Proving the Draw^H^H^H^H Win After 58...Qf5

by Kenneth W. Regan


 Diagram 1: Position after 58...Qf5

With Irina Krush's analysis never being posted by MSN, the World chose the inferior 58...Qe4?, which was already known to lose albeit in long lines. Krush and the St. Petersburg GM School had worked out a much tougher defense with 58...Qf5, one that would have forced Kasparov to exhibit his best chess. White's main winning ideas are:

Status Update 11/20: From Peter Karrer's 6-piece KQP(g6-g7)KQP(d5-d2) EGTB compilation completed Friday, White has two separate ways to win: by 66. Kf6! Qb6+ 67. Kf7! Qb7+, and now not Kasparov-Alterman's "{Key idea} 68. Kg8 Qc8+ 69. Qf8 Qe6+ 70. Kg7" (which is about the only natural sequence from this position that throws away the win permanently) but 68. Ke6!. Now 68...d4 fails not to the seemingly killing 69. Qf7 Qa6+!!=, but to 69. Qf2+ Kb1 70. Qg1+! Ka2 71. g7, and the cruel truth is that unlike the analogous "Position G" line where Black's Queen checks White's King on f7 from f5, White's King will reach the key square h6 in a way that Black cannot reply ...Qh4+ any more, and then Black cannot prevent a killing interposition by the White Queen way-down-there on g1. Since a 6th-rank check allows a mighty Qd6 interposition, Black has only 68...Qc8+ 69. Kf6 Qd8+ (best) 70. Kf5, but then faces the problem that White can't be allowed to scoot his King back into h6 here: 70...Qd7+? 71. Kg5 Qe7+ 72. Kh6 Qe6 73. Qf3! is the "splat" we all knew well. Hence Black has to transpose into Fritz's 65...Qc3 defense via 70...Qc8+ 71. Kg5 Qc3, but that is undone in several ways and optimally by a second delectable Zugzwang: 72. Qh2+! Ka1 73. Qe2! Kb1 (nothing better---Black's Queen has no good moves and ...d4 allows g7) 74. Qf2!. Now 74...Ka1 allows 75. Qf6! d4 76. Qa6+!! Kb1 77. Qf1+ Ka2/b2 78. Qg2+! Ka1 and finally now 79. g7, and see above note about h6 and White's Queen on the g-file. Thus Black must play 74...Qc1+ 75. Kg4! Qc3 76. Qf1+ Kb2 77. Kf5!, and now the point is that White has taken a bite out of the e5 square on which ...Qe5+ was previously saving Black in lines where White checks around to b7 and plays g7. Now Black has to fall back with 77...Qc7, but at this point the defense is recognizably broken and White concludes with 78. Qe2+ Kb1 79. Qd3+ Ka2 80. Qa6+ Kb3 (80...Kb1 meets 81. Ke6 to tighten the ring) 81. Qe6! Ka2 82. Qf7! Qc2+ (if only Black could check on b1 this would be completely drawn!!) 83. Ke6! (Tableau!) Qe2+ 84. Kxd5, and at this point we recognize EGTB +/- from the 5-piece tables we had during the game. ("Tableau!" means "Look at this beautiful set-up!", and "^H" as in my revised title is the code for backspace.)

White has a second, independent winning line, with 66. Kf6! Qb6+ 67. Kg7, also deserving "!". Now the line citing Gorski down in my "Line B" below is incorrect: 67...d4 68. Kh7! Qb5!? is smashed by 69. Qh2+!---here I and all simply forgot our knowledge of earlier winning ideas. (My 69. Qg4"!" also wins---I didn't notice a key difference between White's Queen being on g4 versus g5 in my Position-G "Part II" line beginning 1. g7 d4 2. Qf3 d3! and the main line given there.) So 67...Qe6"!" is Black's best after all, and then 68. Qf6(!) takes play into what I had labeled "Main Line B" in
my original 58...Qf5 analysis file as of 10/28, and last week started calling the "Last-Ditch" line. That line I originally gave with "68. Qf3!" in place of 68. Qf6, before I knew Position G was a draw. On 68. Qf3 Qe5+, Black cannot /force/ Position G (Black can do that only against Kasparov's moves), but Black can force White to repeat by 69. Kh6(!) Qh8/h2+ 70. Kg5 Qe5+ 71. Qf5 Qe7+ (technically going into "Last-Ditch" with 71...Qg3+ takes longer to lose, but making White repeat is more in the spirit of the game of course). Now the independent attempts 72. Kh5, 72. Kf4, and 72. Kg4 all lead after repetition after 72...Qg7!, the last continuing 73. Qf7 Qd4+ 74. Qf4 Qd1+! 75. Kg5! Qg1+ 76. Kf6! and White is back to the earlier wins---so the GM-School's Qg7 idea had real holding power. Thus White has nothing better here than to repeat with 72. Qf6 Qe3+, etc.

Other moves besides 68. Qf3 and 68. Qf6 listed there are more efficient only because they rewind to the Zugzwang to play the more-efficient 67. Kf7 Qb7+ 68. Ke6 line. It follows that White has exactly two winning lines, and forcing "Last-Ditch" by 68. Qf6 is White's only independent way to win. This is the plan shown on the first two pages of the Kasparov-Alterman file, and transposes into what I gave as "Line B" (main line before my knowing G was a draw---this was written on 10/19!) by transposition after 68...Qd7+ 69. Qf7 Qg4/c8 70. Qa7+ Kb1/b2 71. Qb6+ Ka2 (EGTB shows 71...Kc2 holding out longer and best with 69...Qg4, but all(?) the long-hold lines are where Black plays an early ...d4 or lets White munch Kxd5 producing 60/70-odd move EGTB wins we knew about in the game) 72. Kf7 Qf5+ 73. Ke7 Qe5+ 74. Qe6 (74. Kd8? d4! draw! since White can do no better than 75. Qa6+ Kb2 76. Qb7+ Ka1! 77. g7 Qa5+! and we've transposed into the wildest trickiest draw line in I.D of my "G" analysis!) 74...Qc7+ 75. Ke8 Qb8+. Here Kasparov and Alterman gave 76. Kf7"!" heading toward Position G, but now we know White must go the other way. Now this "Pin Battery" too can be defused by Black by a sequence of what the 5-piece EGTBs identified as only-moves: 76. Kd7 Qa7/b7+ 77. Kd6 Qb8+! 78. Kc5! Qf8+! (my treatment inserted Kc6 Qa8+! to show that point too, but this causes move numbers to be up by 1) 79. Kb5 Qb8+! Here 80. Qb6? actually only draws, and while White had alternatives earlier that (also) win (but sub-optimally), they all involve the key idea 80. Ka5!! that I gave two weeks ago. I will complete what happens after this's gorgeous, but what I'd posted and collected in the "Addendum" in
58Qf5extra.html is correct. (Artistically I really wish the win with 68. Ke6 did not exist.)

Absent my two oversights as noted in the first two paragraphs, everything else here and in
PositionG.html checks out to the very letter, including my defenses in Lines C and D of 58Qf5.html being the only ones (though the D one only forces White to repeat as noted above). Written quickly during kids' naps and sans-board, apologies for any typos. My original 58...Qf5 analysis as of 10/19 and 10/28 was in some ways closer to the truth than the rest of this file, but I have to leave the rest of this file as-was, just a few tiny "Update" inserts. --KWR)

  1. "Brute-Force Breakthrough": forcing through g7 with White's King nearby, having tricked Black's Queen away from checks or pins that could prevent this advance. Kasparov and Alterman show that White can achieve "Position G" this way.
  2. "Queenside Run": White's King runs behind Black's d-pawn, so that this pawn blocks the checks Black would need to hold. Further, by placing his King on the same file as Black's---or on the a-file after driving Black's King to the b-file---White can either avoid checks by having pinning or cross-checking interpositions, or maneuver his Queen behind his King to cover g7 or g8 and set up a discovered check. The Kasparov-Alterman main line ends this way.
  3. "Cannon": White lines up King, Queen, and Pawn on the g-file, loads his King back to g2 so that Black runs out of checks and must block by ...Qg7, and then "fires" by some f-file checks followed by Qf7. This forces through g7, and if Black manages to make White's Queen interpose and then occupies g8, then the "Cannon" fires again with White's King on g1 by Qf8.

For other general strategic themes and major patterns of this endgame, see the accompanying article 58Qf5themes.html.

The moves up to 65. Qf4 are agreed on as the main line by all sources---the World Team and Kasparov worked out the same drawing defenses against other White tries as shown in the first parts of the last SCO FAQ and K-A analysis. Then Black has an independent defense with Fritz's 65...Qc3!? that still begs for attention---Kasparov and Alterman handle it only by transposing to their main line against 65...Qg1+. We will, however, concentrate on 65...Qg1+ and make White's Move 66 the main branch point.

58. ...


59. Kh6


 Diagram 2: The "GM School Position", fodder for Zugzwang.

Black's "holding strategy" exploits the fact that White's blocking Queen on d4 does not protect g6, so that Black can force White's King to stay near the g-pawn with checks. For example, on 60. Kh5 Qf5+ or 60. Kg5 Qe7+ 61. Kg4 Qe6+ or 60. Kg5 Qe7+ 61. Kf5 Qd7+ 62. Kf4 Qd6+, White has to turn back. Here 60. Kg5 Qe7+ 61. Kh5!? is interesting, but Black can continue 61...Qe2+ 62. Qg4 Qe5+, and faces much less danger than with his King on a2 as in Lines C and D. (In particular, 63. Kh4 Qe7+ 64. Kh3 Qe3+ 65. Qg3 Qe6+ 66. Kh2 is no longer winning for White: 66...d4! 67. Qg1+ Kc2! 68. Qg2+ Kb1 69. g7 Qh6+ is a perpetual, and 68. Qg5 d3 69. g7 d2 is just a draw. Also here 63...Qg7!? seems well-timed for Black, and Black has other checking ideas to boot.)

The only square Black's Queen can hold from is e6, however, so Black is susceptible to Zugzwang. White's next objective is to bring about this position with Black to move, when Black will have to move his King away from the safest square, b1. There seem to be several ways to achieve this, including 60. Qb4+ Ka2 61. Qf4, when 61...d4 is impossible owing to 62. Qf7+! (see point 3. above), and Black seems to have nothing better than 61...Kb1 62. Qd4! Sentiment on the BBS favored the pretty triangulation sequence 60. Qd3+ Kc1 61. Qc3+ Kb1 62. Qd4, though it gives Black possible extra chances by 60...Kb2!? Kasparov was going to play 60. Qd1+ Kb2 61. Qd2+ Kb1 62. Qd4, and equally forcing and simple seems:

60. Qg1+


61. Qf2+


62. Qd4!


(on 61...Kc1, White can at least reach the same position by 62. Qf4+ Kb1 63. Qd4). This Zugzwang leaves Black not much more than 62...Kc2 or 62...Ka2 since 62...Qh3+ seems to help White and other Queen moves cede ground. Now the drawback of 62...Kc2, which was part of the original GM-School strategy, is revealed by 63. Kg5 Qe7+ 64. Qf6 Qe3+ 65. Qf4 (65. Kg4! looks even better) Qe7+ 66. Kg4 (Kh5 may also be good), when the hoped-for liberating 66...d4 simply meets 67. Qxd4! with a computer-proven checkmate in 45 moves! Kasparov and Alterman state that Black's next is an "only-move".

62. ...


63. Kg5


64. Qf6


65. Qf4


Now going back by 65...Qe7+ runs into a diabolical problem: 66. Kh6 Qe6 67. Qf3! White's last move prevents 67...d4 by the 68. Qf7+ motif, and equally important, covers the squares h3, e4, and f5, one of which Black needs to continue his holding strategy after Kh7 by White. There appears to be no sturdy defense left here---this is a case where Strategy (1) ("brute force breakthrough") works. Eventually we promise a full treatment of Fritz's idea 65...Qc3!?, but priority was given to the thematic idea of stopping Strategy (3) by checking from behind on White's first rank:

65. ...


Diagram 3: Main branch point for the 58...Qf5 endgame.

Now all of White's legal replies form four independent plans, all promising in different ways!

    A. Run King back to cover on g7: 66. Kh6 Qh1+ 67. Kg7 or 66. Kf6 Qb6+ 67. Kg7.
    B. Kasparov-Alterman plan: 66. Kf6 Qb6+ 67. Kf7!
    C. "Cannon" with 66. Qg4
    D. "Cannon" with
66. Kh5.

In addition, in the separate file 58Qf5extra.html we will cover important lines where White appears to be winning but some questions remain:

    K. White's win against 62...Kc2.
    L. "Last-Ditch" defense if Black fails to achieve Position G.
    M. Kasparov-Alterman Main Line---final part.

A. Running back to g7, with Black's Queen unable to return to the h3-c8 diagonal. Two independent lines embody this idea:

    A1. 66. Kh6. This is AOK for Black after 66...Qh1+ 67. Kg7 d4, intuitively because White's King is still on g7, Black's Queen controls the important square d5, and Black has many checking options when White's King moves. Indeed, the strength of Black's setup is exemplified by the position being drawn even if White is allowed to capture on d4 with check: 8/6K1/6P1/8/3Q4/8/8/k6q+b. Since ...Kb1 as well as ...Ka2 draws in that position, Black's King can even travel to c2 to assist the d-pawn without fearing an EGTB loss after Qxd4! White's two best lines seem to be:

     (i) 68. Kf6 (to leave Black's King on a2) Qc6+ 69. Kg5 (69. Kf5 Qd7+ and White can't even play 70. Ke4 Qe6+, while 70. Ke5/g5 d3 71. Qf7+ Qxf7 72. gxf7 d2 is =) Qd5+ 70. Qf5 Qd8+ 71. Kf4 d3=; and

     (ii) 68. Qd2+ Ka1(!) (so that White does not even achieve Qd3 with check) 69. Qa5+ Kb2 70. Qb4+ Kc2 (probably ...Kc3 is even better) 71. Qc4+ Kb2, and the point is that White's inability to cover g6 and e4 kills all King-run hopes. And 72. Qd3 Qd5 73. Kf6 Qd6+ 74. Kf5 Qd5+ 75. Kf4 Qd6+ 76. Kg4 Qe6+ ultimately goes nowhere because Black gets to check on e5 or f6. With the better White move order 69. Qd3 Qd5 70. Kf6 Qd6+ 71. Kf5 Qd5+ 72. Kf4 Qd6+ 73. Kf3 Qf6+! 74. Kg2, Black even has time for 74...Kb2!, when 75. Qg3 d3! 76. Qxd3 is EGTB= and 76. g7 Qc6+ is a perpetual check with winning chances only for Black if White retreats his King to the first rank.

    A2. 66. Kf6 Qb6+ 67. Kg7!? We initially feared this suggestion by IM2429, and the true antidote was found only recently, following up a suggestion by Rafal Gorski! White's idea is that after 67...d4 68. Kh7!, Black cannot hold even by achieving "...d3 before g7".

Diagram 4: Rafal Gorski's find.

Here 68...d3? loses to 69. g7 Qb7 70. Qd2+ Ka1 71. Qc1+ Ka2 72. Kh8---and thematically also to 70. Qd4(!) entering the "Hogtie" setup.

11/20: This is where my edits on Wednesday evening stopped, and I left "B" in quite a mess---partly explained by its being where I was wrong!---see the note on 68. Kh7! Qb5 69. Qh2+! above and my flip-flops below.

B. 66. Kf6, when Black can stop a run to the Q-side but cannot play ...d4 as in A, so this "runaround" by White's King displaces Black's Queen. It is worth displaying as a main line:

66. Kf6


67. Kg7!

Qe6! (no, ??)

Here 67...d4 does not quite work: 68. Kh7! and now Black cannot stop g7 (68...Qb1 69. Qf7+ leads to a trade of Queens and win). All lines we looked at after IM2429 suggested 67. Kg7"!" seem to meet fates similar to: 68...d3 69. g7 Qb7 70. Qd2+ Ka1 71. Qc1+! (to cover h1) Ka2 72. Kh8 and Black cannot stop White from queening. Rafael Gorski sent me 68...Qb5!? (here or on Move 67), but White seems to have a bust with 69. Qg4!? d3 70. g7 Qb7 71. Qd4!, a theme that comes up in other lines. Even if White played 71...Qh1+ 72. Kg8? Black would still be unable to let the d-pawn fall with check (this note is relevant to A.), but 72. Kg6! sets up the theme of running White's King around to capture Black's d-pawn, after which the position is a known EGTB win.

Update, 11/15: Both the claimed bust of Gorski's move and my 67...Qe6(?) are wrong, the latter because 68. Qf6! transposes into Kasparov-Alterman lines that lead the same place as 67...Qe6 68. Qf3 Qc8 does below. The answer to 67. Kg7 is 67...d4! 68. Kh7 Qb5! 69. Qg4 Qd3!, a move I initially discounted because it blocks Black's d-pawn, but it's fine: 70. Kh8 Qf1! (does 70...Qe3 71. g7 Qh6+ 72. Kg8 d3 lose?) 71. g7 Qh1+ 72. Kg8 d3 73. Qd4 Qf3! is not the same problem as before because White's King is contained.

[Update 11/20: White still wins here by 73. Kf7 as it's not the same as in my Position G, line II.H with 1. g7 d4 2. Qf3 d3!---White's Queen is better on g4 here than when it reaches g5 there. But anyway all this is academic owing to the overlooked, smashing 69. Qh2+!]

And on 71. Qe6+ Kb2 White cannot maneuver with check to arrange an interposition on h7, while Black can even afford to pitch the pawn on d4 with check! Nor does 70. Qe6+ bring anything. Thus 67. Kf7! as given by Kasparov-Alterman does deserve its "!", with the idea being that any check now by Black makes ...d4 impossible on the next move. However, Black can still force "Position G" as shown in multiple lines, with the hairiest being 67. Kf7 Qb7+ 68. Kg8 Qc8+ 69. Qf8 Qe6+ 70. Qf7 (instead of K-A 70. Kg7) Qc8+ 71. Kg7 Qc3+ 72. Kf8 Qb4+! 73. Qe7 (73. Ke8 gives Black an immediate perpetual from the squares b8,b7,b6,b2,f2,d2,h2 as Kxd5 is EGTB= with Black's Queen on b6 or b2) Qf4+!, and since 74. Ke8 Qb8+ 75. Kf7 Qf4+ 76. Qf6 Qc7+ 77. Kg8 Qc8+ leads back to the beginning, White has nothing better than 74. Kg8 Qb8+ 75. Qf8 Qe5 = G. Now we re-join the analysis file the way it originally was.

However, Black's last move seeks to re-establish the GM-School holding strategy. Were White's King on h6 White's next would be winning as mentioned above, but even here it is strong:

68. Qf3! (no, ?: Qf6!)


Now available checks all seem to put Black's Queen "on the wrong circuits." For instance, 68...Qe5+ 69. Kh7! Qh2+ 70. Kg8! leaves Black unable to stop g7 after 70...Qb8+ 71. Qf8---Black's Queen needs to be on c8 when White interposes on f8. The major question, however, is whether White is winning after 71...Qe5! here. This critical position

(WKg8, Qf8, Pg6; Bka2, qe5, pd5; White to move) is one even I have not been able to exhaust, nor the position resulting from 72. g7 d4---the latter was also relevant to the viability of 56...Qe3 if I recall correctly. If White plays 72. Qf7 Black just resumes checking, and all other non-check Queen moves allow ...d4. If White has a win here then this is a major step toward proving the whole ending won for White, but if not then several lines of defense besides this one become open for Black. (All diagrams are screen-captured from MacChess 5.0.1.)

It is worth noting that 70...d4 seems to fail quickly once again: 71. Qd5+! Kb1 (allowing d4 to fall with check is a computer-proven loss, and ...Ka3 is no help) 72. g7 Qb8+ (weirdly, 72...Qg3!? 73. Qxd4? Qb8+ is a computer-proven draw, but 73. Kf7 leaves Black nothing but transposition into this line) 73. Kf7 Qf4/c7+ 74. Kg6 Qg3+ 75. Kf6!, and Black is gone after 75...Qh4+ 76. Qg5. And 68...d4 is still impossible owing to 69. Qf7! Black has one other option whose point is to meet 69. Kh7 by ...Qc2!, when from prior analyses of lines with 51...Ka1 Black looks fine. Although it seems inferior to 68...Qe5+ and is probably a forced loss, the lines it leads to are interesting and potentially important enough to warrant mainline treatment here:

C. 66. Qg4

This try seems inferior to line D, but is worth including since some themes are similar. Both 66...Qe3+ and 66...Qc1+ seem acceptable in reply, with not much to choose between them. Now running Q-side with White's Queen only on g4 seems not to work, e.g. 66. Qg4 Qc1/e3+ 67. Kf5 Qc2/d3+ 68. Ke5 Qc3+ 69. Qd4 Qc7+! 70. Ke6 Qc6+ 71. Kf5 Qd7+ 72. Kf4 Qd6+ 73. Kg5 Qe7+ resets the game back to Move 63, and 69. Kd6 Qa3+! leads to no better as Black has enough room for checks. Hence most thematic seems:

66. Qg4


67. Kh4


68. Kg3


69. Kg2/h2


Black avoided 67...Qh1+?! on account of 68. Qh3! Qe4+ 69. Kg5! and suddenly Black is in a spot of bother after 69...Qe7+ 70. Kh6! since White's Queen controls key squares, or 69...Qe5+?! 70. Qf5! with themes from Line D. And Black is avoiding 69. Kg2 Qd2+ (or 69. Kh2 Qh6+ 70. Kg2 Qd2+) on account of 70. Kf1! Qc1+ 71. Kf2!, when 71...Qc5+ 72. Kg2 leaves Black no checks---although black may still be OK with 72...d4. This version of ...d4 is quite a bit better, however, and leads to a twin pair of positions that to me seem critical for line C:

I haven't had time to analyze it all, but these lines offer good comparison to themes in line D:

70. g7 Qd2+/h6+ is perpetual check along the c1-h6 diagonal, as White never gets a chance to interpose profitably.

69. Kh2 d4 70. Qg2+ Ka1 71. Qg1+ trades Queens but does not win, while 71. g7 Qh6+ is perpetual on h6, e3, and c1, and 71. Qa8+ Kb2 72. Qb7+ Ka1 73. g7 Qh6+ sets up a block and perpetual using h5,g5,f5,c2,c1.

69. Kh2 d4 70. Qg3!? Qh6+! 71. Kg2/g1 d3! seems to draw as White has no productive checks and 72. g7 Qc6+ 73. Kf2 Qc5+! 74. Kf1 Qc1+ 75. Qe1 d2 76. g8=Q+ Kb2! and 74. Kf3 Qd5+ 75. Ke3 Qc5+! (...d2 76. g8=Q d1=Q is a 4Q loss) 76. Ke4 Qc6+, and since White cannot do better than EGTB= on taking Black's d-pawn, we do not see what White is accomplishing: 77. Kf5 Qd7+ 78. Kg6 Qe6+, and the best White has is a "known pattern" typified by 79. Kh5 (Kg5? d2!=) Qf5+ 80. Kh6 Qe6+ 81. Qg6 Qe3+ 82. Qg5 Qe6+ 83. Kh5 Qh3+ 84. Kg6 Qg4+ 85. Kf7 Qd7+ 86. Qe7 Qf5+ 87. Ke8 Qc8+ "etc."---even with Black's pawn at d4 we suspect this is holding.

69. Kg2 d4 70. Qg3!? Qe6!? Now 71. g7 yields a perpetual check that constitutes Black's main defensive idea hereabouts: 71...Qc6+ 72. Kf1 Qc4+! 73. Kg1 Qc1+ 74. Kh2 Qh6+ 75. Kg2 Qc6+ 76. Kf2 Qc2+ 77. Kf3 Qc6+! 78. Kf4 Qd6+ 79. Kg4 Qg6+ 80. Kh3 Qh6+! 81. Kg2 Qc6+!, and Black is always able to reach the squares needed to prevent White from escaping without losing his g-pawn.

The first question is whether 71. Qg5 wins with White's King on g2 the same way it does with White's King on h2 in line D. if Black tries 66. Kh5 Qh1+ 67. Kg4 Qg2+? 68. Qg3! Qe4+ 69. Kh3 Qe6+ 70. Kh2! d4. Black's only possible defense seems to be 71. Qg5 Qe4+ 72. K-any Ka1!, when 73. g7 seems to yield a perpetual again---I think I can illustrate all the many points in one line: 72. Kg1 Ka1! 73. g7 Qb1+ 74. Kf2 Qc2+ 75. Kf1 Qd3+! 76. Ke1 Qc3+! 77. Ke2 (Qd2 Qg3+ 78. Kd1 is not close enough to give mate) Qc4+! 78. Kd1 Qb3+! 79. Kd2 Qc3+ 80. Ke2 Qc4+ 81. Kf3 Qd3+ 82. Kg4 Qe4+ 83. Qf4 Qg6+ 84. Qg5 Qe4+ 85. Kh3 Qh7+! 86. Kg2 Qe4+ 87. Kg3 Qd3+ 88. Kf4 Qd2+! 89. Kf5 Qc2+! 90. Ke5 Qc7+! 91. Ke4 Qc2+ 92. Kd5 Qc6+! 93. Kf4 (here Qd5 is not check and loses the g-pawn) Qc1+! 94. Kg4 Qd1+! 95. Kf5 Qc2+ 96. Ke6 Qc6+ 97. Ke7 Qc7+ 98. Ke8 Qc8+ 99. Qd8 Qe6+ 100. Qe7 Qc8+ 101. Kf7 Qf5+ 102. Qf6 Qd7+ 103. Kf8 Qc8+ 104. Ke7 Qc7+ 105. Ke6 Qc6+ 106. Ke5 Qb5+! (handle Pin Battery by keeping contact with the counter-skewer ...Qb2+) 107. Kf4 Qf1+ 108. Kg5 Qg2+ 109. Kh6 Qh3+, and hopefully by now it is clear (without my having to show King-run lines with White's Queen on e7 as well as g5,f6) that Black can always reach the needed squares.

The second question is whether Black survives being pinned down on g8 after 71. Kg1!? d3 72. g7 Qg8 because White cannot hop to d4 in one move, and 73. Qf2+ Kb1 74. Qf8 fails to 74...d2! 75. Qxg8 d1=Q+ and perpetual! This was a needed resource in some GM-School lines, but maybe with Black's king further away on a2, White has a way to use his King too... However, facts like 73. Kf2 d2 74. Ke2 Qe6+?! 75. Kxd2 Qd5+ being EGTB= (notice also 8/6P1/4q3/6Q1/8/8/k7/3K4+b with White's Queen better placed on g5: Black still draws with ...Qb3+) give Black cause for optimism.

69. Kg2 d4 70. Qf3!? Qe5!, and now the sneaky 71. Qf2+ Ka1 72. Qg1+ Kb2! does not win. Now 71. Qg3 Qe6 transposes into the most-delicate last variation, while supporting g7 from f7, b7, or a7 seems to yield Black a perpetual after ...Qg5+, a fact that was crucial to certain GM-School lines. I don't see how White makes progress after 71. Qd3 Kb2---since once Black's King escapes a2 White cannot win by opposing Queens---though this may contain hidden venom. And other non-checking moves allow ...d3, which seems always to be an EGTB draw if it is captured.

Whew! I'll admit that the above volume far from exhausts this line---and any one hole I may have missed makes the whole thing go blooey! (Please someone test it with better computer-chess software than I have.) But that's my manifesto for a Black draw in Line C.


D. 66. Kh5!?

66. Kh5


67. Kg4


68. Qf3


Along the way, Black has avoided a pretty loss with 67...Qg2+? 68. Qg3! Qe4+ 69. Kh3 (White's plan is to retreat the King to the first rank, when Black may be forced to block White's pawn by ...Qg7 but will have no check in response to Qf2+/f1+/f7. This seems inexorable unless Black can take evasive action.) Qe6+!? 70. Kh2! d4 71. Qg5! as found by JQB last Sunday. All defenses other than 71...d3 72. g7 Qg8 run out of checks as White gets a new Queen, but now 73. Qd2+ Ka1 74. Qc3+ Ka2 75. Qd4! leads to a standard winning theme in which Black is hogtied: losing the pawn with check is an EGTB loss when White's Queen returns to d4, and otherwise Black cannot stop White's King from wandering over and munching the pawn. In particular, 75...Qf7 76. Kg3 is another Zugzwang.

The amazing point of 68...Qc2! is that 69. Qxd5+ is EGTB=, even though the move seems to lead either to 69...Kb1 70. Qf5 pinning the Queen and game-over, or 69...Ka1/b2 70. Qd4+ K-any 71. g7 setting up the same win I just mentioned, or 69...Ka3 70. g7 and Black has no really useful checks and will soon die from the King's exposure on a3. The point is that 69...Kb1 70. Qf5 Ka1! escapes the pin by stalemate! Also amazing is that then 71. Qe5+ and 72. g7 is not enough to win---White's Queen needs to be on d4 for the EGTB trap to close. Hence this is actually a critical position too:

Now it is important to note that White may have made progress after 69. Qf5(!), because the Queen on this square assists White's King into f6. A second nasty point is that if White's King gets to g2, then Black has absolutely no checks at all! Hence Black may have to hope that 69...Qg2+ followed by 70...d4 works, or if 70. Kf4 Qf2+ 71. Ke5, that 71...Qb2+ is OK.

Update: There seems to be no reason why 69...Qg2+ 70. K-any d4! shouldn't work---with Black's Queen on g2 this seems the best possible form of the resource that Black could imagine. And 69. Kg5 and Kh5 seem at best to transpose into line C (which admittedly is very tenuous for Black---Kasparov said in his chat today that the analysis was "incredibly delicate", so this being the line may be no surprise. Hence my current belief (Wed. 10/27 late-night) is that 68...Qc2! completely shuts down this major branch for White!---and White has to go looking for wins in line B or line C---where quite possibly they may still be found, either by winning the critical position in line B or by punching a hole in my fragile line-C defense.

In any event, the above lines seem to be the major backbone of White winning plans, and bears analysis. I will post this on both the MSN and 99%'s bulletin boards, but if you have major additional analysis, please also send it to me by e-mail to Hopefully we're getting to the core of this fascinating endgame.


---Ken Regan.


11/20: I've gone back to a simpler copyright notice, although this material will soon be integrated with Irina's and thence covered by SCO's notice.

This article is copyright (c) Kenneth W. Regan, 10/19/99. Permission to reproduce in whole or in part is granted for non-commercial uses.

This file originally had a notice intended to protect against my analysis being merged with someone else's without asking. My musings on this being a better test case of "should chess moves always be free?" than Fischer's quest to copyright game scores (mine here being in main and essence the work of only one person, not two) extended to observing that my analysis as of 10/28 and file released by Kasparov and Alterman later that day had "essentially no overlap of content---they do not give 67. Kg7 or 66. Qg4 and give "!" to a losing move in reply to 66. Kh5. (Indeed, I wonder if any of their lines after 62...Ka2 are primary at all...)" The EGTB answer to my wonderment rates their essential content not already in the SCO FAQ of 11/14 as being: exactly one move, 67. Kf7!