FWIW, here is how Stockfish 5 considers Anand's 17...Qa1 to be Black's top move after 17.Qg4. Here is how it evaluates the position at d=41; it evaluates other Black alternative moves as much worse, [+2.24] after 17...Be7 and [+4.03] after 17...Kf7.

[+0.81] 17...Qxa1 18.Bxe6 Rd8 19.Bh6 Qb2 (here it deviates from Anand's 19...Qc3) 20.Bxg7 Qe2 21.Bxh8 Qxg4 22.Bxg4

click for larger view

Black has managed to exchange queens and blunt White's attack, but it is a pawn down and White has the 2 bishops on an open board, so the game may be effectively lost at this point. Stockfish's line continues:

22...Be7 23.e6 Rd5 24.Rd1 c5 25.Kf1 Bg5 26.Bf3 Rd6 (maybe 26...Rd8 is better, not losing a tempo and surrendering the Pc5 after the exchanges that follow) 27.Be5 Rxe6 28.Bxd4 cxd4 29.Rxd4 Ke7 (the BOC might make this a difficult game to win as White for us mere mortals) 30.Bd5 Rd6 31.g3 a5 32.Kg2 Bf6 33.Rd2 (I would have thought that 33.Rd1 to allow unpinning the bishop by 34.Bf3 or 34.Bb3 might be better, but White certainly doesn't want to exchange rooks and reach a BOC ending) 33...b4 34.f4 Bc3 35.Rd3 a4 (and here 35...b4 might be better, but it results in just a transposition of moves, see below) 36.Kf3 Rd8 37.Ke4

click for larger view

White is a passed pawn up with the more active king so clearly it has the advantage, but whether it is enough to win is another issue. From this position Stockfish evaluates the resulting position at [+1.49], d=38 after 37...Rc8 38.g4 b4 39.g5 h6 40.h4 Rh8 41.a3 hxg5 (maybe 41...Be1 intending to follow anything other than 42.axb4 with 42...b3 should be considered, although the side behind in material is usually better off if they can exchange pawns) 42.hxg5 Bb2 (but this can't be best since Black cannot hope to advance the a-pawn beyond a3) 43.axb4 Rb8 44.f5 Rxb4+ 45.Kf3 Rb6 46.Re3+ Kf8 47.Re4 Rd6 48.Be6 Rd3+ 49.Kg4 Kg7 (49...Rd4 first with a goal of exchanging rooks looks attractive but fails after 50.Rxd4 Bxd4 51.f6 (shutting off Black's bishop from the k-side) 51...a3 52.Kf5 a2 (horizon effect move, White's g-pawn cannot be stopped from queening) 53.Bxa2 Ke8 54.g6. So 49...Kg7 was necessary to keep Black's king in front of White's pawns) 50.Rxa4 Rd4+ (Black finally succeeds in his goal to exchange rooks but it cost Black its a-pawn) 51.Rxd4 Bxd4 (however, this is a draw per the Nalimov tablebases but Stockfish, which in its "official" version does not use tablebases, could not see this and so went for this position which it evaluated as very advantageous for White. Another example why I think it is <essential> in endgame analyses to select engines that use tablebases).

In contrast, when using a Stockfish 4-based version and 5-piece Syzygy tablebases, Stockfish evaluated the position after 37.Ke4 at [+1.03], d=36 after 37...b4 38.g4 h6 39.h4 Bb2 40.g5 b3 41.axb3 a3 42.b4 hxg5 43.hxg5 Rb8 44.Kf3 Rxb4 45.Re3+ Kd6 (this looks almost fatal; as the Stockfish 5 line showed, Black's king is needed on the k-side to slow down White's pawns so 45...Kf8 was likely better) 46.Ba2 Bc1 47.Re6+ Kc5 48.Rf6 Rd4 49.Rf8 Rd3+ 50.Ke4 Rd4+ 51.Ke5 Bb2 52.Rc8+ Kb5 53.Kf5 Rd2 54.Be6 Rg2 55.Bd5 Rd2 56.Bc4+ Kb4 57.Bf7 Rf2 (this looks to me like a loss of time) 58.g6 Re2 59.Rc7 Re1 60.Be6 Rg1 61.Rc4+ Kb5 62.Rc8 Bg7 63.Bf7 Kb4 64.Rc7 Bb2

click for larger view

That was a surprise to me. I would have suspected that Stockfish + tablebases would reach either a substantially different evaluation or a different position (or both) when compared with Stockfish without tablebases. So maybe using tablebases doesn't help in evaluating this position, and I should note that at d=36 Stockfish 5 without tablebases evaluated the position similarly at [+1.06] but had a completely different top line.

And restarting the analysis at this position doesn't help; Stockfish 4 + Syzygy tablebases still evaluates the resulting position at [+0.96], d=42 after 65.Rc4+ Kb5 66.Rc8 Re1 67.Rg8 Rg1 68.Ke4 Kb6 69.Ra8 Re1+ 70.Kf3 Rf1+ 71.Kg4 Rg1+ 72.Kf5 Kb5 73.Ke4 Re1+ 74.Kd5 Rd1+ 75.Ke6 Rf1 76.Rb8+ Kc5 77.Kf5 Rg1

click for larger view

And this is headed for a draw by repetition since the bishops hold the opposing pawns at bay, but Stockfish can't see this, even when using tablebases. Maybe with 6-piece tablebases it could.

Instead, sliding forward, after 22.Bxg4 Stockfish evaluates the resulting position as follows:

[+0.83], d=32: 22...Be7 23.e6 Bb4 (so why not 22...Bb4?) 24.Bf6 Be7 (even stranger, if 22...Bb4 23.Bf6 Be7 a similar position is reached but with White's pawn on e5 instead of the seemingly more advantageous e6, so Black has gained a tempo since e5-e6 is likely to be needed) 25.Be5 c5 26.f4 Rd5 27.Kf2 Bd6 28.Ke3 Bxe5 29.Ke4 Rd8 30.fxe5 (strange, why does Stockfish accept doubled and isolated pawns?) 30...Nc6 31.Bh5+ Ke7 32.Rf7+ Kxe6 33.Rf6+ Kd7 (33...Ke7 seems better since 34.Rf7+ Ke6 clearly leads White nowhere) 34.Bg4+ Kc7 35.Rf7+ Kb6 36.Bd7 c4 37.e6 Kc7 38.g3 b4 39.Rxh7 b3 40.axb3 cxb3

click for larger view

A very different best line compared with Stockfish 5's best line. Hard to see how Black's q-side pawns can be stopped since White's B+P block White's rook from getting over to the a-wide. But restarting the analysis at this point causes Stockfish 5 to evaluate the position at [0.00], d=41 after 41.Rf7 a5 42.Rf1 a4 43.e7 Nxe7 44.Bxa4 b2 45.Bb3 Rd2 46.h4 Nf5 (cute, if 47.Rxf5, the Black b-pawn queens and if 47.Rxf5 then 47...Rf2+! 48.Rxf2 b8=Q, so Black picks up one of White's k-side pawns) 47.Rb1 Nxg3+ 48.Ke3 Rh2 49.Kd3 Nf5 (the simple 49...Rxh4 50.Rxb2 is a Nalimov tablebase draw) 50.Bc2 Rh3+ 51.Kd2 Nd6 52.Bd3 Rh2+ 53.Kc3 Rxh4 54.Rxb2 (another tablebase draw) 54...Ne4+ 55.Bxe4 Rxe4 and K+R vs. K+R is clearly a draw.

I think he was.

http://www.kasparov.com/chess-arche...

Another pun that could have been used for this game...

<Slammin' the door>, or even,

Kasparov vs Anand, 1995 (kibitz #37)

but let me cite some comments from Nunn/Burgress/Emms' <The Mammoth Book>:

11.Ng5!?

<This staggering move, an invention of Karpov's trainer Zeitsev, caused a sensation in game 10 of the Karpov--Korchnoi WC match in Baguio City in 1978>

14.Bc2!

<Tal suggested this idea at the time of the Baguio match, without giving any further analysis. [...] The 6th game saw Kasparov following Karpov's recipe from Baguio with 14.Nf3, but Anand improved on Korchnoi's play and after 14...O-O-O! (etc.)>

14....Qxc3

<Anand spent just four minutes on this move, suggesting that he was still following his pre-match prep...>

15.Nb3

Daniel King reports in his book of the match (page 72) that Kramnik had left New York before this game was played and as far as I can see, he never came back.