best move found so far is +1.0 pawns for White. The search speed on our somewhat outdated computer is 750 kN/s = thousand nodes (positions) per second, and the total number of positions Fritz examined to come up with the currently best line is 39,642,000 – yes, thirty nine million in fifty seconds.
