300. 6 November 2005: Rook handling

A highly original game from the Dutch Team Competition (the Meesterklasse), played yesterday. Look at White's Rook handling.

Bosboom (Homburg Apeldoorn) - Kuijf (U Boat Worx), Breda, 5 November 2005

1.d4 d5 2.c4 e6 3.Nf3 c6 4.Qc2 Nf6 5.Bg5 h6 6.Bxf6 Qxf6 7.a3 dxc4 8.Nbd2 b5
(see diagram left)
9.g4 Bb7 10.Rg1 c5 11.g5 hxg5 12.Rxg5 Nd7 13.a4 cxd4 14.axb5 c3 15.bxc3 dxc3 16.Ne4 Qf4
(see diagram right)
17.Ra4 Rc8 18.Nc5 Qc7 19.Nxb7 Qxb7 20.Bg2 Qc7 21.Ra6 Nb6 22.Ne5 Be7 23.Bc6+ Kf8 24.Nxf7 Bxg5 25.Nxg5 Qxh2 26.Nxe6+ Ke7 (Better 26...Kg8) 27.Rxa7+ Kxe6 27...Kf6 28.Qxc3+ is also winning for White. 28.Qg6+ Ke5 29.Re7+ Kd4
(see diagram)
30.Qd3+ 30.Re4+ Kc5 31.Qg5+ would have mated quickly. 30...Kc5 31.Qxc3+ throwing away the next half point; 31.Qf5+ Kb4 32.Qb1+ Kc5 33.Qf5+ etc. would still have been a draw. 31...Nc4 32.Re4 and White resigned. (The final moves as I first gave them, 32.Rd7 Qh1+ (0-1) were wrong, according to Kuijf.)

Thanks to Johan Hut for tipping me off.

299. 22 October 2005: The legend of the Curaçao conspiracy (+ PS #298)

Even the paranoid are sometimes persecuted. The chessworld has always agreed that in the Candidates Tournament of Curaçao 1962, there was a Soviet conspiracy against Bobby Fischer.
    Among eight participants, there were five Soviets: Tal, Korchnoi, Geller, Keres and Petrosyan. And in those days you didn't have to be paranoid to think that their first objective might have been to make sure that Fischer would not win.
    Tal, who was in bad form and bad health, and who dropped out after 21 rounds, has never been accused of being part of a Curaçao conspiracy, and Korchnoi has always vehemently denied any such thing. But Geller, Keres and Petrosyan only played fightless draws against each other, which helped them to eight extra resting days each.
    The Soviet Union does not exist anymore, the sins can be admitted. "Of course it was rigged," Yuri Averbakh has recently said (in an interview with Jules Welling in the Dutch magazine Schaaknieuws). According to Averbakh, who was in Curaçao as a member of the Soviet delegation, it had been decided that Keres, being an Estonian, should not win, and neither should the Jewish Ukrainian Geller. It had to the Armenian Petrosyan. Why an Estonian and a Jewish Ukrainian (and the Jewish Russian Korchnoi!) were not suitable and an Armenian was, Averbakh sadly fails to say, or Welling did not ask.
    Well, a conspiracy - but how poorly those evil Soviets managed their conspiracy! With two rounds to go, the disobedient Keres was in first place, and he had to play the outsider Benkö. Had he won, no conspiracy could have prevented an Estonian becoming Botvinnik's challenger - but he lost. Yes, that was a relief, Averbach now says, Benkö's win was a Deus ex Machina which prevented suspicions.
    In his famous piece in Sports Illustrated in 1962, titled How the Russians fixed World Chess, Fischer has given this position as proof of the conspiracy.

White to play
Keres - Petrosyan, Curaçao 1962

After 14...a5 a draw was agreed. But Black is winning, said Fischer, and Timman, in his new book Curaçao 1962, agrees. After 15.Qa3 (15.Qb3 a4 and a3) h6 16.Bf4 Nc4 17.Qb3 Rfc8 Timman gives two main lines: 18.Rd1 a4 19.Qb4 Qe6 20.O-O Nxb2 21.Qxb2 Bxc3 22.Qxb7 Qxa2 and the a4-pawn runs; and 18.O-O a4 19.Qb4 Nxb2 20.Nd5 Nd3 21.Qxe7 Rd8 22.Be3 Bxa1 23.Rxa1 Qxe7 24.Nxe7+ Kh7, which is also winning for Black.

To me, this draw proves that there was no conspiracy at all. This game was played in round 25. Before that round, with four rounds to go (and, as Tal had dropped out, 3 or 4 games to play), the standings of the top-4 were:
Keres 15½ (out of 23)
Petrosyan 15½ (23)
Geller 15 (24)
Fischer 12½ (24)

So what could those bad Soviets have been conspiring about? Fischer didn't even have a theoretical chance anymore of winning the tournament. And if Petrosyan was the designated winner, and Keres should not win, then what could have been easier than to make Keres lose this lost position, and give the point to Petrosyan?
    Fischer started the tournament with 0 out of 2. He fought back, and at midway, after 14 rounds, the standings were:
Petrosyan 9
Geller 9
Keres 8½
Korchnoi 8
Fischer 7

The inevitable insight here is that Fischer only had a chance of still winning the tournament, if the three Soviet leaders kept playing draws against each other. And they did! But Fischer's form was not good enough to take advantage of this help.
    There is no reason to think that at Curaçao, there was anything going on beside Geller, Keres and Petrosyan having great respect for each other; not minding a few extra resting days, and Petrosyan being the luckiest - or perhaps the strongest. Fischer was never an issue.
    Of course it is unfair to have a qualification tournament with five players out of eight from one country - too much chance of foul play. Fischer's accusation, even if it was based on nonsense, led to Curaçao 1962 being the last Candidates Tournament. It was replaced by a series of matches.

PS 4 November: Thomas Henrich makes a good point: "If the Soviets wanted to cheat, why didn't they cheat at their zonal, their own championship? It would have been much easier. If they didn't want Geller or Korchnoi to win, why did they let them go there? There were Smyslov, Spasski (who had crushed Fischer at Mar del Plata 1960), Taimanov and others around. Why the hell should they send the unexperienced Korchnoi, and leave Smyslov at home?"

298. 17 October 2005: New endgame record - a 290 move win

The record of the longest win in an EGTB (database-) endgame has been broken. 15 years after Lewis Stiller discovered the famous 243 move win in KRNknn (Rook + Knight vs. 2 Knights), Marc Bourzutschky and Yakov Konoval found a 290 move win in the 7-man endgame KRRNkrr, or 2 Rooks plus Knight vs. 2 Rooks. "Since only a small number of 7-man endgames has been explored so far," Bourzutschky writes, "there is a good chance even deeper wins will be found" - the length of these wins being measured as the shortest number of moves to mate or to conversion to a won sub-endgame.

According to Bourzutschky, KRRNkrr "appears to be a draw in general," just as KRRBKRR. The longest win in that endgame is 138 moves. They can both be played over in the Palview board on the left.

Asked for any likely candidates to break the 300-move barier, Bourzutschky says: "No clue. I did not expect krrnkrr to be so deep. Perhaps krnnkrn will be another deep one, at least that is one of the next on the list to generate. I also expect krbnkrn to be a general win with very deep lines."

The moves below are amazing. For the first 260 or so, you see no progress at all - and then Black is busted. It's like walking a treadmill in the gym. All you ever see is the walls of the place, and suddenly you're in Kathmandu.

Win in 290 moves
Bourzutschky & Konoval, 2005

Only-winning-moves have an exclamation mark. There are some utterly mysterious ones, like 98.Kg8
1.Rg1+! Kf2 2.R3g2+ Ke3 3.Re2+! Kd4 4.Rd2+! Kc5 5.Rg5+! Kc4 6.Rg4+! Kc3 7.Rg3+! Kc4 8.Ng5 Rg8 9.Re2! Rf4 10.Re5! Rgf8 11.Kd2! Rd4+ 12.Ke1! Rb8 13.Ra3! Rg4 14.Nh3! Rb3 15.Ra4+ Rb4 16.Ra8 Rb2 17.Kf1 Rg3 18.Re4+! Kd5 19.Rh4! Rf3+ 20.Kg1! Kc6 21.Re8! Rg3+ 22.Kh1! Rf3 23.Re1 Rf6 24.Rd1 Rc2 25.Ng5! Rcf2 26.Rd3 R6f4 27.Rh6+! Kb5 28.Nh3! Rf6 29.Rh5+! R6f5 30.Rh4 Rb2 31.Rd1! Ka5 32.Ra1+! Kb5 33.Rh8 Kc6 34.Re1 Rd2 35.Rc1+! Kb7 36.Rb1+ Kc7 37.Rh7+ Kc6 38.Rh6+! Kc7 39.Rg1 Rdd5 40.Rg7+! Rd7 41.Rg2 Rdd5 42.Kg1 Rh5 43.Rg7+! Kc8 44.Rc6+! Kb8 45.Re6 Rde5 46.Rb6+ Ka8 47.Rg3 Rb5 48.Ra3+! Ra5 49.Rf3 Rhf5 50.Rfb3 Rf7 51.Rb8+! Ka7 52.Rg8! Ra2 53.Re3! Rc2 54.Ra3+! Kb7 55.Rb3+! Ka7 56.Rg4 Ra2 57.Rb1 Rb7 58.Re1 Rf7 59.Rd1 Kb7 60.Rg5 Rf3 61.Rg7+! Kc6 62.Rg6+! Kb7 63.Rd7+ Kc8 64.Rh7! Rf8 65.Rc6+! Kb8 66.Rc1! Ra7 67.Rb1+! Ka8 68.Rh5! Rg7+ 69.Kh2 Rf6 70.Ng5! Rgg6 71.Ne4! Rb6 72.Ra5+! Ra6 73.Rf5 Rg8 74.Ng5 Ra2+ 75.Kg3 Ra3+ 76.Kg4 Ra4+ 77.Kh5 Rh8+ 78.Kg6! Ra6+ 79.Kg7! Rb8 80.Rd1! Ra7+ 81.Kg6 Rg8+ 82.Kh5 Ra4 83.Nf3 Ra6 84.Rb5 Ra4 85.Rdb1 Ra7 86.Ng5 Rb7 87.Ra1+ Ra7 88.Rd1 Rb7 89.Rf5 Rb4 90.Nf3! Ra4 91.Kh6 Ra6+ 92.Kh7 Re8 93.Ne5 Ra5 94.Re1! Ra6 95.Kg7 Re7+ 96.Kf8! Rc7 97.Rd1! Ra5 98.Kg8! Rac5 99.Rd2 Rc2 100.Rd4 R2c5 101.Rdf4 Kb7 102.Rb4+! Ka8 103.Kh8 Re7 104.Rf8+! Ka7 105.Ra4+! Kb6 106.Ng6! Rb7 107.Nf4 Rc6 108.Ra3 Rd7 109.Rfa8! Kc7 110.R3a7+ Kd6 111.Ra5 Rf7 112.Ra4! Rcc7 113.R8a6+! Ke5 114.R6a5+ Kd6 115.Rd4+ Kc6 116.Ne6! Kb6 117.Ra8 Ra7 118.Rd6+ Ka5 119.Rd5+ Kb4 120.Rad8 Rfe7 121.Ng5 Re2 122.Rd4+ Kb3 123.Rg4 Rae7 124.Rdd4 Ra2 125.Rb4+ Kc2 126.Rgc4+ Kd3 127.Rc8 Rea7 128.Rb3+ Ke2 129.Rc1 Re7 130.Rb8 Ra5 131.Rg1 Raa7 132.Rh1 Kd3 133.Rh6 Kc4 134.Rf8 Rac7 135.Rh3 Kc5 136.Rhf3 Rc6 137.Rb8 Rg6 138.Nf7 Kc4 139.Rd8 Kb4 140.Kh7 Rg1 141.Rd4+ Kc5 142.Rh4 Kc6 143.Rf6+ Kd7 144.Rd4+ Kc7 145.Kh6 Rh1+ 146.Kg7 Rhe1 147.Rff4 R1e6 148.Rb4 Kc6 149.Rfc4+ Kd5 150.Rc8 Re8 151.Rc1 R8e7 152.Rb8 Re8 153.Rb2 R8e7 154.Rd2+ Ke4 155.Rd8 Kf4 156.Kf8 Re3 157.Rc4+ R7e4 158.Rc5 Re6 159.Ra5 R3e4 160.Rda8 Re2 161.Rc5 Re1 162.Rd5 Rf1 163.Raa5 Ke3 164.Ra7 Ke4 165.Rg5 Kf4 166.Kg7 Rfe1 167.Raa5 Re7 168.Raf5+ Ke3 169.Rf6 Kd4 170.Rd6+ Kc4 171.Rc6+ Kd4 172.Kf6 Re8 173.Nh6 Rf1+ 174.Kg6 Ke4 175.Rd6 Ra1 176.Rc5 Rea8 177.Rc4+ Kf3 178.Nf7 R1a6 179.Ne5+! Kg2 180.Rc2+ Kh1 181.Nc6 Ra1 182.Rd4 R8a2 183.Rh4+ Kg1 184.Rc3 Ra3 185.Rc5 Rg3+ 186.Kf5 Rf1+ 187.Ke4 Re1+ 188.Kf4 Rg8 189.Ne5 Rf8+ 190.Kg5! Ra1 191.Rc3 Raa8 192.Rb4 Ra5 193.Rc1+ Kg2 194.Rc2+ Rf2 195.Rg4+! Kf1 196.Rc1+! Ke2 197.Re4+ Kd2 198.Rb1! Rf8 199.Rb6 Rc5 200.Kg6 Rcc8 201.Rb2+ Kc3 202.Ree2 Rce8 203.Rbc2+ Kb3 204.Rcd2 Rc8 205.Rb2+ Ka3 206.Ra2+ Kb3 207.Reb2+ Kc3 208.Rd2 Kb4 209.Rab2+ Kc3 210.Nd7 Rh8 211.Nf6 Kc4 212.Rb1 Ra8 213.Kg5 Ra5+ 214.Kg4 Raa8 215.Nd5 Rag8+ 216.Kf5 Rf8+ 217.Nf6 Rc8 218.Rdb2 Ra8 219.Rb7 Ra5+ 220.Kg6 Ra6 221.Rc1+ Kd3 222.Rd7+ Ke3 223.Rc3+ Ke2 224.Rc2+ Ke1 225.Kg5 Raa8 226.Rd3 Rhc8 227.Rh2 Rh8 228.Rhd2 Ra3 229.Rd1+ Ke2 230.R3d2+ Kf3 231.Rf1+ Ke3 232.Rdd1 Rha8 233.Nd5+ Ke2 234.Nf4+ Ke3 235.Rde1+ Kd4 236.Ne6+ Kc4 237.Rf4+ Kc3 238.Rf2 Kc4 239.Kf6 Rd3 240.Rc2+ Kb3 241.Rh2 Kc4 242.Kf5 Rd5+ 243.Kf4 Rd3 244.Rc2+ Kb3 245.Rc7 Rd2 246.Nc5+ Ka2 247.Rh7 Rc2 248.Ne4 Rb8 249.Rhh1 Ka3 250.Ra1+ Ra2 251.Rac1 Rg2 252.Rc6 Rgg8 253.Rh3+ Kb2 254.Rh2+ Ka3 255.Ke5 Rb5+ 256.Kf6 Rf8+ 257.Ke7 Rff5 258.Nd6 Rbe5+ 259.Kd7 Rh5 260.Ra6+ Kb4 261.Rb2+ Kc3 262.Rab6 Rh7+ 263.Kc6 Re1 264.R6b3+ Kd4 265.Rf3 Rg7 266.Rd2+ Ke5 267.Nc4+ Ke4 268.Rf6 Rc1 269.Re2+ Kd4 270.Rf4+! Kc3 271.Re3+ Kb4 272.Kd5 Rd1+ 273.Ke6 Kc5 274.Re5+ Kc6 275.Na5+ Kc7 276.Rc4+ Kb6 277.Rb4+ Ka6 278.Nb3 Rg6+ 279.Kf7 Rb6 280.Ra4+ Kb7 281.Na5+ Kc7 282.Rc4+! Kb8 283.Re8+ Ka7 284.Nc6+ Ka6 285.Ra4+ Kb7 286.Na5+ Kc7 287.Rc4+! Kd7 288.Re7+ Kd6 289.Re6+ Kd7 290.Rxb6 and it's mate in 6 more moves: 290...Rf1+ 291.Rf6 Rh1 292.Rd4+ Kc7 293.Rc6+ Kb8 294.Rb4+ Ka7 295.Rb7+ Ka8 296.Rc8 mate.

PS 23 October: Marc Bourzutschky and Yakov Konoval have made available a zipped pgn.file with the maximum wins in all fifty-two 7-man endgames they have explored so far. (48, if the distinction between same- and opposite colored Bishops is ignored.)
    As already said in item 294 - although play is always shown all the way to mate, mate is not what the moves necessarily aim for. The program may choose a conversion to a won sub-endgame over a mate, which may lead to "bizarre moves [that] are fully correct from a game theoretic point of view," as Bourzutschky says.

297. 14 October 2005: An astonishing computer move from San Luis

Rolf Knobel sent me a very nice computer move he found in one of the San Luis games.

Anand - Kasimdzhanov, San Luis round 11
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.f3 e5 7.Nb3 Be6 8.Be3 Nbd7 9.Qd2 b5 10.O-O-O Nb6 11.Qf2 Nc4 12.Bxc4 bxc4 13.Na5 Rc8 14.Bb6 Qd7 15.g3 g6 16.Rd2 Bh6 17.f4 Ng4 18.Qf3 Rb8 19.h3 (see diagram)
Now, after 19...Nf6 Black was lost: 20.Bc5 exf4 21.gxf4 Rc8 22.Bxd6 Qd8 23.Bb4 Qb6 24.a3 Nh5 25.Kb1 Bxf4 26.Nd5 Bxd5 27.Rxd5 Bb8 28.Rhd1 c3 29.Rd7 and he resigned. No better was 19...Rxb6 20.hxg4 and Black loses a piece; Bg7 21.f5 or 20...Bxg4 21.Qe3

But in the diagram, as Knobel's computer analysis shows, Black could have saved himself with the amazing 19...Nh2! After 20.Rd(h)xh2 Rxb6, Black has nothing to fear, and after 20.Qe3, the Knight hops back to safety: 20...Nf3! 21.Rf2 exf4 22.gxf4 Ne5
    On 20.Qf2 too, Black has 20...Nf3 and after a few more computer moves, he seems to survive: 21.Rd5 (21.Nd5 Qa4 22.Nc7+ Kf8 23.Rxd6 Qxa2 24.Nxe6+ fxe6 25.Qxf3 Rxb6!) 21...exf4 22.Qxf3 Rxb6 23.Nxc4 fxg3+ 24.Kb1 Rc6 25.Ne5 Rxc3 26.Qxc3 (26.Qf6 Qe7! 27.Qxh8+ Qf8) 26...Bg7 27.Nxd7 Bxc3 28.Rxd6 Bb4 29.Rb6 Kxd7 30.Rxb4 f5 and here the computer rewards itself for all of its brilliant moves by thinking that Black is equal or nearly equal.

For more examples of common-sense defying computer moves, see next item.

PS: Never underestimate humans! In his press-conference after this game, (click Round 11, Anand), Anand mentions having seen Nh2 - and Nf3 too! Thanks to Rob Eisler for letting me know.

296. 4 October 2005: Astonishing computer moves (Computers can't play chess ct'd)

In July, Shredder, the strongest commercially available chess program, won a master tournament in Argentina with 8½ out of 10. But it also produced a blunder of a type not seen before, even among computers.

White to play
Lafuente - Shredder
Mercosur Cup, Vicente Lopez, Argentina 2005

Here White innocently exchanged Bishops with 19.Bxb7 To everybody's disbelief, Black did not recapture, and lost after 19...Rfd8 20.Bc6 etc.

How is something like that possible? Shredder's creator, Stefan Meyer-Kahlen, simply did not understand. It never does things like that, he mailed me. It was thought something might have been wrong with Shredder's hash tables, where it keeps track of earlier evaluations. Gian-Carlo Pascutto (programmer of Deep Sjeng) explained to me that hash collisions sometimes occur, when the computer retrieves the wrong position from its tables. But: "The chance that this was caused by a hash collision, must be smaller than the chance that both of us will win the lotto every week next year, and will be exterminated subsequently by a comet destroying the Earth". He thought the mistake might have been caused by a hardware problem.

On the other hand, computers are able to find brilliant moves humans could hardly have thought of. Here's a little showcase.

White to play
Fazekas - N. Littlewood, Bath 1963 (analysis)

Despite his huge material advantage, White seems to be dead lost (1.Bxh4 Qxh4 2.Qg1 Qh3+ 3.Kf2 Qe3+ 4.Ke1 Rxg1 mate), but the computer move 1.Bg3 saves a draw. Black must play 1...Rxg3+ 2.Kf2 Rh3+ 3.Kg2 Rh2+ 4.Kxh2 Bg3+ 5.Kg2 Qh2+ 6.Kf3 Qh5+ etc., repeating moves, as 1... Qh5+ 2.Kg2 Rxg3+ 3.Kf1 loses.

Black to play
Shirov - Nikolenko, URS ch, Moscow 1991 (analysis)

In Fire on Board, Shirov writes that he first believed this position to better for White, e.g. 1...Qxa1 2.Dxg7+ Ke6 3.Qe5+ Kd7 4.Nf6+ Kc8 5.Bf4 winning or 1...Bxe4 2.Qxg7+ Ke6 3.Rxf1 Rg8 4.Qe5+ Kd7 5.Rc1 Nc6 6.Qf6
    But then his computer (Fritz4 at the time) found 1...Qf4!! and all White has is 2.Bxf4 Bxe4 3.Be5 Rg8 with equality.

Szilagyi - Benkö, ch Hungary, Budapest 1950
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.d4 exd4 5.Nd5 Nxe4 6.Qe2 f5 7.Ng5 d3 8.cxd3 Nd4 9.Nxe4 A funny trick, but better is Qh5+. 9...Nxe2 10.Bg5 (see diagram)
After the simple 10...fxe4 11.Bxd8 Kxd8 12.Bxe2 exd3 13.Bxd3 c6 14.Nc3 d5 Black was a pawn up, and had no problems. But when I had a look with the computer because this had popped up in a game of mine, it immediately suggested 10...Nf4! which leaves Black a piece up. This was not only missed by Benkö (and by me, of course), but also by the 1974 ECO, which gave 10...Be7 11.Bxe7 Qxe7 12.Nxe7 fxe4 13.Nxc8 exd3 14.Bxe2 dxe2 15.Nd6+ with equality.

PS 18 October: Antonio Torrecillas tells me 10...Nf4! was played in a correspondence game Imperiali - Lumachi, 1994. After just a few more moves, 11.Bxf4 fxe4 12.Nxc7+ Kf7 13.Nxa8 Bb4+ 14.Bd2 Bxd2+ 15.Kxd2 Qa5+ 16.Kd1 d5, White resigned.

I invite readers to send me other examples of this kind of briliant computer move, from their own analysis, or from publications that they have seen.

295. 1 October 2005: Triple play

Douglas Bellizzi of New York's Marshall Chess Club sent me a very nice variation of the well-known castling trap that happened at his club.

Kopiecki - Bonin, New York, Marshall Chess Club, 25 September 2005
1.e4 c6 2.d4 d5 3.Nc3 g6 4.Be3 Nh6 5.f3 f6 6.Qd2 Nf7 7.h4 e6 8.Nh3 Bb4 9.Nf2 Qa5 10.Bd3 e5 11.O-O Nd7 12.a3 Bxc3 13.bxc3 Nb6 14.Ng4 Bxg4 15.fxg4 dxe4 16.Bxe4 Nd6 17.Qd3 Nxe4 18.Qxe4 Qxc3 19.dxe5 Qxe5 20.Qxe5+ fxe5 21.Rad1 Nd5 22.Bg5 h6 23.c4 Nc3 24.Bf6 Nxd1 25.Bxh8 Ne3 26.Rb1 Nxg4 27.Rxb7 A blunder in a losing position - (see diagram)
and now 27...O-O-O was a triple play; attacking the Rook, the Bishop, and threatening mate. White resigned. "Has there ever been a castling move that contained as many threats?" Bellizzi writes, and I do believe there has not.

Although this trick is, for obvious reasons, always seen on the Queen's side, it could happen with King's side castling, and there is one game in which it almost did.

Kupreichik - Kapengut, Minsk 1978
1.e4 e6 2.Nf3 d5 3.e5 c5 4.c3 Bd7 5.Na3 Nc6 6.Nc2 Rc8 7.d4 cxd4 8.Ncxd4 Nge7 9.Bd3 Nxd4 10.Nxd4 Ng6 11.Qe2 Bc5 12.Nb3 Bb6 13.h4 Nxh4 14.Qg4 Ng6 15.Bxg6 fxg6 16.Rxh7 (see diagram)
Now 16...Rxh7 17.Qxg6+ would just cost a pawn, and 16...0-0 17.Qxg6 would lead to mate (Qe7 18.Qh5 etc.), but Black had a nice countertrick.
16...Bxf2+ 17.Ke2 And not 17.Kxf2 0-0+ and Black wins.
There followed: 17...Rxh7 18.Qxg6+ Kf8 19.Qxh7 Qh4 20.Qxh4 Bxh4 21.Be3 b6 22.a4 Rc4 23.a5 Re4 24.Rf1+ Ke8 25.Rf4 Bb5+ 26.Kf3 Ba4 27.Rxe4 dxe4+ 28.Kg4 Bxb3 29.Kxh4 bxa5 30.Bxa7 Bd5 31.g3 a4 and a draw was agreed.

A simple example from an endgame study:

White to play and win
V. Prigunov
"64", 1990

1.0-0 Txh4 would be a draw, so: 1.e7+ Kf7 2.e8Q+ Kxe8 3.f7+ Kf8 4.Be7+ Kxe7 5.f8Q+ Kxf8 Whatever Black plays, his King ends up on the f-line. There follows 6.O-O+ and White wins.

294. 6 September 2005: The mindboggling perfection of 7-man endgames (+ PS to 292)

A few months ago (see item 282 of 12 April) I wrote about the first 7-man endgame database, constructed by Marc Bourzutschky, which solved Troitzky's infamous 4 Knights vs. Queen endgame. In the meantime Bourzutschky, in collaboration with Yakov Konoval, has solved several other pawnless 7-man endgames. One of them, 3 Queens vs. 2 Queens is a nice counterpart to item 292 below, with its two recent 5-Queen games.

Win in 76 moves
Bourzutschky & Konoval, 2005

(Exclamation marks are for only-winning moves.)
1.Kd1! Qd4+ 2.Ke2! Qg4+ 3.Kd3! Qd1+ 4.Qd2 Qa3+ 5.Ke4! Qg4+ 6.Ke5! Qag3+ 7.Qff4! Qg5+ 8.Ke4! Qg6+ 9.Kd5! Qb3+ 10.Kd4! Qg7+ 11.Kc5 Qa7+ 12.Kd6! Qaa3+ 13.Ke5 Qc5+ 14.Kf6! Qbb6+ 15.Qe6! Qf8+ 16.Ke5! Qg7+ 17.Qef6 Qg5+ 18.Ke4! Qb1+ 19.Kd4! Qa1+ 20.Kc4 Qa4+ 21.Kd3! Qb3+ 22.Qfc3 Qb1+ 23.Kd4! Qb6+ 24.Ke4 Qbg6+ 25.Ke3 Qg1+ 26.Ke2! Q6g2+ 27.Kd3! Qb1+ 28.Kd4! Qb6+ 29.Qc5 Qf2+ 30.Qde3! Qfb2+ 31.Kd5! Qa2+ 32.Qfc4! Qd2+ 33.Qcd3 Qa2+ 34.Kd4 Qf6+ 35.Qee5! Qff2+ 36.Qde3 Qfb2+ 37.Qec3! Qf2+ 38.Qee3 Qf6+ 39.Kd3 Qb1+ 40.Kc4! Qa6+ 41.Kd5 Qf5+ 42.Qee5! Qf7+ 43.Ke4 Qe2+ 44.Kd4! Qef2+ 45.Qee3 Q2f6+ 46.Kd3 Qf1+ 47.Ke4 Qe6+ 48.Q5e5! Qg4+ 49.Q3f4! Qfg2+ 50.Kd4 Qd7+ 51.Qd6! Qf2+ 52.Kc4! Qa2+ 53.Kd3 Qb5+ 54.Kd4! Qa7+ 55.Ke4! Qe2+ 56.Qfe3 Qa4+ 57.Kd5 Qeb5+ 58.Qec5 Qa2+ 59.Kd4! Qf2+ 60.Qe3! Qbb2+ 61.Ke4 Qg2+ 62.Kf5! Qb1+ 63.Kf4 Qbf1+ 64.Ke5! Qg7+ 65.Kd5! Qfg2+ 66.Kc4 Qa2+ 67.Qb3 Qg4+ 68.Qcd4! Qc8+ 69.Kb4 Qb7+ 70.Q6b6 Qe7+ 71.Kc4 Qae2+ 72.Kc3 Qf3+ 73.Kb2 Qee2+ 74.Qc2 Qef2 75.Qh8+ Kg1 76.Qcxf2+ Qxf2+ 77.Qxf2+! Kxf2 78.Qg8 Ke3 79.Qg4 Kf2 80.Kc1 Kf1 81.Kd1 Kf2 82.Kd2 Kf1 83.Ke3 Ke1 84.Qg1 mate. A mindboggling view of perfection.

This the longest quickest win for this endgame, White always aiming for the quickest mate or transposition to a winning sub-endgame, Black for the slowest. Bourzutschky and Konoval did not calculate the DTM (distance to mate), as that would cost, in most cases, at least a factor of 2 in calculation speed. This KQQQkqq endgame takes about 1½ day to generate and verify.

Bourzutschky and Konoval worked with basic hardware worth not more than $ 3000, on which some new programming tricks make it possible to solve the famous 6-man krnknn endgame (243 moves without mate or capture; still the record) in a little over an hour.

By far the largest endgame that Bourzutschky and Konoval solved (and that was ever solved) is KRBNkrb. It took their configuration 17½ days to generate (plus another 5 days to verify), yielding two files (one for same-color Bishops, the other for opposite-color Bishops) totalling 168 Gigabytes. Although that is beside the point, it does have some practical relevance, and it also occurs in endgame studies.

Win in 226 moves
Bourzutschky & Konoval, 2005

1.Rb1+! Kc4 2.Nd2+! Kd3 3.Nf3! Rh6 4.Kd1! Ba5 5.Rb3+! Bc3 6.Ne5+! Ke4 7.Ng4! Rg6 8.Nf2+! Kd4 9.Rb8! Re6 10.Rd8+! Kc4 11.Rc8+! Kd4 12.Ng4 Re1+ 13.Kc2 Re2+ 14.Kb3! Rb2+ 15.Ka3! Rc2 16.Bg1+ Kd3 17.Rd8+! Ke4 18.Nf2+ Kf5 19.Rd3! Rc1 20.Rf3+! Ke6 21.Nh3! Be5 22.Kb3! Rb1+ 23.Kc2 Rb2+ 24.Kd1! Rb1+ 25.Ke2 Rb2+ 26.Kf1 Rb1+ 27.Kg2 Rb2+ 28.Kh1 Rb4 29.Ng5+! Kd5 30.Rd3+! Kc4 31.Rd1! Rb3 32.Re1! Kd5 33.Rf1 Ra3 34.Rf2! Rb3 35.Rf5 Rb4 36.Nf3 Ke4 37.Rf7! Rb2 38.Ng5+! Kd5 39.Ra7 Bd6 40.Ra5+! Kc4 41.Ne4! Bb8 42.Ra8 Bc7 43.Nf6 Kb5 44.Rg8! Kc6 45.Rg6! Kd6 46.Ng4+! Kd5 47.Ne3+! Ke4 48.Re6+! Kf3 49.Nd5 Ba5 50.Re3+! Kg4 51.Re4+! Kg3 52.Re8 Kf3 53.Rf8+ Kg3 54.Rf1 Bb4 55.Rf7! Bd6 56.Rd7 Bb8 57.Rd8 Be5 58.Re8! Bd6 59.Re6 Bb8 60.Re3+ Kg4 61.Ra3 Kf5 62.Ra4! Rd2 63.Ne3+! Ke6 64.Nc4 Rc2 65.Ra6+ Kd7 66.Ne3! Rc6 67.Ra4! Bc7 68.Ng4! Rc2 69.Re4 Bd6 70.Nf6+! Kc7 71.Re6! Bc5 72.Bh2+! Kb7 73.Ne4 Bb6 74.Be5 Bc5 75.Rh6 Bb6 76.Nc3 Ba5 77.Nd5 Rc5 78.Rh7+! Ka6 79.Rd7! Kb5 80.Bg3! Kc6 81.Nf6! Rg5 82.Rd6+! Kb5 83.Rd3! Rg6 84.Ne4 Kc6 85.Rd1! Re6 86.Ng5 Rg6 87.Bf4 Rf6 88.Nh3 Rg6 89.Rc1+ Kd7 90.Nf2 Ke6 91.Rb1! Kf5 92.Bh2! Rb6 93.Rc1! Rb5 94.Rc8! Bb6 95.Rf8+! Ke6 96.Nd3! Rb3 97.Nf4+! Ke7 98.Rh8 Rb2 99.Nd3! Rd2 100.Ne5! Bc5 101.Bf4 Re2 102.Nd3 Be3 103.Bg3 Rd2 104.Bh4+ Kd6 105.Be1! Rc2 106.Rh7 Ra2 107.Bg3+ Ke6 108.Rh5! Bd4 109.Be1 Ba7 110.Rg5 Bd4 111.Nf4+ Kf7 112.Rf5+ Ke8 113.Nd3 Ra3 114.Rf3! Ra2 115.Rf4 Ba7 116.Nc1 Rb2 117.Rf5 Kd8 118.Nd3 Ra2 119.Bg3 Kd7 120.Rh5 Be3 121.Rd5+ Kc8 122.Rf5 Rd2 123.Rf8+ Kb7 124.Rd8 Ka7 125.Be1 Rc2 126.Re8 Bb6 127.Nb4 Rc4 128.Re6 Kb7 129.Nd5! Ba5 130.Bg3 Rc5 131.Re7+ Kc6 132.Nf6 Rf5 133.Re6+ Kc5 134.Kg2 Bc3 135.Ne8 Kd5 136.Nc7+! Kc4 137. Re7 Rf6 138.Ne6 Bd2 139.Ng7 Bh6 140.Ne8 Rf8 141.Nd6+ Kd5 142.Re4 Bd2 143.Rg4 Be3 144.Rh4 Kc5 145.Ne4+ Kc6 146.Nc3 Kc5 147.Re4 Bd4 148.Ne2 Bg7 149.Nf4 Rf7 150.Bf2+ Kc6 151.Rc4+ Kb5 152.Rc5+ Ka4 153.Ne6 Re7 154.Nc7 Rd7 155.Bg3 Kb4 156.Rc6 Rd3 157.Bd6+ Kb3 158.Ne6 Bd4 159.Bf4 Be3 160.Be5 Ba7 161.Nf4 Rd2+ 162.Kf3 Bd4 163.Bb8 Bg1 164.Ne2 Rd3+ 165.Kg4 Rd2 166.Re6 Bc5 167.Bf4 Rd7 168.Kf3 Kc4 169.Re4+ Kd5 170.Nc3+ Kc6 171.Re6+ Kb7 172.Ne4 Rd3+ 173.Kg4 Rd5 174.Rh6 Ba3 175.Rh3 Ra5 176.Rc3 Bf8 177.Rc7+ Ka6 178.Be3 Re5 179.Kf3 Bb4 180.Bd4 Re6 181.Rc5 Ba5 182.Rd5 Bb6 183.Ba1 Re8 184.Rd1 Rd8 185.Nc5+! Kb5 186.Nd7 Bc7 187.Rb1+ Kc6 188.Nf6! Bd6 189.Ke4 Bc5 190.Rc1 Kb6 191.Be5 Kc6 192.Nh5 Kb5 193.Ng7 Rd2 194.Ne6 Re2+ 195.Kf5 Bf2 196.Nc7+ Kb4 197.Nd5+ Kb3 198.Rb1+ Kc4 199.Ke6 Rc2 200.Rb4+ Kd3 201.Ra4 Rd2 202.Kf5 Bc5 203.Bg7 Rc2 204.Bf6 Rd2 205.Nf4+ Ke3 206.Rc4 Bd6 207.Rc3+ Kf2 208.Ne6 Re2 209.Ng5 Kg2 210.Bd4 Rd2 211.Rc4 Bg3 212.Ne4 Ra2 213.Rc1 Bb8 214.Kg4 Ba7 215.Nc3 Ra5 216.Rc2+ Kf1 217.Bf6 Bb8 218.Bh4 Ba7 219.Bg3 Ra1 220.Kf3 Bd4 221.Bh4 Be5 222.Be7 Re1 223.Bc5 Ra1 224.Rf2+ Kg1 225.Re2+ Bd4 226.Bxd4+ Kf1 227.Rf2+ Kg1 228.Ra2+ Kh1 229.Rxa1+ Kh2 230.Rh1+ Kxh1 231.Ne4 Kh2 232.Ng3 Kh3 233.Bg1 Kh4 234.Ne4 Kh5 235.Bd4 Kg6 236.Nd6 Kg5 237.Kg3 Kh5 238.Bf6 Kg6 239.Be7 Kh5 240.Bd8 Kg6 241.Kh4 Kh6 242.Be7 Kg6 243.Bg5 Kh7 244.Kh5 Kg7 245.Be7 Kh8 246.Kh6 Kg8 247.Kg6 Kh8 248.Nf7+ Kg8 249.Nh6+ Kh8 250.Bf6 mate. "Particularly striking," writes Bourzutschky, "is the retreat by the white king to h1 on move 28, where he remains for over a hundred moves until finally reemerging on move 134."

This ending (and several other 7-man endings) can be played over on the Palview board on the left. Play is always shown all the way to mate. It should be noted however, that in the final phases, the play for conversion instead of mate sometimes leads to some seemingly illogical moves. See, for instance, move 230 in the above ending. Almost any move is mate in 2, but White sacrifices a Rook for a conversion in 1, even if that is only mate in 21.

PS 7 February 2014: Guy Haworth points to another detour. After 226...Kf1, White can mate in 3 with 227.Rg(h)2 Ke1 228.Be3 and 229.Rg(h)1 mate. Even if that mate is the same distance as the conversion, the conversion is chosen.

All of these endings are the quickest longest wins in their class, and utterly fascinating to watch.

KNNNknn, win in 93 moves
KRBNkrb, win in 192 moves

KRRRkqb, win in 200 moves
KBNNNkq, win in 224 moves

KBBBBkq, win in 101 moves

Other endings solved by Bourzutschky & Konoval are KRRRkqn (longest shortest win 131 moves), KRRkbbb (50), KRRknnn (49), KQQkrrr (40), KBBBknn (36), KRRRkrr (34), KBBBkbb (32), and KNNNkbb (30). They have good hopes that they will find a 7-man ending that breaks the above-mentioned 243-move record.

293. 1 September 2005: 10,000 meters, without oxygen

Exactly two years ago, under the title "A new chapter in the Babson saga", I published the first cyclic Babson Task, by Peter Hoffmann. (See Diary 221-240, item 226.) Although the variations worked as they should, showing the scheme QB-BR-RN-NQ, the problem had an obvious shortcoming - in the initial position, there were four white Bishops and three white Knights. In later versions, Hoffmann managed one Knight less, and eventually with one promoted piece (see Diary item 284 of 7 May), but of course the real challenge was to construct a cyclic Babson without any promoted men. "If the regular Babson is the Everest, then achieving the cyclic Babson must be compared to the ascent of a 10.000 meter mountain, without oxygen," he quoted a problem friend - the oxygen being the use of promoted pieces.

In today's (September) issue of the German magazine Schach, Hoffmann has done it - the first cyclic Babson without promoted men. I am honoured that he dedicates this historical problem to me.

Mate in 4
Peter Hoffmann
dedicated to Tim Krabbé
Schach, September 2005

The matrix is partly different from the 2003 one. There the white pawn promoted on g8 - here, it promotes on f8.
    After 1.Bxc6 (in a cyclic Babson such a key should be forgiven), White has many threats, mainly 2.Qxd2 and 2.gxf8Q. Black's only defense is to promote on d1.

1...d1Q 2.gxf8B! And not 2.gxf8Q? Qd4+! 3.exd4 stalemate, nor 2.gxf8N+ Kd6! and White will be too late. Now, the threat is 3.c8Q+. 2...Qd4+ Or 2... Qd7 3.c8Q Qxc8 4.bxc8Q mate 3.exd4 Kxf6 4.d5 mate.

After 1...d1B, 2.gxf8Q would be stalemate, and 2.gxf8N+ Kd6 also doesn't work; e.g. 3.c8N+ Kc5! or 3.Qd2+ Kxc7! But after 2.gxf8R! Black is in Zugzwang; there follows 2...Kd6 3.Qd2+ and mate next move, e.g. Kc5 4.Qd4 mate.

1... d1R Now 2. gxf8Q? Rd4+! 3.exd4 would be stalemate again, just as 2.gxf8B? Rd7! 3.c8Q But after 2.gxf8N+! there follows 2...Kd6 3.c8N+ Kc5 4.Qxc2 mate (or 3...Kc7 4.Ne6 mate.)

Finally, 1... d1N threatens a nasty check, so White must be quick. 2.gxf8N+ Kd6 is too slow, but 2.gxf8Q! works: 2...Nc3+ (or 2...Nxb2+ 3.Qxb2 and mate next move) 3.Kxa5 Nd5 (otherwise Qe7 mate) 4.c8Q or 4.Bd7 mate.

PS 8 Januari 2006: In Schach for December, Peter Hoffmann showed a new version with, in his words, the best possible key. In German, this is probably the Letztform.

Mate in 4
Peter Hoffmann (dedicated to Tim Krabbé)
SCHACH, December 2005

1.Nxb6 and now

1...d1Q 2.exf8B! (2.exf8Q? Qd4+ 3.Bxd4 stalemate or 2.exf8N+? Kd6 3.Be5+ Kc5 and no mate) 2...Qd4+ 3.exd4 Kxf6 4.d5 mate

1...d1R 2.exf8N+! (2.exf8Q? Rd4+ 3.Bxd4 stalemate but also 2...Rd7! 3.Qe8+ Kd6 and no mate; or 2.exf8B? Rd7! 4.c8Q(B) stalemate) 2...Kd6 3.Be5+ Kc5 4.Qxc2 mate

1...d1B 2.exf8R! (2.exf8Q? stalemate or 2.exf8N+? Kd6 3.Be5+ Kc5 and no mate) 2...Kd6 3.Qd2+ and mate next move

1...d1N 2.exf8Q! (2.exf8N+? Kd6 3.Be5+ Kc5 and no mate) 2...Nxc3+ 3.Kxa5 Ne4 4.c8Q mate

Side variations: 1...Rxh8 2.Qxd2 Kf7 3.Qd5+ Ke8 4.c8Q mate and 1...dxc1Q 2.exf8Q Qa1+ 3.Rxa1 c1Q 4.c8Q mate.

See also: Sons of Babson.

292. 21 August 2005: New five-Queen games

Serious games where five Queens appear on the board simultaneously are extremely rare. In all tournament play, only two such games were known, (see my records page), which makes it something of a miracle that two new examples were added within a few months this year:

Miton - Benjamin, World Open, Philadelphia 2005

1.d4 Nf6 2.Nf3 Nc6 3.c4 e6 4.Nc3 Bb4 5.Qd3 d6 6.a3 Bxc3+ 7.Qxc3 O-O 8.e3 a5 9.Be2 Re8 10.b3 e5 11.d5 Ne7 12.Nd2 Bf5 13.O-O c6 14.dxc6 Nxc6 15.Rd1 Rc8 16.Qb2 Qe7 17.Nb1 b5 18.Nc3 bxc4 19.bxc4 Ne4 20.Nb5 Red8 21.f3 Nc5 22.e4 Be6 23.Be3 Rb8 24.Qc3 a4 25.Rac1 Nb3 26.Rc2 h6 27.f4 f6 28.Qe1 Bf7 29.Qf2 exf4 30.Qxf4 Ne5 31.Nc3 Nc5 32.Nd5 Bxd5 33.cxd5 Rb3 34.Bxc5 dxc5 35.Bc4 Nxc4 36.Rxc4 Rxa3 37.d6 Qe6 38.Rxc5 Rb3 39.Ra5 a3 40.Ra7 Rd7 41.Rxd7 Qxd7 42.e5 Rb2 43.Qc4+ Kh7 44.e6 Qa7+ 45.Qd4 Qb7 46.Qd3+ g6 47.Qf1 a2 48.e7 Rb1 49.e8Q a1Q 50.d7 Qb6+ 51.Kh1 Rxd1 52.Qe7+ Kg8 53.d8Q+ (see diagram)
53...Qxd8 54.Qxd8+ Rxd8 55.Qxa1 and four Queens have disappeared in four half-moves, which may be a record in itself. 55...Rc8 56.h4 Rd8 57.Qa2+ Kg7 58.Kg1 Rd7 59.Kf2 Re7 60.Kf3 Rf7 61.Qa8 Re7 62.Qd8 Re5 63.Qd7+ Kf8 64.Qh7 Re6 65.Qh8+ Kf7 66.Qc8 Ke7 67.Qh8 g5 68.Qh7+ Kd6 69.Qg7 Ke5 70.Qxh6 Rd6 71.Qh7 Re6 72.Qh5 and Black resigned.

Hickl - Sokolov, Ordix Open Mainz, August 2005

1.g3 d5 2.Nf3 c6 3.Bg2 Bg4 4.b3 Nd7 5.Bb2 Bxf3 6.Bxf3 e5 7.d3 Ngf6 8.O-O h5 9.c4 d4 10.Re1 Bc5 11.Na3 Bb4 12.Rf1 Nf8 13.Nc2 Be7 14.e3 Ne6 15.exd4 exd4 16.Re1 Qd7 17.Qd2 h4 18.Rad1 Rd8 19.b4 Kf8 20.Re2 c5 21.bxc5 Bxc5 22.Nb4 hxg3 23.hxg3 Nc7 24.Rde1 a5 25.Nc2 b6 26.Ba3 Ne6 27.Re5 g6 28.Rxe6 fxe6 29.Bxc5+ bxc5 30.Qxa5 Rc8 31.Na3 Kg7 32.Re5 Qe7 33.Nb5 Nd7 34.Re2 Qg5 35.Qe1 e5 36.Nd6 Rb8 37.Bg2 Rb6 38.Nb5 Ra6 39.Rb2 Rb8 40.Bh3 Nf6 41.Re2 e4 42.dxe4 Nh5 43.Bg2 Rab6 44.e5 d3 45.Rd2 Rxb5 46.cxb5 c4 47.e6 Rd8 48.Rd1 d2 49.Qe4 c3 50.b6 Qf5 51.e7 c2 52.Rf1 d1Q 53.exd8Q Qxd8 54.Qb4 Qd1 55.b7 c1Q 56.b8Q (see diagram)
56...Qxf1+ 57.Bxf1 Nf6 58.Q4f8+ and Black resigned.

This game has the additional oddity that the 2-Queen side wins, but that also happened in one of the games on my records page. The record of the other game there, the five Queens remaining on the board for nine half-moves, still stands.

PS 6 September: Hauke Reddmann sent me a game he once played in which 5 Queens appeared:

Schulz - Reddmann, ch Hamburg 1983
1.e4 c5 2.b4 cxb4 3.Bb2 Nf6 4.Qf3 Qc7 5.Bd3 d6 6.Qg3 Nc6 7.Nf3 e5 8.O-O Be7 9.Ne1 O-O 10.a3 Be6 11.axb4 Nxb4 12.Nc3 Nh5 13.Qf3 Nf4 14.Ne2 Nbxd3 15.cxd3 Nxe2+ 16.Qxe2 Bf6 17.Bc3 a5 18.f4 b5 19.g4 h6 20.Ng2 b4 21.Bb2 a4 22.g5 hxg5 23.fxe5 dxe5 24.Rxf6 gxf6 25.Qf2 Qe7 26.Bc1 a3 27.Ne3 Rfc8 28.Kg2 Rxc1 29.Rxc1 b3 30.h4 b2 31.Rh1 g4 32.Qg3 Kf8 33.h5 a2 34.h6 b1Q 35.h7 Kg7 36.Rh4 a1Q 37.h8Q+ (see diagram) Rxh8 38.Rxg4+ Kf8 and White resigned.

And then of course, there is this old Dutch game, which was however probably not played under tournament conditions.

Tresling - Benima, Winschoten 1896
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.Nc3 Be7 6.O-O b5 7.Bb3 d6 8.d3 Be6 9.Qe2 Qd7 10.Nd5 Bxd5 11.exd5 Na5 12.d4 Nxb3 13.axb3 e4 14.Nd2 O-O 15.c4 Rfe8 16.Nxe4 Nxe4 17.Qxe4 Bf6 18.Qd3 Qg4 19.Be3 b4 20.f4 h5 21.h3 Qd7 22.f5 Qe7 23.Rae1 a5 24.Kh1 Qe4 25.Qd1 c5 26.dxc5 Bxb2 27.Rf4 Qe7 28.Qxh5 Bf6 29.cxd6 Qxd6 30.Qf3 Re5 31.Re4 Rxe4 32.Qxe4 Qg3 33.Bf4 Qxb3 34. d6 Rf8 35.c5 a4 36.c6 a3 37.Be5 Bxe5 38.Qxe5 a2 39.d7 Qa3 40.c7 b3 41.d8Q b2 42.c8Q b1Q (see diagram) 43.Qxf8+ Qxf8 44.Qxf8+ Kxf8 45.Qe8 mate

291. 17 August 2005: Endgame study quiz

Harold van der Heijden offers 3 copies of his famous updated Endgame Study Database (see item 285) as Prizes in a quiz he devised. One CD is for the highest score, two further ones are for the next best, provided they score 5 points or more. Answers as per 17 August; 1 point per question; bonus points for answers that correct Harold's intended answer. Partially correct answers score parts of points.
    Solutions before 1 October to hhquiz@gmail.com. One submission per competitor, corrections not allowed. In case of equal points, earlier submissions prevail. Any unclarities will be decided by Harold.

1) Sometimes it takes a long time before the award of a tourney is published. What is the world's record? (Awards that never appeared do not count.)

2) It is very difficult to become a composition grandmaster. At present there are four grandmasters who earned their title (mainly) with endgame studies. Who?

3) Who is the oldest living endgame study composer? (With at least one study in the hhDBIII database.)

4) As far as I know in the whole history of chess only two endgame study composers can be considered as professionals. Who?

5) Some composers have the bad habit of sending studies to more than one tourney. Which actually published study holds the world's record?

6) Name at least four magazines that are (were) fully devoted to endgame studies.


What is the primary source of the endgame study in the diagram. (White to play and win; 1.a5 Bf8 2.Kd5 Bh6 3.g5+! Bxg5 4.Ke4 Bh4 5.Kf3 etc.)

8) Which formal endgame study tourney attracted the highest number of studies? (FIDE Albums submissions not counting.)

9) Name the three most prolific endgame study composers.

10) What is the most elementary position (by material) that is illegal and still occurred (as a pattern) in a study?

PS 3 October: Here are the answers.

1) Shakhmaty v SSSR 1940, which was judged in 1987.

2) Dobrescu (Rumania, 1989), Gurgenidze (Georgia, 1990), Rusinek (Poland, 1992) and Nestorescu (Rumania, 2001). (Living composers only.)

3) Paoli, who was born 13 January 1908.

4) Pogosyants and Afek.

5) A study by the foursome Ionov, Krutchkov, Chitkin and Peretyatko was published as an original in Shakhmatna Misl 1989, Szachy 1989, The Problemist 1991, and Schach 1992.

6) EG; EBUR; British Endgame Study News; STES Journal; Studistica; Etyudnaya Mozaika Malyutka.

7) This famous study by H.Otten was originally published in 1891 in the New York Sunday World, as was recently discovered bij Harrie Grondijs.

8) Roycroft Jubilee Tourney in 1978-9 in EG, with 274 original studies.

9) Pogosyants (1884); Rinck (1685); Troitzky (1541).

10) Ba1/Pb2 (and mirrors), used by Horwitz in 1871, Voellmy in 1922 and G & J. van Altena in 1940.

Due to the extreme difficulty, there were only nine entries, of whom the top two were answers themselves. They are: 1. John Roycroft (8.50), 2. Yochanan Afek (6.90), 3. Emil Vlasák (6.75), 4. Elie Solomon (5.75), 5. Arpad Rusz (5.75), 6. John Beasley (5.08), 7. Peter Treffert (4.75), 8. René Olthof (3.75), 9. Maarten de Zeeuw (3.75).

A more detailed solution with comments will be published by Harold van der Heijden in the December issue of EBUR.

290. 31 July 2005: Disaster in the Queen's ending

Yesterday, the Dutch Open had an occurrence of one of my favorite themes, the Disaster in the Queen's ending - the disaster being that one side is one or more pawns up, but loses in a winning attempt.

White to play
Van Lanen - Van Oosterom
Dieren, Dutch Open, 30 July 2005

After 60.Kh4, White would have good winning chances. But with the horrible 60.Qg3, he had missed 60...f4, after which White resigned.

It reminded me of some other more or less recent exemples.

Black to play
Negre - Todorov
Cannes 1997

Two pawns up, Black gambles on the speed of his passed pawn. 45...h5 46.a4 Threatening to win the Queen with Qd8+ and a5+ 46...a6 47.Qxe6 h4 Stronger was Qd2+ 48.Kb3 Qd1+ 49.Ka3 Qc2 followed by a5. 48.Qe7 h3? Ka7 might still have been winning, and after 49.b5 Black should have given a perpetual, or allowed one with axb5. 49...h2? Perhaps this was an intentional Queen's sacrifice, hoping for a win with the new Queen. 50.Qd8+ Qc7 51.a5+ Kxb5 52.Qxc7 h1Q 53.Qxb7+ Kc4 54.Qxa6 mate. Did he overlook that mate? But 53.Qe7 too, was already winning.

White to play
Korobov - Osokin
Kstovo 1998

The extra passed pawn offers some winning chances. 43.Kxh5 Qh3+ 44.Qh4 Qf5+ enz. is a draw, so: 43.Kh4 Qf2 44.h3 Qc2 After 45.Qxf7 Qxa4+ 46.Kxh5 White would be two pawns up, but with all the pawns on one wing, the win would present great technical difficulties. 45.Qe4+ Kh6! 46.Qf4+ (Qxc2 g5 mate) 46...Kh7 47.Kxh5 Now he does have a second extra pawn, but: 47...Qc5+ After 48.Qg5 Qd4 the a4-pawn will be lost or the moves will be repeated with 49.Qf4 Qc5+. And 48.Kh4 f6 even loses; 49.Qg4 Kh6 of 49.Kg4 Kg6 Therefore: 48.Kg4 But that is a terrible underestimation of the Queen's attacking potential. After 48...Kg6 49.h4 White immediately resigned: Qc8 is mate.

Thanks to Piet Peelen for showing me the Dieren example.

289. 16 July 2005: The Back Home Task

In the Humor Composing Tourney (see items 281, 276 and, in Dutch, Het leuke van humor) we awarded a Special Prize to the following study.

White to play and win
Gady Costeff
Special Prize, Humor Tourney, 2005

1.Rh1 Rh4 2.Ng1 Bf7 3.Ra1 axb4 4.Nb1 Be6 5.Bf1 Nc4 6.Bc1 and wins.

White pins a piece, attacks it, and wins it. That is all. What is so funny about that? The fact that White can only win by retracting all his pieces to their starting squares. Two minor blemishes are that the Queen doesn't join in the fun (but that would hardly be possible in a study) and that Rg1 retracts sideways. As a member of the jury (the others were Jan Timman, Hans Böhm and Harold van der Heijden) I voted for a major prize, but Jan Timman said: 'This is not an endgame study, it is an idea.' That was true. But as it was such a witty idea, and this composition would be pointless outside of this theme tourney, we decided to give it a Special Prize.

Many thanks to Joshua Green for showing me a selfmate with the same theme - which could be named the Back Home Task.

Selfmate in 8
M. Kirtley
1st Prize, The Problemist 1986

1.Nb1+ Kb3 2.Qd1+ Rc2 3.Bc1 axb6 4.Ra1 b5 5.Rh1 bxc4 6.Ke1 c3 7.Ng1 f3 8.Bf1 f2 mate.

Strategy and deep themes are absent, Black only has forced moves, but it's one of the funniest chess problems I ever saw. I was tempted to give a diagram of the final position, but it's more rewarding to play it over on the Palview board on the left.

288. 15 June 2005: Surprises

Two entertaining games, shown to me by readers.

De Jonge - Van der Tuuk, Utrecht Open ch, 10 June 2005

1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.O-O Be7 6.Re1 b5 7.Bb3 d6 8.c3 O-O 9.h3 Nb8 10.d4 Nbd7 11.Nbd2 Bb7 12.Bc2 Re8 13.b3 Bf8 14.Bb2 g6 15.Rc1 Bg7 16.d5 Nh5 17.b4 Nb6 18.c4 bxc4 19.a4 Nf4 20.a5 Nbxd5 Black launches on a - healthy - adventure. 21.exd5 Bxd5 22.Nf1 Be6 23.Ne3 Rb8 24.Qd2 f6 25.Ba4 d5 26.Bxe8 Qxe8 (see diagram)
And the pawn roller is certainly worth a rook.
27.Red1 Nd3 28.Ba3 Bh6 29.Rc2 Qa4 30.Qc3 c5 31.Rxd3 cxb4 32.Bxb4 cxd3 33.Qxd3 Bxe3 34.Qxa6 Qxc2 Bb6 was winning 35.Qxe6+ Kg7 36.Qe7+ (see diagram)
And now Black even manages to lose. 36...Kh6 Kg8 was still a draw. 37.Qf8+ and Black resigned.

Solodovnichenko - Filippov, Bydgoszcz Open, 1999

1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 Nc6 6.Ndb5 d6 7.Bf4 e5 8.Bg5 a6 9.Na3 b5 10.Bxf6 gxf6 11.Nd5 f5 12.exf5 Bxf5 13.c3 Bg7 14.Nc2 Be6 15.Nce3 Ne7 16.Nxe7 Qxe7 17.g3 d5 18.Bg2 Rd8 19.O-O O-O 20.Qe2 f5 21.Rfd1 e4 22.Nc2 f4 23.f3 Be5 24.fxe4 fxg3 25.Qh5 gxh2+ 26.Kh1 Qg7 27.Ne3 dxe4 28.Bxe4 Bf4 29.Rxd8 Rxd8 30.Ng2 Bc7 (see diagram)
31.Re1 A farsighted trick, which Black apparently does not appreciate. 31...Bxa2 32.b3 Bxb3 33.c4 Bxc4 Spoiling the last winning chances. 34.Bd5+ Bxd5 35.Re8+ Rxe8 36.Qxe8+ Qf8 37.Qxf8+ Kxf8 (see diagram) with, for a game, an uncommon stalemate.

Thanks to Johan Hut and Jovan Petronic, respectively.

287. 23 May 2005: Father Knows Best (computers can't play chess, ct'd)

One of my first computer opponents, back in the early 80ties, was a certain Chess Champion MK I. When I accidentally made an illegal move against it, it simply played on, which gave me the idea to start with 1.Rh1xh8 in the next game. Probably seeing its Ng8 was attacked, it played 1...Nf6 - a blunder, as this allowed 2.Ng1xg7 mate. The champion acknowledged the mate; its window said: LOSE. In the next game, I opened with 1.Bc1xe8, hoping to settle the question of the shortest game forever. But Chess Champion MK I did not comply - it played 1...Qd8xe8, somehow leaving me without a plan.

Have they progressed much since then? Look at the 3-minute blitz game I won this morning against Fritz 8 on my 2.4 GHz, 512 Mb RAM computer with hashtable size 192 Mb.

TK - Fritz 8, 3 0 blitz, 23 May 2005
1.d4 Nf6 2.f4 d5 3.c3 Bf5 4.e3 Nbd7 (see diagram)
5.Qd3 TN
5...Bxd3 6.Bxd3 Ne4 7.Nf3 e6 8.O-O f5 9.Re1 Qf6 10.Re2 Be7 11.Bd2 O-O 12.Be1 Qh6 13.Nbd2 c5 14.Nf1 c4 15.Bc2 a5 16.Rc1 a4 17.a3 b5 18.Bb1 Qh5 19.Rcc2 Rab8 20.Rc1 Rbe8 21.Bc2 Qg4 22.Bd1 Rf7 23.Rcc2 Qh5 24.Bd2 Rf6 25.Be1 Rh6 26.Rc1 Qg4 27.Rcc2 Rf8 28.Rc1 Qg6 29.Rcc2 Rf7 30.Rc1 Qf6 31.Rcc2 Rf8 32.Rc1 Qf7 33.Rcc2 Qh5 34.Rc1 Qe8 35.Rcc2 Bd6 36.Rc1 Qe7 37.Rcc2 Qf6 38.Rc1 Qg6 39.Rcc2 Qg4 40.Rc1 Rb8 41.Rcc2 Be7 42.Rc1 Rd8 43.Rcc2 Rc8 44.Rc1 Rc7 45.Rcc2 Qg6 46.Rc1 Rc8 47.Rcc2 Rb8 48.Rc1 Rd8 49.Rcc2 Re8 50.g3 Rb8 51.Rg2 Rf8 52.Rc1 Bd6 53.Rcc2 Qh5 54.Rc1 Qe8 55.Rcc2 Ndf6 56.Rc1 Ng4 57.Rcc2 Qg6 58.Rc1 Rh3 59.Rcc2 Qh5 60.Rc1 Kh8 61.Rcc2 Qg6 62.Rc1 Be7 63.Rcc2 Qh6 64.Rc1 Kg8 65.Rcc2 Bf6 66.Rc1 Qg6 67.Rcc2 Be7 68.Rc1 Rf7 69.Rcc2 Bd6 70.Rc1 Qh5 71.Rcc2 Qh6 72.Rc1 Be7 73.Rcc2 Bf6 74.Rc1 Qg6 75.Rcc2 Bd8 76.Rc1 Bc7 77.Rcc2 Bb8 78.Rc1 Rf8 79.Rcc2 Qh6 80.Rc1 Rf6 81.Rcc2 Rf7 82.Rc1 Rd7 83.Rcc2 Bd6 84.Rc1 Qg6 85.Rcc2 Rd8 86.Rc1 Re8 87.Rcc2 Qh6 88.Rc1 Rf8 89.Rcc2 Rd8 90.Rc1 Be7 91.Rcc2 Bf6 92.Rc1 Qg6 93.Rcc2 Be7 94.Rc1 Rb8 95.Rcc2 Re8 96.Rc1 Bf6 97.Rcc2 Rc8 98.Rc1 h6 A keen eye for the 50-move draw rule. 99.Rcc2 Rf8 100.Rc1 Be7 101.Rcc2 Bd6 102.Rc1 Bc7 103.Rcc2 Qf7 104.Rc1 Bd6 105.Rcc2 Be7 106.Rc1 Re8 107.Rcc2 Qg6 108.Rc1 Bf6 109.Rcc2 Ra8 110.Rc1 Rb8 111.Rcc2 Be7 112.Rc1 Bd6 113.Rcc2 Rb7 114.Rc1 Rf7 115.Rcc2 Bc7 116.Rc1 Rd7 117.Rcc2 Bd6 118.Rc1 Rd8 119.Rcc2 Qh5 120.Rc1 Rf8 121.Rcc2 Bb8 122.Rc1 Bc7 123.Rcc2 Rf7 124.Rc1 Re7 125.Rcc2 Bd6 126.Rc1 Qg6 127.Rcc2 Rc7 128.Rc1 Rc8 129.Rcc2 Ra8 130.Rc1 Kh8 131.Rcc2 Rf8 132.Rc1 Be7 133.Rcc2
and in this completely winning position (see diagram), Fritz 8 overstepped the time limit.

I did not invent this game myself. I just followed the strategy devised by Pablo Restrepo, who used it under his handle Father in a 3 0 blitz game against an anonymous computer named Elektrosmoker (Fritz 8?) on ChessBase's Playchess server. Father won on time in 144 moves.
    This was picked up by Eduard Nemeth, a well-known computer-buster; see my story Defending Humanity's Honor. He published the game and used the Father-opening himself to beat Fritz 8 even in a 10-minute blitz game, in 142 moves - on time.
    And now I duplicated the Father Opening and the general strategy, too. I must confess that my above win was only my third or fourth attempt, but that does not matter - the point is that such a game can be played at all against a computer which boasts a rating of 2752.

Related items in Open Chess Diary: 61, 114, 152 and 168.

Thanks to Jan Hondebrink for tipping me about the Father Opening.

286. 20 May 2005: Delay

Just out in Great Britain, with Bloomsbury: my novel Delay, translated by Sam Garrett.

Saturday 28th May, 14:30 I'll be talking about it at the Guardian's Hay Festival in Hay-on-Wye. "A Dutchman looks up an old flame in Sydney," as the site says. Nothing about chess.

285. 13 May 2005: New edition of the chess bible

In 1982, the Dutchman Harold van der Heijden began collecting endgame studies which contained a promotion to Rook or Bishop. He drew diagrams on system cards, and glued pieces to them that he clipped from the diagrams of "boring games" in magazines and papers.
    In 1988, he became fascinated by the possibility to digitalize endgame studies. Next morning he started with his promotions collection, and when he had completed that, he set himself the task of doing all studies.
    In 1991 he distributed a first version of his database on diskettes, which had 23358 studies. In 2000, after 12 years of entering endgame studies at a few hours a day, his collection had expanded to 58801. It was then published and marketed on CD-ROM by ChessBase as StudyDatabase 2000 and became famous as the bible of endgame studies.
    Now, he is privately marketing the newest version of his database, updated to and including April 2005, and containing 67691 studies - again on CD-ROM. It does not only add close to 9000 studies, it also updates around 10000 studies that were already included with new variations, cooks, Endgame Tablebase knowledge, and other information. Van der Heijden estimates that his database contains some 80 % of all endgame studies ever published, and around 99 % of what is relevant. If you would spend two hours a day playing over one study every 5 minutes, you would have eight years of enjoying chess beauty. This must be the chess publication of the year.
    Endgame Study Database III (which also contains the earlier versions of the database for reference) is priced at € 60, including VAT and postage. For information and orders, use this email address. Van der Heijden also has his own website.

Asked for potential classics among the newcomers in his database, Van der Heijden named three studies - numbers 3228 and 1627, and this winner of what was possibly the strongest tournament of the past 10 years:

White to play and win
S. Osintsev
First Prize, Shakhmatnaja Nedelya 2003

After 1.Bxf2+? Kg5 the mating threat on h8 decides. 1.e7 Rh8 Easier for White is 1...fxg1Q+ 2.Kxg1 Rc8 3.gxf7 Rgg8 4.Bh6! The f2-pawn has a different job. 2.gxf7 Rg7 3.e8R! And not 3.e(f)8Q Kg4+ 4.Qxh8 Rh7+ 5.Bh6 Rxh6+ 6.Qxh6 f1N+ and 7.Rxf1 stalemate, or 7.Kh1 Ng3+ 8.Kh2 Nf1+ with a repetition. After 3.f8R Kg4+ 4.Rxh8 Rh7+ 5.Bh6(?) Rxh8 Black even wins 3...Kg4+ 4.Rxh8 Rh7+ 5.Bh6 5.Rxh7 f1N+ is a draw once more. 5...Rxh8 Now 6.f8Q Rxh6+ 7.Qxh6 f1N+ would draw, but: 6.f8B! fxg1Q+ 7.Kxg1 (see diagram right)
The Rook is imprisoned by two unicolored Bishops. 7...Kg3 8.Kf1! 8.Kh1? Kf2 9.Kh2 Ke2 10.Kg3 Kd3 11.Kf3 Kc2 12.Ke2 Kb1 13.Kd2 Kxa2 14.Kc2 Rg8 15.Bhg7 Rxg7! would only result in stalemate. 8...Rg8 9.Bhg7 Kg4 After Kf4 10.Kf2 Ke4 11.Ke2 Kf4 12.Kd3 the a-pawn will promote. 10.Kf2 Kf4 11.g3+ Kg4 12.Kg2 Kf5 13.Kf3 Kg5 14.g4 Kg6 15.Kf4 Kf7 Even now White still has a chance to go wrong. After 16.Kg5? Rxg7! Black obtains the opposition. But: 16.Kf5! Rxg7 17.Bxg7 Kxg7 18.Kg5 and White wins.
    A brilliant study which, apart from everything else, shows all four promotions.

Maintaining his database, and being a central figure in the endgame studies world in many other respects, it is a miracle that Van der Heijden also develops spectacularly as a composer himself. In 2000, his database only had 14 of his own studies, and now there are 103 (a few jointly), 21 having won prizes, among which 6 First Prizes. In the most recent individual composing World Championship, he shared 7th place, ahead of all other westerners.

White to play and draw
Harold van der Heijden,
Prize, Rozlov-40 JT, 2003

1.d3! A very unlikely-looking key move with two hidden points. As Van der Heijden says: "Who would find this in an over-the-board game?" 1.h4? loses; 1...Rh3 2.Kg5 Kb3 3.h5 Kc4 4.h6 Kd5 5.Kg6 Ke6 6.Kg7 Ke7 7.h7 Rg3+ 8.Kh8 (Or 8.Kh6 Kf7 9.h8N+ Kf6 and Black wins) 8...Kf7 and Black mates quickly. Other tries that do not work are 1.Kg4 Rb1 2.h4 Kb3 3.h5 Kc4 4.h6 Kd5 etc. and 1.d4 Rh3 2.d5 Kb3 3.d6 Kc4 etc. 1...Rxd3 Black has no good alternatives, e.g. 1...Rb1 2.h4 Kb3 3.h5 Rh1 4.Kg5 and the first point of 1.d3 is revealed - 4...Kc4 is not possible here. 1...Rb5 also does not win: 2.h4 Kb3 3.Kg4 Kc3 4.h5 Kd4 5.h6 Ke5 6.h7 Rb8 7.Kg5 Ke6 8.Kg6 etc. 2.h4 Kb3 3.h5 Kc4 4.h6 Kd5 5.Kf5 (White can interchange his 4th and 5th moves) 5...Rh3 6.Kg6 Ke6 7.Kg7 Not 7.h7? Rg3+ 7...Rg3+ 8.Kf8! Rh3 9.Kg7 Ke7 10.h7 Rg3+ 11.Kh8 Kf7 stalemate. That is the second point of 1.d3; White had to get rid of that pawn - compare the variation after 1.h4.

284. 7 May 2005: New steps in the Babson saga

In item 226 (1 September 2003) I showed two cyclic Babson Task problems by Peter Hoffmann. As he explained there, the cyclic scheme, in this case QB-BR-RN-NQ, is more difficult to accomplish than the normal, full echo Babson; QQ-RR-BB-NN. Hoffmann had in fact needed some supernumerary pieces; two Bishops and one Knight in a first version; just two Bishops in an improved one.
    In last month's Die Schwalbe (the leading German problemist's magazine), he approached perfection further, with two cyclic Babsons following the same scheme, in which he brought down the number of promoted pieces to one.

The first one has 24 pieces, and a promoted Rook.

Mate in 4
Peter Hoffmann
Die Schwalbe, April 2005

1.Nxg6 threatening hxg8Q
1...d1Q 2.hxg8B! (2.hxg8N+? Kxe6 3.Nf8+ Ke5 4.Rxe2+ Qxe2!) and now 2...Qd4+ 3.c4 Qxb2 4.Bxb2 mate, or 2...Qxd7+ 3.exd7 Kg7 4.c4 mate, or 2...Qxc1 3.Qxc1 b2 4.Qf4 mate.
1...d1B 2.hxg8R! 2.hxg8Q? stalemate, or 2.hxg8N+? Kxe6 3.Nf8+ Ke5 4.Rxe2+ Bxe2! 2...Kxe6 3.Rhh7 Kf6 4.Rd6 mate.
1...d1R 2.hxg8N+! (2.hxg8Q? Rd4+ 3.c4 stalemate) 2...Kxe6 3.Nf8+ Ke5 4.Rxe2 mate.
1...d1N 2.hxg8Q! (2.hxg8N+? Kxe6 3.Nf8+ Ke5 4.Rxe2+ Ne3!) 2...Nxb2+ 3.Bxb2 or 3.Kb5 and 4.Qf7 mate 3...c1Q 4.Qf7 mate

Some side variations:
1...Kxe6 2.hxg8Q+ Kxd7 3.Qd5+ Kc7 4.b8Q mate
1...Rxg6 2.Rxe2 Rxg4+ 3.c4+ Kf5 4.Rf7 mate
1...dxc1Q 2.hxg8Q Qf4+ 3.c4+ Qe5 4.Qxe5 mate
1...Ra8+ 2.bxa8Q dxe1Q 3.Qf8+ Kxe6 4.Qf7 mate
1...Rb8 2.Rxe2 Ra8+ 3.bxa8N d1Q 4.Rf7 mate
1...Rg7 2.c4+ Kxe6 3.Nf8 mate
1.Rxg8? Rg7!

A second new, and quite different version is even lighter with only 22 pieces, including one promoted Knight.

Mate in 4
Peter Hoffmann
Die Schwalbe, April 2005

1.Bxf7 (1.Be4? exd1N! or 1.Ne3? Bxg6!) and now:
1...e1Q 2.dxe8B! Not 2.dxe8N+? Kxf7 3.Rfxf2+ Qxf2!, or 2.dxe8Q? Qxe7+ 3.Qxe7 stalemate. And now: 2...Qe5+ 3.d5 Qxa1 4.Qxa1 mate, or 2...Qe3 3.Nxe3 d1Q 4.Nf5 mate, or 2...Qxe7+ 3.Nxe7 Kh8 4.d5 mate.
1...e1B 2.dxe8R! and not 2.dxe8Q stalemate or 2.dxe8N+? Kxf7 3.Rfxf2+ Bxf2! 2...Kxf7 3.Nh8+! and mate next move.
1...e1R 2.dxe8N+! Not 2.dxe8Q? Re5+ 3.d5 stalemate. 2...Kxf7 3.Rfxf2+ and mate next move.
1...e1N 2.dxe8Q! (2.dxe8N+? Kxf7 3.Rfxf2+ Nf3!) 2...Nd3+ 3.Kb6 Nc5 4.Qg8 mate.

283. 2 May 2005: A smaller board

The endgame databases technique is approaching perfect chess from two directions. In item 282 just below, I reported on the first 7-men database. The 32-men database is out of reach because it has too many positions (more than there are atoms in the universe) - but 3x3 chess has been solved.
    This has been done independently by two researchers, Kirill Kryukov and the first name-less Aloril. They use slightly different rules (Kryukov allows pawns on the first ranks, but not the double step; Aloril only on the second rank) but with both, pawns promote on the first and third ranks, and they agree that 3x3 chess does not have a starting position. They constructed the database of all legal positions with 9 or fewer men on the 3x3 board; with Kryukov's rules, there are just over 300 million.

White can mate in 16 moves here - the longest quickest mate in 3x3 chess. (There are some other mates in 16.) White aims for the quickest mate, Black for the slowest.

1.Nxb1 Kxc1 2.Nc3 Bb1 3.Nxb1 Kc2 4.Nc3 Kc1 5.Na2+ Kc2 6.Nc1 Kxc1 7.Kxb3 Kb1 8.Kc3 Kc1 9.b3N+ Kb1 10.Bb2 Ka2 11.Na1 Kb1 12.Nc2 Ka2 13.Ba1 Kb1 14.Kb3 Kc1 15.Bb2+ Kb1 16.Na3 mate.

Alternatives can be tested on Kryukov's site - in fact, God's Algorithm for any position can be looked up there.

In this unique position, the lone King survives for 32 plies - after White's first move, it is again a mate in 16 by Black.

1.Kb3 Na3 2.Kxa3 Rc2 3.Kb3 Kc1 4.Ka3 Qb1 5.Kb3 a1N+ 6.Ka3 Qa2+ 7.Kxa2 Nb3 8.Kxb3 Kb1 9.Ka3 Ka1 10.Kb3 Rc1 11.Ka3 Rb1 12.Kb3 c2 13.Ka3 c1Q 14.Kb3 Qc2+ 15.Kxc2 Ka2 16.Kc3 Rc1 mate.

There is an interesting page on Aloril's site where the two grandmasters of 3x3 chess discuss further developments. Kryukov is working on smaller databases on the 3x4 board, and Aloril says: "3x4 chess is solvable, ditto 4x4 chess but not with current algorithms and our resources." And he adds: I think that maybe 5x5 starting position is also in our reach (maybe draw)."

PS 3 May: As Manny Rayner points out, I fell for the oldest trap in chess - there are not more chess positions than atoms in the universe. An accepted number (see link) for the latter is "between 10^78 and 10^81", and a serious approximation (see link) of the number of possible chess positions is 2.28*10^46. And that is even less than the estimate for the number of atoms in the Earth - 9 x 10^49.

282. 12 April 2005: First 7-men endgame database

In today's EG (no. 156) I found an interesting article by Marc Bourzutschky about the famous 4 Knights vs. Queen endgame, which was first analysed by Troitzky in 1912. The largest endgame databases were for six men up to now, but in just four days, Bourzutschky's home computer compiled this 7-men database. It is much smaller (15 billion positions) than the average 7-men database would be (around a trillion (1,000,000,000,000) positions) - the absence of pawns and the presence of 4 identical pieces greatly reduces the number of essentially different positions.

Bourzutschky found that 62.5 % of the initial positions are winning for the Knights, the longest quickest win lasting 85 moves.

White to play and win
Bourzutschky, 2005

All the moves aim for the quickest win, or the slowest loss. Only-winning moves are marked with an asterisk.

1.Nac8* Qa3 2.Nh4* Qb3+ 3.Ke2* Qb5+ 4.Kf2 Qb2+ 5.Kf1* Qb5+ 6.Kg1 Qg5+ 7.Neg2* Qf6 8.Kh2* Qe5+ 9.Kh3 Qe4 10.Kg3* Qe5+ 11.Kf3 Qf6+ 12.Ke3 Qh6+ 13.Kd3 Kc1 14.Kd4 Qg7+ 15.Ke4 Qg4+ 16.Ke5 Qg7+ 17.Ke6 Qc7 18.Kf6 Qc3+ 19.Kg6 Qg3+ 20.Kh5 Qe5+ 21.Kh6 Qh8+ 22.Kg5 Qd8+ 23.Kf4 Qc7+ 24.Kf3 Qh7 25.Kf2 Qc2+ 26.Kg3 Qc3+ 27.Nf3 Qg7+ 28.Kh3 Qf6 29.Kg4 Qe6+ 30.Kg3 Qg6+ 31.Kh2 Qc2 32.Nfh4 Qc7+ 33.Kg1 Qc5+ 34.Kh1 Kb1 35.Ng6 Qh5+ 36.N2h4 Qe2 37.Kg1 Kc2 38.Ng2* Qd1+ 39.Kh2* Kd3 40.Nd7 Ke4 41.Nc5+* Kd4 42.Ne6+* Ke4 43.Ng6f4 Ke5 44.Nb6 Qc2 45.Nd7+* Kd6 46.Ndf8* Ke7 47.Kg3* Qb1 48.Kg4 Kf7 49.Nh4* Qg1+ 50.Kh5 Qd1+ 51.Kh6 Qe1 52.Nhg6 Qa1 53.Nh5 Qh1 54.Nef4 Ke8 55.N8e6 Kd7 56.Kg5 Kd6 57.Nf6 Qe1 58.Ne8+ Kc6 59.Kf6 Kb6 60.Nd5+ Kb5 61.N8c7+ Ka5 62.Ngf4 Qa1+ 63.Ke7 Qa3+ 64.Kd7 Qa4+ 65.Kd6 Qa1 66.Nd8 Qa3+ 67.Kd7 Ka4 68.Nc6 Qa1 69.Nce6 Ka3 70.Kd6 Qc1 71.Ned4 Qf1 72.Kc5 Qc1+ 73.Kb5 Qf1+ 74.Nfe2 Qf7 75.Kc4 Qb7 76.Ndb4 Qh7 77.Nd3 Qg8+ 78.Kc3 Qd5 79.Nb2 Qg5 80.Ne5 Qh6 81.Nec4+ Ka2 82.Nd3 Kb1 83.Kb3 Qg7 84.Na3+ Ka1 85.Ndc2 mate. The by now well-known eerie beauty of incomprehensible chess.

Bourzutschky also found a few full-point mutual Zugzwangs.

White to play loses, Black to play loses
Bourzutschky, 2005

Again: quickest win, slowest loss; only-winning moves marked with an asterisk.

1.Kb2 Qg2+ 2.Kb3 Qb7+* 3.Ka3 Qb6 4.Nf4+ Kc4* 5.Ka2 Qb3+* 6.Ka1 Kb4 7.Ng7 Ka3 8.Nge6 Qb2 mate

1...Kc4 2.Kd2* Kd5 3.Ne3+ Ke5 4.Ng7 Qa8 5.N7f5 Qa5+ 6.Ke2 Qa2+ 7.Kf3 Qa8+ 8.Kf2 Qa1 9.Nh4 Ke6 10.Nf3 Qh8 11.Kg2 Qg8 12.Nf2 Kd6 13.Nd3 Qa2+ 14.Kh3 Qb1 15.Nfe5 Qa1 16.Ne4+ Ke6 17.Nf3 Kd7 18.Nde5+ Kd8 19.Kg4 Qc1 20.Nd5 Qc8+ 21.Kf4 Qe6 22.Nef6 Qh3 23.Nc6+ Kc8 24.Nfd4 Qh2+ 25.Kf5 Qh3+ 26.Ke5 Kb7 27.Ncb4 Qh2+ 28.Ke6 Kc8 29.Nf5 Qe2+ 30.Kd6 Qh2+ 31.Kc6 Kd8 32.Nd4 Qe5 33.Nb5 Qg3 34.Nd6 Qa3 35.Nb7+ Kc8 36.Nb6+ Kb8 37.Nfd5 Qa1 38.Nd6 Qc1+ 39.Kd7 Ka7 40.Nbc8+ Kb8 41.Na6+ Ka8 42.Ndb6 with a unique pure mate - a Knight for each square.

281. 3 April 2005: Dancing down the staircase

As already mentioned in item 276, Hans Böhm's Humor Tourney for endgame studies (jury: Böhm, Jan Timman, Harold van der Heijden and me) was won by Sergiy Didukh from Ukraina.

White to play and draw
S. Didukh
First Prize Humor-tourney, 2005

White's only asset is his attack against the black King. 1.Bd6 would threaten a perpetual on h7 and g7, but 1...Qc3 refutes that: 2.Re5 Qxe5 3.Bxe5 Rxe5 4.Rxh7+ Kg8 5.Rg7+ Kf8 6.Nxc5 Rd8 7.Nxd7+ Rxd7 8.Rxd7 Ke8 etc. and Black wins.
1.Be6 White must play for stalemate and therefore, get rid of his pieces. This is a double mating threat, so: 1...dxe6 and only now 2.Bd6 Qc3 3.Nd4! Shutting the Queen off from g7, and threatening a perpetual again. 3...Qxd4 Or Qd3 4.Be5 Rxe2 (4...Qxd4 5.Ra2 Rxa2 6.Rg6+ Rxe5 7.Rg8+ Kxg8 stalemate) 5.Rxh7+ Kg8 6.Rg7+ Kf8 7.Bd6+ Ke8 8.Re7+ and Black must go back, with a draw. 4.Re5 A shut-off of g7 again, and 4...Qxe5 5.Bxe5 Rxe5 6.Rg8+ Kxg8 would be stalemate once more. Black must guard h7 4...Qd3 5.Re4 Shutting off h7. 5...Qc3 After Qxe4 6.Be5! Black must settle for something like 6...f3 7.Rxg3+ Qxe5 8.Rg8+ and stalemate. And 5...Qd4 is met by 6.Re5 again. 6.Rd4 The Rook must follow the Queen on her heels; after 6.Re5 Qc2 7.Re4 Qb2 8.Re5 Qb1 9.Re4 Qa1 10.Re5, Black can play 10...Ra7 11.Rxa7 Qxa7 12.Rg5 f3 and wins. 6...Qc2 7.Rd3 Qb2 8.Rc3 Qb1 9.Rc2 Qa1 10.Rb2 Not 10.Rc3 Ra7! - the idea of this mutual dance down the staircase is that as soon as the black Queen reaches a1 and Ra7 would thereby become possible, White must be able to play Rb8 mate 10...Qxb2 Settling for stalemate after all. 11.Be5 f3 Or 11...R(Q)xe5 12.Rg8+ Kxg8 stalemate, or 11...Rf8 12.Rxh7+ Kg8 13.Rg7+ with a perpetual. 12.Rxg3+ The last finesse; otherwise the Bishop takes on e5 and there is no stalemate. 12...Rxe5 13.Rg8+ Kxg8 stalemate.

I copied the picture from the study magazine EG no. 155, which mentions that Didukh is 28 and a teacher of French in the little West Ukrainian village of Andrivka, at a salary of $ 80 a month. So his € 250 for this study and a further € 50 for the 1st Honorary Mention that he also won, might come in handy. Only starting in 2003, he has published some 40 studies by now, and must be seen as one of a promising new generation of composers.

© Tim Krabbé 2005

Previous | Current
Top | Main chess page | Main page | Contact me