Fritz and the Grandmasters By Andre Schulz It was a GM tournament like any other. Hectic activity, the organisers scurrying around, solving final problems, players slightly nervous, waiting for the pairings. Only one participant was completely uneffected by all of this. Just two feet tall, with only one request - to always play in the proximity of a 220 volt power socket. Incredible: a regular commercial Fritz3 program, taken out of the standard box, playing in a full Grandmaster tournament under normal time controls. The chances are anything but good for the computer. True, the same sleek Pentium-based Olivetti PC wrecked havoc among the world's top players in Munich and London, but that was blitz and rapid chess. At two hours for 40 moves Grandmasters are able to avoid the kind of mistakes made at faster speeds and use their strategic understanding to outplay the machine. Everyone knows that in the extended time controls of tournament chess humans reign supreme. The computer can expect to get zero points out of eleven rounds, maybe one point if some poor devil drops a piece, in time trouble perhaps. Some Grandmasters actually show traces of respect. "Computer beat Kasparov, no?" asks GM Glek. In blitz games, he is told. "Okay, no problem." The first opponent, Edvins Kengis, Elo 2575, begins his game with 1.b3. "Who knows how much openings theory the machine knows." This is the way to take it out of "book" and make it rely on its own primitive algorithms. But Fritz holds its ground and even offers a pawn for compensation in the middlegame. Kengis is afraid of the complications and goes for a draw by repetition. Relieved he pats the Pentium monitor and asks, "Do you want to analyse?". Laughter all around. Edvins Kengis - Fritz3: [A01: Larsen Opening] *1.b3 e5 2.Bb2 d6 3.d4 exd4 4.Qxd4 Nc6 5.Qe3+ Be6 6.Nf3 Nf6 7.Ng5 Qd7 8.Nxe6 fxe6 9.g3 d5 10.Bg2 Bb4+ 11.Nd2 0-0 12.0-0 Qe7 13.Nf3 Bc5 14.Qg5 h6 15.Qg6 Rad8 16.e3 e5 17.Nh4 Qf7 18.Rad1 Qxg6 19.Nxg6 Rfe8 20.c4 d4 21.exd4 Bxd4 22.Bxd4 Nxd4.* DIAGRAM (wKg1,Ng6,Bg2,Rd1,f1,Pa2,b3,c4,f2,g3,h2; bKg8,Nd4,f6,Rd8,e8,Pa7,b7,c7,e5,g7,h6) 22...exd4 looks perfectly normal. Why does Fritz give up the pawn on b7? After 23.Bxb7!? we have: A) 23...Kh7 (probably best) 24.Nh4 c6 25.Kh1 (25.Kg2 Rb8 26.Rxd4 exd4 27.Bxc6 Re2 28.Nf3 Rd8 29.Rd1 Ng4 30.Rd2 d3 31.Bd5) 25...Ne4 26.Rxd4 exd4 27.Bxc6=; or B) 23...c6 24.f4 e4 25.Ne5; or C) 23...Kf7 24.Nh4 c6 25.Kh1! g5 (25...Rb8 26.f4!; 25...Re7 26.f4!) 26.Nf3!! (Fritz) Nxf3 27.Bxc6. *23.Rfe1 e4 24.h3 c5.* According to Kengis 24...c6 is better, with the idea 25...Kf7 26.Nf4 g5 27.Nd5 Nf3 28.Bxf3 cxd5 and 29...d4; or 27.Ne2 Nc2 28.Rf1 e3, in each case with an advantage for Fritz. But also after 25.Nf4 g5 26.Nd5 Nf3+ 27.Bxf3 cxd5 28.Bh1 d4 Black is better. *25.Kh2 b6 26.g4 Kf7 27.Nf4 Nc2 28.Rxd8 Rxd8 29.Re2 Nd4 30.Re3 Nc2 31.Re2 Nd4 32.Re3 Nc2 33.Re2 draw.* In the second round the computer is faced with a Slav Queen's Gambit and a very well-prepared opponent. International Master Roman Slobodjan, Elo 2465, plays a fine novelty and Fritz loses touch with the position. Like a tigress it protects a totally irrelevant pawn on h2 instead of sacrificing it for the attack. The endgame is hopelessly lost. The third round is a fiasco for Fritz. In an exchange variation of the Queen's gambit a simple position arises with clear strategic elements and no tactical complications. IM Alfred Kertesz, Elo 2365, can cash in on all principle weaknesses of computer play and take Fritz apart like a beginner. Alfred Kertesz - Fritz3 [D35: Queen's Gambit, Exchange Variation] *1.c4 e6 2.Nf3 Nf6 3.Nc3 d5 4.d4 Be7 5.cxd5 exd5 6.Bf4 c6 7.Qc2 Na6 8.a3 Qb6 9.e3 Bg4 10.b4 0-0 11.Ne5 Bh5 12.Bd3 Nb8 13.0-0 Qd8 14.Na4 a6 15.Rab1 Re8 16.Nc5.* DIAGRAM (wKg1,Qc2,Nc5,e5,Bd3,f4,Rb1,f1,Pa3,b4,d4,e3,f2,g2,h2; bKg8,Qd8,Nb8,f6,Be7,h5,Ra8,e8,Pa6,b7,c6,d5,f7,g7,h7) A position that Fritz doesn't understand at all. There is nothing to calculate, the long-term threat is the entrapment of the bishop on h5, which the program will not recognise until material loss appears on the horizon. For the moment it has only one concern: to cover b7. *16...Ra7 17.Rfc1 Bd6 18.Bf5 h6 (18...Bg6 19.Nxg6 hxg6 20.Bxd6 Qxd6) 19.h3 b6 20.Na4 Bxe5 21.Bxe5 Nfd7 22.Bh2 Rb7 23.g4 Bg6 24.Bxg6 fxg6 25.Qxg6 Nf8 26.Qd6 Qh4 27.Qg3 Qf6 28.Qg2 Ng6 29.Bg3 Nh4.* Fritz offers the exchange of the knight in order to get rid of the troublesome bishop which is threatening b8. Perhaps it would have been wiser to retain the knight and with it some swindle chances. *30.Bxh4 Qxh4 31.Nb2 h5 32.Nd3 hxg4 33.Qxg4 Qxg4+ 34.hxg4 g5 35.Ne5 Re6 36.Kg2 Rg7 37.Kg3 Kf8 38.f4 gxf4+ 39.exf4 Nd7 40.Nxd7+ Rxd7 41.Re1 Rde7 42.Re5 Rxe5 43.dxe5 Ke8 44.Rc1 Re6 45.g5 Rg6 46.Kg4 Re6 47.Kf5 Kf7 48.Rh1 Re8 49.Rh7+ Kg8 50.Rd7 c5 51.bxc5 bxc5 52.Rxd5 Rc8 53.Ke6 c4 54.Kd7 Rb8 55.Rc5 1-0.* Disaster. Was the draw in round one just a fluke? In the ChessBase offices in Hamburg the makers of Fritz feverishly analyse the positions the program misplayed. They issue new directives to Fritz for the opening: "With White only play 1.e4, with Black try to play the Tarrasch against 1.d4, 1.Nf3 and 1.c4. Avoid blocked positions and strive for positions with an isolated pawn." The ChessBase people knew from past games that Fritz plays well with an isolated pawn - on either side. If it has the weak pawn it defends it imaginatively and at the same time develops dangerous play on the open lines. If it is the opponent then Fritz has a clear goal and attacks the isolated pawn viciously, developing something that looks remarkably like the deep strategic planning of strong human players. Will the strategy work? In the fourth round the Pentium is paired against GM Yury Piskov, Elo 2540. Fritz plays 1.e4 and the game continues 1...e6 2.d4 d5 - the French Defence. But how do you get an isolated pawn in this opening? Very simple: 3.exd5 exd5 4.c4 and voila! Piskov plays his usual Grandmaster game, aesthetic and sound. But Fritz is unrecognizable. Gone is the novice of rounds two and three, the program really seems to understand things, develops its pieces, threatens, launches attacks, all the while deftly protecting its isolated pawn. In the endgame the GM presses too hard - why shouldn't he win like everybody else? - and Fritz is able to capitalise on his pawn weaknesses. It is the first GM scalp of the tournament. Fritz3 - Yuri Piskov (2540) [C01: French Defence, Exchange Variation] *1.e4 e6 2.d4 d5 3.exd5 exd5 4.c4 Nf6 5.Nc3 Be7 6.cxd5 Nxd5 7.Nf3 0-0 8.Bd3 Bg4 9.0-0 Nc6 10.Bb5 Ndb4 11.Be3 a6 12.Be2 Nd5 13.Bc4 Be6 14.Bb3 Na5 15.Nxd5 Bxd5 16.Bxd5 Qxd5 17.Bf4 Bd6 18.Bxd6 cxd6 19.b3 Rfe8 20.Qd2 Nc6 21.Rfe1 h6 22.Rac1 f5?! 23.Rcd1 a5 24.Qc3 b5 25.Rc1 Nb4? 26.a3 Na2 27.Qc6.* DIAGRAM (wKg1,Qc6,Nf3,Rc1,e1,Pa3,b3,d4,f2,g2,h2; bKg8,Qd5,Na2,Ra8,e8,Pa5,b5,d6,f5,g7,h6) This is the kind of position Fritz really likes: lots of points of attack, lots of calculation to do. *27...Qxc6 28.Rxc6 Rxe1+ 29.Nxe1 Re8 30.Kf1 b4 31.axb4 Nxb4 32.Rxd6 Rc8 33.Rb6 Rc1 34.Ke2 Rb1 35.Rb5 Rb2+ 36.Kd1 Nc6 37.Nf3 Nb4 38.Nd2 f4 39.Rxa5* (Fritz knows its winning) *Nd3 40.f3 g5 41.d5 Kf7 42.d6 Ke8 43.Rd5 Nb4 44.d7+ Kd8 45.Rd6 Ra2 46.Rxh6 Nd5 47.Ne4 Rxg2 48.Kc1! Ne3 49.Rh7 g4 50.fxg4 f3 51.Rf7 f2 52.Nxf2 Nxg4 53.Nd3 1-0.* Fritz saves the game without any display of emotion. So I beat a Grandmaster - great. The players are puzzled. How is it possible that the computer can play like a rank amateur in one round and like a Grandmaster in the next? "Doesn't that happen to humans as well?" asks the Fritz operator. The strategic instructions from Hamburg are of course not revealed to the tournament participants. In round five Fritz chooses an unsound variation of the Tarrasch, and Joachim Wintzer, Elo 2335, wins a pawn in the opening. The endgame is theoretically won for the human, but Wintzer is unable to overcome the tenacious defence of the computer. A pawn down, Fritz eventually even starts to play for a win. However, in the opponent's time trouble Fritz graciously accepts move repetition and a draw - it has not been instructed to relinquish its objectivity and try to swindle the opponent if he is running out of time. Joachim Wintzer - Fritz3 [D33: Queen's Gambit, Tarrasch Defence] 1.c4 e6 2.Nc3 d5 3.d4 c5 4.cxd5 exd5 5.Nf3 Nc6 6.g3 c4 7.Bg2 Bb4 8.0-0 Nge7 9.e4 0-0 10.exd5 Nxd5 11.Nxd5 Qxd5 12.Be3 Bg4 13.Ne5 Qxg2+ 14.Kxg2 Bxd1 15.Nxc6 bxc6 16.Rfxd1 Bd6 17.Rac1 Rfb8 18.Rxc4 Rxb2 19.Rxc6 Rd8 20.Ra6 Rd7 21.Rd2 Rxd2 22.Bxd2 f5 23.Kf3 g6 24.Ra4 Kf7 25.Bf4 Be7 26.Ke3 Bd8 27.Kd3 Ke6 28.Bd2 Kd5 29.Ba5 Be7 30.Bb4 Bd8 31.Bc5 a5 32.Rc4 Rb7 33.Ba3 Rc7 34.Bc5 Rb7 35.Rc2 Bf6 36.Re2 Rb1 37.Rd2 g5 38.f3 f4 39.Ba7 Rb7 40.Bc5 Rb1 41.Ba7 Rb7 42.Bc5 Rb1 draw. The next two rounds against Maiwald and Podzielny are fighting draws. Karl-Heinz Podzielny, Elo 2455, tries for 100 moves to push the Pentium over the edge. Fritz appears somewhat preoccupied ("It's probably flirting with the disk controller again," complained the operator) and must defend an unpleasant endgame, but it does so with great virtuosity, with study-like manoeuvres, driving the unfortunate Podzielny to near distraction. "I think I've forgotten how the rooks move", he said after the titanic eight-hour struggle. The eighth round is very exciting. The top seed Igor Glek, Elo 2590, avoids his usual French (after what happened to colleague Piskov) and opts for a sharp Sicilian against Fritz. The Russian GM usually covers his eyes with his hands, so the opponent cannot see where he is looking. But with the Pentium it is different and Glek's eyes dart around the castled king of his opponent. Then, like a bolt of lightening, the bishop is sacrificed on h3, with three pawns and a devastating attack in compensation. This is vintage GM tactics, and any normal human opponent would quickly succumb to the pressure. But of course Fritz is not human. The program coolly calculates: "I got a bishop, worth three pawns. He got three pawns in return, so that's okay. I cannot see how he can mate me, but I do see a stray pawn on b7. Perhaps I can snap that up." It plays its queen from d1 to b3, grabs the pawn and then returns with a florish to h2. Glek stares in disbelief, like a puppy that had a bone moved back and forth in front of its face and doesn't get hold of it. Afterwards he complains of a headache. "I've never had to do so much calculation before in a game. All that work, and in the end I had nothing." GM Rainer Knaak after playing through the moves: "It's a lovely game between two 2600 players." Fritz3 - Igor Glek [B22: Sicilian Defence, Alapin Variation] *1.e4 c5 2.c3 e6 3.d4 d5 4.exd5 exd5 5.Nf3 a6 6.Be2 Nf6 7.dxc5 Bxc5 8.0-0 h6 9.Nbd2 0-0 10.Nb3 Bb6 11.Re1 Nc6 12.Nbd4 Re8 13.h3 Qd6 14.Be3 Bxh3 15.gxh3 Rxe3 16.Nf5 Qf4 17.Nxe3 Bxe3 18.fxe3 Qg3+ 19.Kh1 Ne4* (19...Qxh3+ 20.Nh2 Ne4 21.Qxd5) *20.Rf1 Qxh3+* (20...Nf2+ 21.Rxf2 Qxf2 22.Qd2 Re8 23.Rf1 Qxe3 24.Qxe3 Rxe3 and White is better) *21.Nh2 Qxe3 22.Bf3 Ng3+ 23.Kg2 Nxf1 24.Nxf1 Qg5+.* White must play very accurately if he is not to lose. But exactly this is the strong point of the program. *25.Kh1* (25.Kf2? Rd8 26.Qb3 Ne5 27.Nh2 Nd3+ with advantage for Black) *25...Rd8 26.Qb3* (Of course Fritz is after the pawn on b7) *26...Ne5* (26...b5!? 27.Rd1) *27.Nh2 Nxf3 28.Nxf3 Qh5+ 29.Kg2 Rd6 30.Qxb7.* DIAGRAM (wKg2,Qb7,Nf3,Ra1,Pa2,b2,c3; bKg8,Qh5,Rd6,Pa6,d5,f7,g7,h6) "Unbelievable," says Glek. "This simply cannot work." *30...Rf6 31.Qb8+ Kh7 32.Rf1 Qf5 33.Qh2.* "Unbelievable, now I have nothing" (Glek) *33...Qe4 34.Qh3 Rf4 35.Kg3 f6 36.a3 g5 37.Qd7+.* 37.Nxg5+? is not possible because of 37...fxg5 38.Qd7+ Kg6 39.Qd6+ Kh5 40.Rxf4. After QxR and QxQ the endgame would be won for White. However there is also 40...Qe3+! 41.Rf3 Qg1+ 42.Kh3 g4# (Fritz). *37...Kg6 38.Qc8 h5 39.Qg8+ Kh6 40.Qh8+ Kg6 41.Qg8+ Kh6 42.Qf8+ Kg6 43.Qg8+ draw.* In round nine it is Detlev Heinbuch's turn. The Elo 2355 IM has carefully prepared an anti-computer variation against his own personal copy of Fritz3, which can be bought in any chess shop and is identical to the version used in the tournament. He does everything correctly, locking the position, moving his pieces to the kingside and launching his g-pawn to smash open the opponent's king fortress. Just like he has done a number of times at home. Well, almost. The problem is that Fritz refuses to castle. The program seems to sense the danger. At one stage it plays Rh8-f8. Heinbuch thinks the operator may be mistaken and Fritz must have castled. When it plays Ra8-c8 he enquires again: "Not castles? Are you sure?" Sorry, Fritz needs its king in the middle. Heinbuch's brutal attack wins a pawn, but Fritz finds all the tricks in the position. In the endgame the human still has the extra pawn, but Fritz uses its initiative and the bishop pair to full effect. It wins back the pawn and then boldly marches its king across the board to finish off the opponent. "He was outfritzed," explains the operator, who has seen this kind of thing on numerous occasions before. Detlev Heinbuch - Fritz3 [D00: Queen's Pawn Game] *1.d4 d5 2.e3 Nf6 3.Bd3 c5 4.c3 c4.* In positions like this all chess programs make the same strategical mistakes. *5.Bc2 e6 6.Nd2 Be7 7.f4.* This is the typical anti-computer system that specialists employ. *7...Qc7 8.Qf3 Nbd7 9.g4 h6 10.Qh3 Nb6.* Fritz is playing like a beginner. It doesn't notice what is brewing. *11.Ngf3 Bd6 12.Ne5 Rf8.* Now Fritz can smell something fishy and will not commit itself to castling kingside. *13.Ndf3 Bd7 14.g5 Ng8 15.gxh6 gxh6 16.Bh7 Nf6 17.Qxh6 Nxh7 18.Qxh7 Ba4 19.Bd2 Rc8 20.Rg1 Nd7 21.Nxd7 Qxd7 22.Ke2 f6!?* Fritz decides to go for the endgame. *23.Rg7 Rf7 24.Qg6.* DIAGRAM (wKe2,Qg6,Nf3,Bd2,Ra1,g7,Pa2,b2,c3,d4,e3,f4,h2; bKe8,Qd7,Ba4,d6,Rc8,f7,Pa7,b7,c4,d5,e6,f6) *24...Bc2!* With this tactical trick the program occupies the best diagonal for its bishop. *25.Qxf7+* (25.Qh5 Bf8 26.Rg8 Kd8 27.Rag1 Kc7 and Black is better) *25...Qxf7 26.Rxf7 Kxf7 27.Be1 Bd3+ 28.Kd2 Be4 29.Ng1* (29.Ke2 Bd3+ 30.Kd2 - Fritz may have taken the draw, after all it's a pawn down) *29...Rh8 30.Bg3 b5 31.Ne2 b4 32.Kc1 Bd3 33.Kd1 Rb8 34.Be1 bxc3 35.Bxc3 Rh8 36.Kd2 Rxh2 37.Re1 Kg6.* Fritz, with not a trace of fear in its heart, throws its king into the fray. *38.Kd1 Kf5 39.Bd2 Ke4 40.Nc3+ Kf3.* "This machine knows no fear," exclaims Heinbuch. *41.Nb5 Bb8 42.Nc3 Bc7 43.Na4 Kg3 44.Nc5 Bf5 45.Nd7?* The endgame was unpleasant enough for White, but it is this move that actually loses. *45...Bg4+* and White resigned because of 46.Kc1 Rxd2 47.Kxd2 Ba5+. 0-1. Fritz has 50% and is on its best way to get an IM norm. In round ten Sascha Grimm (Elo 2315), the only untitled player in the tournament, tricks the program into committing a typical computer error. Fritz grabs a pawn on b7 and Grimm can trap its rook. Fritz finds all kinds of tricks to prevent the actual capture of the rook, but it gets a "lost" endgame, which it naturally holds to a draw. In the last round Fritz needs just a draw for the IM norm. The opponent is the Lithuanian Grandmaster Aloyzas Kveinys, Elo 2525. He offers to assist operator Christoph Wulfken in unloading the Pentium: "Can I carry Fritz for you?" Christoph is a bright lad and refuses: "No, no, you might just drop it!" In the game Fritz is never in any trouble and makes his GM opponent work hard. The program refuses a draw by repetition with the rook on the second rank, and receives an admonition from kiebitzing GM Kengis: "Come on, Fritz, take the draw, then we go and drink to your IM norm!" Soon after Fritz does accept the draw and Kveinys breaths a sigh of relief: I got a draw against Fritz! Fritz3 on the Pentium is the first PC program to earn an IM norm in a regular Grandmaster tournament. In fact the International Computer Chess Association has no knowledge of any machine ever earning an IM norm, in any kind of tournament. The overall Elo performance is 2452, but interestingly it was much higher against the stronger players. If you take only the GMs (Glek, Piskov, Kengis, Kveinys) into account the performance is 2652. This is because the weaker players were in general the ones who undertook special preparations with their private copies of the program; the GMs tended to rely on their experience and general strength. The tournament direction is very pleased with the event, as are the players. They consider Fritz an interesting experience. The collective assessment after the tournament: "A very strong opponent, with a slightly instable playing style, tactically extremely dangerous, incredibly tenacious in worse positions, good endgame treatment. Perhaps a bit uncommunicative, but generally quite pleasant. It doesn't smoke, sweat or smell, doesn't talk during the game, doesn't jiggle or totter, doesn't kick you under the chair or insult you after the game. It's fun to play against it, and a real challenge." After the prize-giving ceremony there are some blitz games. Kveinys and Fritz have become friends and play a quick match. The result was 3:1 for Fritz, with one game going like this: Aloyzas Kveinys - Fritz3: [C22: Center Gambit] *1.e4 e5 2.d4.* A mistake against Fritz - the position is too open. *2...exd4 3.Qxd4 Nc6 4.Qe3 Nf6 5.e5 Ng4 6.Qe4* (6.Qe2 d6!) *6...Ngxe5!?* A theoretical novelty. But doesn't Black lose a piece? 6...d5 7.exd6+ Be6 8.Ba6 Qxd6 is better for White. *7.f4 d5 8.Qe3? 8.Qe2* had a better chance of punishing the greed of the program. *8...Bg4 9.Nf3 Bxf3 10.gxf3 Qe7* (10...Qh4+ 11.Kd1 f6 12.fxe5 fxe5) 11. fxe5 Nd4 12.Qf2 Qxe5+ 13.Kd1) *8...d4. 9.Qe2 Bb4+ 10.c3 Bg4.* DIAGRAM There is one more move that delays winning the piece. *11.Nf3.* 11.Qd2 dxc3 12.bxc3 (12.Qxd8+? Rxd8 13.Nxc3 Rd1+ 14.Kf2 Bc5+ 15.Kg3 Rxf1 16.fxe5 Be6 is winning for Black) 12...Qe7 13.fxe5 Rd8 14.Qc2 Rd1+ 15.Qxd1 (15.Kf2 Qxe5-+) 15...Bxd1 16.Kxd1 Qxe5 and Black can win. *11...Bxf3 12.gxf3 dxc3 13.bxc3 Qh4+ 14.Qf2.* At last! After 14...Qxf2 15.Kxf2 two pieces, Ne5 and Bb4 are hanging. *14...Bc5! 15.Qxh4 Nxf3+ 16. Kd1 Nxh4* and Kveinys resigned after a few more moves. 0-1. Even Karl-Heinz Podzielny who has remained somewhat suspicious about the machine, sits down for a game. One of Germany's strongest blitz players, he promises: "If I lose this one I'll never play blitz again." During the game he hurls insults at the Pentium: "Now I've got you, scumbag. Missing a piece, sardine tin? What are you going to do now? What does he say?" Podzielny cannot see the screen, so Kengis helps out: "He says you will be mate in four!" Here is the historical last blitz game of "Potzblitz" Podzielny: Podzielny - Fritz3: 1.e3 d5 2.Nf3 Bg4 3.b3 e5 4.h3 Bxf3 5.Qxf3 e4 6.Qe2 Qf6 7.Nc3 c6 8.Bb2 Nd7 9.0-0-0 Qg6 10.f3 0-0-0 11.g4 Ngf6 12.Qg2 Bd6 13.f4 h5 14.f5 Qg5 15.Be2 hxg4 16.hxg4 Ne5 17.Kb1 Rdg8 18.d3 exd3 19.cxd3 Qxe3 20.g5 Rxh1 21.Rxh1 Nxd3 22.gxf6 Nxb2 23.Rh3 Qd4 24.Kxb2 and Fritz announces mate in four. 24...Qd2+ 25.Kb1 Ba3 26.Na4 Qc1 mate. 0-1. What is the future going to bring? Are we soon going to see a full-blown IM Fritz, or even a computer Grandmaster? Will Fritz demand an appearence fee or threaten to leave the tournament site if special requests are not fulfilled? (A member of the Godesberg organisation claims that Fritz demanded a CD with pictures from the Caribbeans for relaxation in the hotel room). And will we have to stick a bill into the drive slot before it plays a tournament game against us patzers? Thank heavens not. We predict that Fritz will remain the way it has always been, modest and unassuming, always prepared to play a nice game of chess. Final standings of the Third Godesberg GM Tournament =========================================================================== Player Elo 1 2 3 4 5 6 7 8 9 10 11 12 Pts 1. GM Igor Glek (RUS) 2590 = = = = 1 = 1 1 1 1 1 8,5 2. GM Yury Piskov (RUS) 2540 = = = = = 0 1 = 1 1 1 7,0 3. GM Edvins Kengis (LAT) 2575 = = = = 0 1 1 1 = = = 7,0 4. IM Roman Slobodjan (GER) 2465 = = = = 0 1 1 1 = = = 6,5 5. IM K.-H. Podzielny (GER) 2455 = = = = 1 = = 0 = 1 1 6,5 6. IM J.-U. Maiwald (GER) 2455 0 = 1 1 0 = 0 = = 1 1 6,0 7. Fritz3 (GER) = 1 = 0 = = = = 0 = 1 5,5 8. GM Aloyzas Kveinys (LIT) 2525 0 0 = 0 = 1 = = 1 1 = 5,5 9. FM Joachim Wintzer (GER) 2335 0 = 0 0 1 = = = = 1 = 5,0 10. IM Alfred Kerzesz (GER) 2365 0 0 0 = = = 1 0 = 0 = 3,5 11. Sascha Grimm (GER) 2315 0 0 0 = 0 0 = 0 0 1 1 3,0 12. IM Detlev Heinbuch (GER) 2355 0 0 0 = 0 0 0 = = = 0 2,0 ============================================================================ Fritz for beginners The tournament report on the previous pages makes it clear that Fritz3 is an awesomely powerful chess program. It is able to defeat over 99% of all chess players. So does this mean that you, as an amateur, can expect nothing but merciless beatings? Not at all. Fritz has a number of artificially weak ("handicap") levels on which the computer plays like a rank amateur. On one level it may move instantaneously and display a great deal of general short-sightedness. On another it will move aggressively with the queen and neglect the development of the other pieces completely, and on yet another it is solely bent on pushing pawns. This is exactly how beginners tend to play, so you will be learning how to counteract that kind of strategy. It is also possible to define your own handicap levels. This was done at a press conference in Hamburg. Fritz was scheduled to play against the Hamburg soccer star Felix Magath, who loves chess but is a rank amateur. Fritz was programmed to move its king 80% of the time, at the cost of a pawn if necessary. The game that developed turned out to be a real jewel, one Magath will treasure for a long time. Felix Magath - Fritz3 (handicap level): 1.e4 e5 2.Nf3 Nf6 3.Bb5 Ke7. Fritz has been instructed to move it's king and does so the first real opportunity it gets. 4.d4 Ke8 5.dxe5 Nxe4 6.0-0 Ke7 7.Re1. Libero Magath is quietly mobilising his forces while Fritz wastes time moving its king. 7...a6 8.Ba4 b5 9.Bb3 Bb7. The program had to defend the knight on e4. That is worth more than the privilege of another king move. 10.c4 Ke8 11.Nbd2 bxc4 12.Nxc4 Ke7. Fritz is simply following instructions and has obtained a terrible position. DIAGRAM 13.Rxe4! Very imaginative play by Magath, who sacrifices the exchange to get at the black king. 13...Bxe4 14.Bg5+ f6. Forced, since a king move would lose the queen behind it. 15.exf6+ gxf6 16.Qd4! A double attack on the pawn on f6 and the bishop on e4. The amateur Magath is having great fun and developing very nice ideas. 16...Kf7 17.Qxe4 d5 18.Qf4 dxc4 19.Bxc4+ Kg7. Now the mate is forced and Magath finds all the best moves. 20.Bh6+ Kg6 21.Nh4+ Kh5 22.Bf7 mate! The audience stood up and cheered this lovely final position, and Magath swore to play much more chess in the future.