A BREEZE IN THE SLEEPY 4-KNIGHT'S GAME
A Guided Tour of Chess - The Chess Cafe, March 2000

In one of my very first games at ICC, in fact when it was still ICS, a certain Blokje played the following against me: Blokje - Platypussy, ICS 2 12 blitz, 1995: 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 (See Diagram) 4.Nxe5 This irritated me. I vaguely knew about this move - not as a real opening, but as some sort of a student's joke, something silly that might be played at the end of a long blitz-and-boozing session. Wasn't this the 'Irish Gambit', the one with the anecdote where they asked the inventor on his death bed how he had thought of it, and he had answered: 'I didn't see it was protected'? The typically Dutch dimunitive suggested Mr. Blokje was a compatriot - so what was this: Dutch bragging?
    Anyway, the game continued 4...Nxe5 5.d4 Nc6 6.d5 Ne5 7.f4 Ng6 8.e5 Ng8 9.Bd3 Bb4 10.O-O Bxc3 11.bxc3 d6 12.e6 fxe6 13.dxe6 Nf6 14.g4 O-O 15.g5 Ne8 16.f5 Ne5 17.Re1 Nxd3?? 18.e7 and a little later I resigned.
    I was convinced I had only lost because of the aggravation, and I remembered this game (and the other two where 4.Nxe5 had been played against me at ICC, and which I had both lost in 14 moves) when not too long ago, there was a discussion on rec.games.chess.misc about whether it is impolite for Black to answer 1.e4 with 1...a6.
    The idea that an opening can be impolite is interesting. You immediately wonder if there are equivalents in other sports. The underhand service in tennis comes to mind (the Chang Variation), the break-away at the start of a bicycle race, the goalkeeper taking a penalty kick in soccer. They are demonstrations, not dangerous for the opponents, but they do disturb the normal course of events, and the irritation they cause favors the impolite.
    In Internet-blitz, it turned out, there are Whites who, if they don't care about their ratings, immediately resign after 1...a6 because it keeps them from practicing their repertoire. But an underhand service is not in itself losing, and neither is 1...a6. Miles played it in a famous game against Karpov, and won. I've published that game with a question mark for a6 - a question mark for bad manners. Karpov was the reigning World Champion - you don't order pizza when you have the King over for dinner. Karpov's loss in that game must have been partly due to his irritation over 1...a6.

When I decided to write something about 'impolite openings' for my Dutch newspaper column, it was going to be about 1...a6 and 1...g5 and the like, but I also added something about this 4.Nxe5.
    In fact, I did know 4.Nxe5 existed, and even had a name: the 'Müller-Schulze Gambit'. (Irish Gambit, Chicago Gambit or Razzle Dazzle Gambit are names for an even more nonsensical opening: 1.e4 e5 2.Nf3 Nc6 3.Nxe5). 'Müller - Schulze Gambit' is mockery; in German, 'Müller and Schulze' (sometimes 'Meier, Müller, Schulze and Schmidt') stands for 'everybody', comparable to 'Tom, Dick and Harry' in American. Volume 11 of Euwe's opening's series (my mid-fifties edition of it) has a little paragraph about the Müller-Schulze gambit, where 4.Nxe5 gets a question mark, and where it says that after 4... Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 d5 8.Bxd5 c6, Black has a decisive advantage.
    When I looked for the Müller - Schulze in my database, hoping for some prehistoric offhand games, I was in for a huge surprise: I found over three thousand games with it; most of them recent blitz games played by a computer named Brause at ICC and other servers - and in which Brause had scored 73 %, occasionally even beating masters and grandmasters.
    Randomly playing over a few of those games, I was even more astonished when I saw how deadly this gambit could be. For one thing, Brause had won close to three hundred Müller-Schulze games in 15 moves or less - I had stumbled upon an unsuspected treasury of light openings brilliancies.
Brause - Betrueger, GICS (the German server) blitz, 1997
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Nc6 6.d5 Nb8 7.e5 Ng8 8.d6 c6 9.Bc4 f6 10.Qh5+ g6 (See Diagram) 11.exf6 Qxf6 12.Qe2+ Kd8 13.Ne4 and Black resigned.

But it was not only Brause who played the Müller - Schulze. I also found a few hundred Internet blitz games by humans who, mostly after having been confronted with this gambit by Brause, had started to play it themselves; the most prominent being Women's Grandmaster Eva Repkova, wellknown at ICC as evbad whose family, she says in her notes, consists of "my husband (Fadi), my son (Christoper) and ICC".
evbad (WGM) - agro (FM), ICC 3 0 blitz, 1999
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 d6 8.Qf3 Be6 9.Qxb7 dxe5 10.Bb5+ Ke7 11.dxe5 Nxe5 12.O-O Nf6 13.Bg5 h6 14.Bh4 Rb8 15.Qxa7 Qc8 16.Rfe1 Ng6 17.Qc5+ Kd8 18.Rad1+ Bd6 (See Diagram) 19.Rxd6+ cxd6 20.Qxd6+ Bd7 21.Nd5 Qb7 22.Nxf6 gxf6 23.Bxf6+ Kc8 24.Qc5+ Qc7 25.Bxd7+ Kxd7 26.Qf5+ Kc6 27.Re3 Rhe8 28.Rc3+ Kb6 29.Bd4+ Ka6 30.Qd3+ Ka5 31.Ra3+ Kb4 32.Qb3 mate

Looking up Euwe's refutation, I saw it does not protect even strong Blacks from going astray horribly.
Brause - Ctoth (IM), ICC 5 0 blitz, 1998
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 d5 8.Bxd5 c6 9.Bb3 Be6 10.O-O Bxb3 11.axb3 f5 12.Qd3 N6e7 13.Bg5 g6 14.d5 cxd5 15.Rfe1 h6 16.Bf6 Rh7 17.Nb5 Bg7 18.Qc3 Kf8 (See Diagram) 19.Nc7 and Black resigned. A heartrending devastation.

As to my own disaster against Blokje, I discovered that after 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Nc6 6.d5 Ne5 7.f4 Ng6 8.e5 Ng8, not Blokje's 9.Bd3, but 9.d6 is the move. In MrAnonymous - RZiat (IM), ICC 4 0 blitz, 1999, there followed: 9...cxd6 10.exd6 a6 Giving back the piece; but in the Müller - Schulze, that doesn't always help. 11.Qe2+ N8e7 12.Be3 b5 (See Diagram) 13.Nd5 and Black resigned.
    Or again, in that variation: Brause - HPROL, ICC 5 0 blitz, 1998 (first 10 moves as above): 10...Qf6 11.Nb5 Kd8 12.Be3 Qe6 Other horrors can arise after 12...b6, for instance 13.Qd5 Rb8 14.O-O-O Bb7 15.Qd2 Ra8 16.Nxa7 Qe6 17.Bxb6+ Ke8 18.Re1 Be4 19.Nb5 Rc8 20.Nc7+ Rxc7 21.dxc7 d5 22.Rxe4 and Black resigned (Brause - babulina, ICC blitz, 1997) 13.Qd4 Nf6 14.Bc4 Qe4 15.O-O-O The Queen's exchange doesn't really stop White's attack. 15...Qxd4 16.Rxd4 This position had already occurred in a game Schlenker - Klostermann, Germany 1993. Schlenker (we'll meet him in a second) gave the claustrophobic line: 16...Ne8 17.Bxf7 a6 18.Re4 Bxd6 19.Bb6+ Bc7 20.Rhe1 d5 21.Rxe8+ Rxe8 22.Bxc7+ Kd7 23.Bxe8 mate. 16...a6 (See Diagram) 17.Re4 Be7 Or Nxe4 18.Bb6+ Ke8 19.Nc7+ Kd8 20.Bxf7 and mate next move. 18.Bxf7 Nd5 19.Bxd5 Ke8 20.Nc7+ Kf8 21.Bc5 and Black was lucky to get away with resigning.

Dizzily putting "Brause" in search engines, I arrived on the Website (http://www.jakob.at/steffen) of Steffen Jakob, a German chessplayer and computer programmer living in Vienna, Austria, who turns out to be responsible for all those Brause-games with 4.Nxe5.
    On his pages, Jakob says he became interested in 4.Nxe5 in 1996, when he saw an article about it in an old (1993) issue of the offbeat German chess magazine Randspringer. Brause (meaning 'sparkle', as in sodapop) was Jakob's clone of the wellknown and very strong freeware chess program crafty by Robert Hyatt.
    "I like bizarre openings, and using the playing strength of crafty combined with my opening book produced some nice games," Jakob mailed me. As "there had never been a Mr. Müller or a Mr. Schulze who played this line," he decided to give the gambit a new name: The Halloween-Attack in the Four Knight's Game. "Many players are shocked, the way they would be frightened by a Halloween mask, when they are mentally prepared for a boring Four Knight's, and then they are faced with Nxe5."
    From 1996 to 1998, Jakob concentrated almost only on his Halloween Attack, analyzing it, and letting Brause play innumerable blitz games with it. On the basis of those games, he compiled an extensive theory of 4.Nxe5 (which can also be found on his pages), using ideas for Black from strong opponents, and adding ideas for White himself. In 1998, he "felt a bit overdosed with Brause", stopped its Internet career, and shifted his attention to his own chessplaying program Hossa. Brause remained dormant until recently, but these days both Brause and Hossa are happily playing the Halloween Attack on ICC against all comers.
    One thing Jakob soon noticed is that most Blacks, strong ones included, when first confronted with the Halloween Attack, tend to allow a white pawn on d6 - we already saw a few examples. 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Nc6 6.d5 Nb8 7.e5 Ng8 8.d6 (See Diagram)
    Jakob calls this the Back-to-the-roots line, saying: 'This paralyses the black position for a long time.' That's putting it mildly: in blitz, the Terrible Pawn on d6 seems to be worth a rook - Brause as well as humans score over 90 % in the diagram.
Crekky - Toranaga (IM), ICC 3 0 blitz, 2000: 8...c6 9.Bc4 b5 10.Bb3 Qh4 Or 10...f6 11.Nd5 Na6 12.exf6 Nxf6 13.Qe2+ and Black resigned (Brause - sekere, ICC 5 0 blitz, 1998) 11.Qf3 Nh6 12.Bxh6 Qxh6 13.Nxb5 Na6 14.Qxf7+ Kd8 15.Nd4 Qg6 16.Qf3 Nc5 17.O-O Ba6 (See Diagram) 18.Bf7 Qe4 19.Qa3 Kc8 20.Qa5 and Black resigned. A good example of the strangling effect the pawn on d6 may have.
Or evbad - Swingking, ICC 3 0 blitz, 1999: (first 10 moves as above) 10...b4 11.Ne4 a5 12.O-O Ba6 13.Re1 (See Diagram) 13...h5 What else? Qh5 was threatening. 14.Nf6+ gxf6 15.exf6+ Be7 16.fxe7 Qb6 17.Qf3 f6 18.Qf5 Nxe7 19.Rxe7+ Kd8 20.Be3 c5 21.Qxf6 and Black resigned.
    Giving back the piece doesn't really help in this variation either: Brause - sveshi (FM), ICC 5 3 blitz, 1997: 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Nc6 6.d5 Nb8 7.e5 Ng8 8.d6 Nc6 9.Nb5 cxd6 10.exd6 Bxd6 11.Nxd6+ Kf8 12.Be2 Nf6 13.O-O h6 14.Re1 g6 15.Bc4 Rh7 (See Diagram)16.Bxh6+ Kg8 17.Bg5 Kg7 18.Bxf6+ Kxf6 19.Ne8+ Kg5 20.Qd2+ Kh5 21.Be2+ and Black resigned.
    And even when Black gives back the piece and manages to exchange queens, things can go wrong quickly.Brause - GMAlex (GM), ICC blitz, 1997 (first 10 moves as above) 11.Qxd6 Qe7+ 12.Be3 Qxd6 13.Nxd6+ Kf8 14.Bc4 Ne5 15.Bb3 Ne7 16.O-O-O f6 17.f4 Ng4 18.Rhe1 (See Diagram) and Black resigned.
    Or Brause - Dorobanov (FM), ICC blitz, 1997: 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Nc6 6.d5 Nb8 7.e5 Ng8 8.d6 cxd6 9.exd6 Bxd6 10.Qxd6 Qe7+ 11.Qxe7+ Nxe7 12.Nb5 Na6 13.Nd6+ Kf8 14.Bd2 Nc7 15.Bc4 Ne6 16.O-O h5 17.Rad1 h4 18.Bb4 Rh5 (See Diagram) 19.Bxe6 dxe6 20.Nf5 and Black resigned.

With that pawn on d6, the e-file is often killing: Brause - ADOLF (FM), ICC blitz, 1997 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Nc6 6.d5 Nb8 7.e5 Ng8 8.d6 cxd6 9.exd6 Qf6 10.Nb5 Na6 11.Bc4 (See Diagram)
    Brause scores 29½ out of 30 here. 11...Qe5+ Or Nh6 12.O-O g6 13.Qe2+ Kd8 14.Re1 Bg7 15.Qe7+ Qxe7 16.dxe7+ Ke8 17.Nd6 mate (Brause - saxon and Brause - roadbiker.) 12.Be3 Nf6 13.O-O b6 14.Bd4 Qf5 15.Re1+ Kd8 16.Qe2 and Black resigned.

All of this really started as long ago as in 1888, when Oskar Cordel, in his Führer der Eröffnungstheorie, was the first to mention 4.Nxe5. "Used to be played frequently in Leipzig," Cordel wrote, "under the nickname Gambit Müller und Schulze. It guarantees a fierce attack and could fittingly be called Leipziger Gambit." He gave: 4...Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 c6 ("Possible is also d5 8.Bxd5 c6") 8.Qf3 d5 9.exd6 Nf6 ("After Qf6 10.Qe2+ Kd8 11.Ne4 Qxd4 12.Be3 Qe5 13.0-0-0 White has a good attack") 10.Qe2+ Kd7 and concluded: "Whether the attack is worth the difference in pieces, is questionable."
    In 1916, this was quoted by Schlechter in his 8th edition of Bilguer's Handbuch. He wrote: "Finally, mention should be made of the original sacrifice 4.Nxe5 (which used to be played often, and was jokingly called "Gambit Müller and Schulze") which, according to Cordel, leads to a fierce attack, but which should finally not suffice."
    One can hardly criticize Euwe, in his openings series, and Keres, in the first (1974) edition of the Encyclopedia of Chess Openings, for not taking the Müller - Schulze seriously and, again, sticking to Cordel's analysis. Keres devoted one line to the gambit, simply adding "-+" to Cordel's 7...d5 - Euwe, as we saw, said that after 7...d5 8.Bxd5 c6 'Black has a decisive advantage,' adding that 7...c6 8.Qf3 d5 9.exd6 Qf6 10.Qe2+ Kd8 is 'another refutation'.
    That Cordel hit the nail on the head is the perhaps surprising conclusion that can be drawn from Brause's Internet practice. Cordel's 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4. Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 is indeed the most dangerous line for White, and the position after 7.Bc4 is crucial. (See Diagram) 7...d6, 7...Bb4 and especially Cordel's own 7...c6 are then good continuations, but the value of the 'refutation' 7...d5 is disputable. Often, a game develops where White has two central pawns against a knight, not unlike the Cochrane Gambit (1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nxf7) which came out of the closet when Topalov played it against Kramnik in Linares 1999 (draw, 31 moves), or Bronstein's famous 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Na5 6.d3 h6 7.Nf3 e4 8.dxe4 against Rojahn in the 1956 Olympiad (1-0, 40 moves.)
    An example with 7...d5 (from the diagram above): evbad - Ceres (IM), ICC 5 0 blitz, 1998: 8.Bxd5 c6 9.Be4 f5 10.Bf3 Nh4 11.O-O f4 Black is giving too many pawns. 12.Bxf4 Nxf3+ 13.Qxf3 Ne7 14.Rad1 Nd5 15.Ne4 Bf5 16.c4 Nxf4 17.Qxf4 Bxe4 18.Qxe4 And White is already winning. (See Diagram) 18...g6 19.f4 Bg7 20.f5 gxf5 21.Qxf5 Rf8 22.Qh5+ Kd7 23.Qg4+ and White won.
    After other 7th moves too, White often gets enough pawns and play. Here is one with 7...d6: Brause - JudgeDredd (GM), ICC 5 0 blitz, 1997 (again from the diagram above): 8.Qf3 Qd7 9.O-O Qf5 10.Qe2 dxe5 11.f4 e4 12.Nxe4 Be7 13.c3 Nf6 14.Bb5+ Bd7 15.Nxf6+ Qxf6 16.Bxd7+ Kxd7 17.f5 Nf8 18.Bf4 Qc6 19.Qc2 Bd6 20.Qb3 Re8 21.Bxd6 cxd6 22.Qxf7+ Re7 23.Qb3 d5 24.Rae1 Kd8 (See Diagram) Black keeps giving pawns, but it does not stop the pressure. 25.Qa3 Rxe1 26.Rxe1 Nd7 27.Qe7+ Kc7 28.Qxg7 Qf6 29.Qxf6 Nxf6 30.Re7+ Kc8 31.h3 h5 32.Kh2 Rh7 33.Rxh7 Nxh7 34.Kg3 Kd7 35.Kh4 Nf6 36.g4 and White won. Agreed; this (and many other Brause games) is as typical of the Halloween Attack as of computer play.
    In one of Cordel's old variations with 7...c6 (1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4. Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 c6 8.Qf3 d5 9.exd6 Qf6 10.Qe2+ Kd8 11.Ne4 Qxd4), Brause deviates with 12.Bxf7 (Cordel: 12.Be3), winning quickly against BuffaloJim: 12...Nh6? 13.Bg5+ Kd7 14.Rd1 Qe5 15.Nc5+ and Black resigned (ICC 4 3 blitz, 1998.)
    Another disaster with 7...c6: Brause - Pfiffigunde, GICS blitz, 1997 (first 8 moves as above): 8...f6 9.O-O d5 10.exd6 Bxd6 11.Ne4 N8e7? (See Diagram) 12.Qxf6! gxf6? 13.Nxf6+ Kf8 14.Bh6 mate. Brause won this game more than once.
    A critical line here, instead of 8...f6, is 8...d5 9.exd6 Be6! and White does not have much. Brause only scored 42 % against that, which may have been the reason Jakob switched to 8.Qe2. With occasional nice results, as in Brause - lhg, ICC blitz, 1997: 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 c6 8.Qe2 d5After 8...b5, Brause plays 9.Nxb5 cxb5 10.Bxf7+ Kxf7 11.Qf3+ Ke8 12.Qxa8, but that strange position, where White has a rook and three pawns against three light pieces, just can't be good. Better seems simply 9.Bb3 9.exd6+ Kd7 10.Ne4 Qe8 11.d5 f5 12.dxc6+ bxc6 (See Diagram) 13.Be6+ Kd8 14.Bg5+ Nf6 15.Nxf6 Qxe6 16.Nd5+ Ke8 17.Nc7+ and Black resigned.

In that 1993 Randspringer article, where Steffen Jakob first found his Halloween Attack, the author (and editor in chief) R. Schlenker had written: 'To wake up ghosts in the sleepy Four Knight's Game is much more than a coarse joke. In fact, we haven't seen such sound joking in a long while.' He mentioned a few games he had played with the Müller - Schulze himself, (we've seen one) but mainly copied his source - an article in the april 1991 issue of the club magazine of a Dutch club - the venerable Discendo Discimus from The Hague, better known as DD - Hein Donner's old club; one of the oldest clubs in Holland. This article would have been buried in the club's archives forever, had not a DD-member sent it to his German friend Schlenker.
    When 4.Nxe5 can be said to have become, through Internet chess, a world-wide fad today, then the author of that club magazine article must be seen as the originator. He was Maurits Wind, whom I vaguely knew as being close to a team of Blackjack professionals I had once written a story about, and as a strong player: his FIDE-rating these days is 2224.
    In that ground-breaking article, Wind called Keres' remark that Black is winning after 7...d5 in Cordel's old variation, "nonsense". "White has excellent compensation," Wind wrote, and after 8.Bxd5 c6 9.Bb3 Bb4 10.0-0 "with nice play for White", he gave two variations: a) 10...Bxc3 11.bxc3 N8e7 12.f4 Be6 13.g4! and b) 10...N8e7 11.Ne4 Bf5 12.Ng5 0-0 13.g4! Both later turned up in Brause's games, suggesting Wind was right: it always won. (Jakob and Brause later discovered that 9...Be6, instead of Bb4, is much better.)
    Wind also gave a serious game he had played in the second league of the Dutch national team competition:Wind - Van der Kraan, tt Nederland 1991: 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 d6 8.Qf3 Qd7 9.O-O c6 10.exd6 Bxd6 11.Re1+ Kf8 11...Kd8 is often played here. 12.Ne4 Bb4 Another possibility is 12...f6; in Brause - sekere, ICC 5 0 blitz, 1998, there followed 13.Bxg8 Rxg8 14.Nxf6 Qf5 15.Nxh7+ Kf7 16.Qb3+ Be6 17.Rxe6 Qd5 18.Ng5+ Kf8 19.Qxb7 and Black resigned. 13.c3 Ba5 14.b3 f6 15.Ba3+ (See Diagram) 15...N6e7? Better N8e7 16.Nd6! (16.Nxf6? Qf5!) and the game goes on. Now, White wins: 16.Qh5 g6 17.Qxa5 Kg7? 18.Nxf6 and Black resigned.
    An interesting variation which Wind analysed happens when, after 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 d6 8.Qf3 Qd7 9.O-O, Black deviates with 9...dxe5. Six years later, his analysis was played move for move at ICC in Brause - evbad, ICC blitz, 1997: 10.dxe5 Nxe5 11.Re1 Bd6 12.Bf4 f6 13.Bb5 c6 14.Rad1 (See Diagram)
    Here, Wind wrote: "A complicated position. Black is a full piece up, White has a threatening attack. Chances seem about even." Brause scored 92 % here, but I'm aware all those percentages of Brause's may be mainly caused by its being a computer. Anyway, evbad lost quickly after 14...a6 15.Bxe5 fxe5 16.Ne4 and she resigned. Better are, according to Wind, 14...cxb5 15.Bxe5 fxe5 16.Ne4 Bc7 17.Rxd7 Bxd7 and 14...Qe6 15.Qg3 Kf8 16.Bxe5 Bxe5 17.f4 "with attack".In Brause - SCHACH1, ICC 2 12 blitz, 1997, there followed, in that last line: 17...cxb5 18.fxe5 Qb6+ 19.Re3 Typical computer move. 19...Bf5 20.Rd6 Qc5 21.Rd5 Brause had this position 46 times, scoring 45½. 21...Qc8 22.Qf4 Ne7 (See Diagram) 23.exf6 Nxd5 24.Nxd5 Qd7 25.fxg7+ and Black resigned.

When I asked Wind how he got the idea to play 4.Nxe5, he wrote: "The first time I read something about this gambit was in the Encyclopedia, the first edition of 1974. The idea to sacrifice a knight in a symmetrical position at move 4 fascinated me. And I found the 'refutation' in ECO, 4...Nxe5 5.d4 Ng6 6.e5 Ng8 7.Bc4 d5! 8.Bxd5 c6 -+, far from convincing. At the time, I looked into it with Michiel (his brother, also a strong player - TK.) Around 1987 - 1991, I often - and succesfully - played it in blitz games at DD. So, when in a team game against Van der Kraan I got the Four Knight's Game on the board, I thought: why not? When I won that game quickly, this was an incentive to really get into it. I analysed a lot, a.o. with the first reasonable computer (Mephisto Lyon) and with Edgar Blokhuis. Hence my DD-story. I still like to play it, especially in Internet blitz games. Against weaker players it is deadly!"
    "If you analyze the gambit, then on the one hand, you see it is dubious (objectively, after 4.Nxe5, White is probably lost), but on the other hand, White has real compensation. It is not just that White has a few primitive threats, and that if Black wards them off, White can resign. No, White really does have long term compensation, even if in the end, it will turn out to be insufficient. In that respect, it is comparable to the Cochrane gambit."
    "In blitz," Wind concludes, "it is quite a good system. In serious games I think it can be played too, but that's of course a gamble: a strong player, somebody over 2200, might be able to refute the gambit over the board. I wouldn't play it against somebody who'd had an opportunity to prepare against it, although even in a correspondence game, where you have unlimited time, White can conduct his attack in the nastiest possible way, and then it is not so sure whether Black will be able to extricate himself and get the full point."

Blokhuis - that Blokje who, before Brause even existed, had introduced me so painfully to the Halloween Attack had to have been Edgar or Jeroen Blokhuis, members of Wind's club DD - and indeed, as Wind revealed to me now, it had been Jeroen, 16 then, close to 2300 and DD's first board now.
    So I had been ignorant to have been angry - five years later, I know that 4.Nxe5 is not a student's joke, but a real gambit, an interesting adventure, an opening in the modern spirit, where activity is more important than material. At the core, it is probably unsound (both Jakob and Wind think so), but the discovery that even something like this is playable at a fairly high level enriches chess.
    Being based mostly on Internet blitz games, this article is clearly not intended as a serious theoretical treatise - had it been, I would not only have given White wins, and I would have shown the various ways in which Black can give back the piece and reach an even game. The point is that these wins happen, and frequently. Brause has several strong human followers at ICC now. Against strong opponents, WGM evbad scores 2 out of 3 with the Halloween Attack. As a blitz gambit, it is a very dangerous weapon; any Black under GM-level who thinks it is plain nonsense, will be wiped off the board with it by a strong White who knows what he is doing. The tactical potential is enormous.
    I might even try it myself.

With special thanks to Steffen Jakob and Maurits Wind.

© Tim Krabbé, 2000




Top | Main Chess Page