That's a very interesting question. My engine does not find 18. Rd5. Indeed, it maintains that is very equal and up through move 22, after which 23...Qc7 was needed, which my engine evaluates as +0.19 for white at d=22. For example, after 23...Qc7! 24. Bxb5+ Ke7 25. Qxe4 Rb8 26. Bc4 Bc5 27. g5 Bxe3+ fxe3, the following position is reached:

click for larger view

This looks relatively drawish, and white will have few winning chances. Black's position is secure, and while his king is still in the center and exposed, white has no means by which to exploit this. My engine evaluates this as +0.24. Maybe Karpov could win this endgame, but even for him it wouldn't be easy.

Compare this to 18. g5, which is both what my engine recommends and the natural move in this position. My engine evaluates this at +0.63 for white at d=22, and it certainly looks like white has a promising attack. For example, after 18. g5 hxg5 19. hxg5 Rc8 20. Bd5 Qa4 21. g6 fxg6 22. Rxg6 Qxb3 23. Bxb3 Rxg7, the following position arises:

click for larger view

To me, this looks more promising for white. White has a rook on the 7th rank, a queenside pawn majority, and very strongly placed bishops. I would much rather play this position than the previous one. My engine agrees, evaluating this position as +0.72 for white at d=24.

click for larger view

<20.Qxb7 Rc8 21.Qxc8+ Ke7 22.Bc5+ Qxc5 23.Qxc5+ Kf6 24.Qxf8 g6 25.Qxf7+ Kg5 26.Qe7+ Kf4 27.Qxh4 Kf3 28.Kd2 g5 29.Qg3#>

click for larger view

Black can really never castle due to the direct but very strong g5. Say instead of 17...Rf8!? black played 17...0-0, 18. g5 hxg5 19. Bxg5 followed by Qh3 with a crushing attack.

"Sacs" is short for "sacrifices", from the Latin "sacrificāre", meaning "to make holy". The Latin "sacer" ("holy") came from Proto-Indo-European *seh₂k- referring to rituals or treaties. The h₂ represents a mystery consonant that was probably guttural.

"Sack" was a noun before it was a verb; . This noun can be traced to Proto-Germanic *sakkuz, a borrowing from Latin "saccus" (large bag), which was itself a borrowing from Greek "σάκκος" ("sakkos", bag of coarse cloth), which was itself apparently a borrowing from Hebrew "שַׂק" ("śaq", sackcloth).

1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 d6 6.g4 h6 7.Rg1 Be7 8.Be3 Nc6 9.Qe2 Bd7 10.h4 Nxd4 11.Bxd4 e5 12.Be3 Bc6 13.Qd3

<13.Qf3 and the center, with the incoming O-O-O, will be dominated by White. 13.Bg2 and 13.f3 are weaker.>

13...Qa5

<Black starts going wrong. 13...Nh7 was perfectly reasonable with a tiny pull for White. 13...Nd7 is worse: 14.O-O-O Bxh4 15.Qxd6 favours White.>

14.O-O-O Nxe4

<Now White is almost winning. 14...h5 15.g5 Ng4 16.g6 f5 was the only hope, but White has a decent advantage. In the annotations there is 'Black can also play: 14...0-0-0', but then 15.g5 hxg5 16.hxg5 Ne8 17.Qc4 follows and sooner or later it is curtains.>

15.Nxe4 d5 16.Qb3

<16.Nd2 is way better: 16.Nd2 d4 (16...Qxa2 17.Qf5 Bd6 18.Nb3) 17.Nb3 Qc7 (17...Qxa2 18.Qf5 Bd5 [18...dxe3 19.Bc4] 19.Bxd4 Bxb3 20.Qxe5) 18.Bd2 Bxh4 19.Bg2 and after a several precise moves White is essentially winning.>

16...dxe4 17.Bc4 Rf8 18.Rd5

<This is hilarious, as 18.g5 is objectively and definitely better: 18.g5 hxg5 19.hxg5 Rc8 (the suggested 19...g6 is very bad: 20.Rh1 Rd8 21.Rxd8+ Bxd8 22.Kb1 a6 23.Rd1 Qc7 24.a4 (+5.13); 19...Qa4 20.g6 Qxb3 21.gxf7+ Rxf7 22.Bxb3 a6 23.Bxf7+ Kxf7 24.Rh1 and White is singing and dancing.) 20.Bd5 Qa4 21.g6 fxg6 22.Rxg6 Qxb3 23.Bxb3 Rd8 24.Rdg1 Bf6 25.Bxa7.>

18...Bxd5 19.Bxd5 Rd8 20.Bc4 Bb4 21.c3 b5

<According to your analysis, 21...Bd6 22.g5 is better for White. But are we certain that we have really found the best move? 22.Qxb7!>

22.Be2

<22.Bd5 Bd6 23.Bc6+ Ke7 24.Qd5 Qc7 25.Qxe4 Bd6 23.Qd5 is visibly better. The Fish does not assign a huge value to it, but it is pleasant to take a look at White's position. According to your analysis, 23.g5 is good for White, and holds a slight edge. Neither the Fish, nor do I agree: 23...Qa4 24.gxh6 Qxb3 25.axb3 gxh6 26.Bxe4 b4 27.Kc2 Rc8 and where is the breakthrough for White?>

22...Ke7

<In your annotation you write 'This move is probably less than best, but it is not very easy to see from this position, which remains extremely complicated.' However, the Fish thinks that with +0.12 this is a completely reasonable move. Let's see other moves: 23...Qc7 is inferior, as it faces 24.g5 immediately (23...Ke7 24.g5 is not problematic, as 24...Qxc3+ 25.Kb1 Qc8 26.gxh6 gxh6 27.Bxh6 Rg8 28.Bg5+ f6 29.Qxe4 Qe6 controls everything). The immediate 23...Qxc3+ is questionable at the very least least: 24.Kb1 Qc8 (in the annotation there is 24...Qxb2+ 25.Kxb2 Ba3+ 26.Kxa3 Rxd5 27.Bxa7 [27.Bd1 is even better] 27...Ke7, that obviously favours White) 25.Rc1 Qe6 26.Bxb5+ Ke7 27.Qb7+ Rd7 28.Bxd7 Qxd7 29.Qxe4.>

24.Bc5 Bxc5 25.Qxe5+ Kd7 26.Qxc5 Qc7 27.Qf5+ Ke7

<What a position! In the analysis we can read 'After 27...Kc6!? White can quickly gain the upper hand. For example: 27...Kc6 28.Qxb5+ Kd6 29.Qb4+ Qc5 30.Qxe4' - but this is not true. Black needs to play logically, remaining with the King on dark squares. For example: 30...Kc7 31.Bf3 Rd7 32.Qb7+ Kd8 33.Rd1 Rxd1+ 34.Bxd1 Qb6 35.Qd5+ Kc7 36.Bc2 g6 and although White has some practical chances, the position is very close to an objective equality. The text move 27...Ke7 is simply incorrect.>

28.Qxe4+ Kd7 29.Qf5+ Ke7 30.Re1

<Your commentary here is that 'Fritz 8.0 prefers instead the very materialistic Bxb5± or even the dynamic move of g4-g5!' - It seems the Fritz miscalculated something. The line you suggest is 30.g5 Rh8 31.Bxb5 Kf8 32.g6 Qe7 (with deviation 32...f6 33.Re1 or 32...Kg8 33.Ba4 with an incoming Bb3 and winning) 33.Rd1 Rxd1+ 34.Kxd1 Qc7 35.Kc2 with an almost complete Zugzwang. However, 30.g5 is refuted by 30...Rd4 31.Rg3 Rd6 32.Qe5+ Kd8 33.Re3 Re6 34.Qxg7 Rxe3 35.fxe3 Qc5 and it seems everything is handled properly by Black. After 30.g5 Rh8 even better is 31.g6 Kf8 32.Bxb5 a6 33.Bxa6 h5, with a horrible future on the Kingside.>

30...Rd6 31.Bc4+ Kd8 32.Bxb5 a6

<An interesting position. 32...g6 seems to be favourable to 32...a6, as it drives away the Queen first, and the Bishop just afterwards (given 33.Qf3 is played). The difference is that 33.Qf4 fails to be a fantastic move because Black has 33...Kc8, and now if 34.Re4, then 34...Kb8 holds a worse, but possibly defendable position. Note, that here Black spared a tempo by omitting a6 in this variation. This is clearly a madness to calculate OTB.>

33.Ba4 g6

<You state here that '[t]his is absolutely and completely forced, that is, as long as Black wants to continue this game'. The Fish does not agree entirely! 33...Rf6 34.Qd5+ Rd6 35.Qe5 Kc8 36.Qxg7 Rfd8 yields a +3.02 at depth 57, while 33...g6 goes for the even worse 34.Qf4 continuation shown after the next move. 33...h5 is not good, as you state it correctly, but the exact variation of the Fish (yours was 34.gxh5 Rh8 35.Re4 Re6 36.Rb4 Re1+ 37.Kd2 Re6 38.Qf3) is somewhat different: 34.Re4 Rf6 35.Qd5+ Rd6 36.Qa8+ Qc8 37.Qa7 Qc7 38.Qe3 and it is over. If 34.gxh5, then the principal variation is 34...Rf6 35.Rd1+ Rd6 36.Rd5 with the obvious result. 34...Rh8 leads to 35.Re4 Re6, indeed, and it is curtains for Black.>

34.Qf3

<In your annotation you wrote (potentially a quotation from Karpov): 'White had Qe5 or even Qf4!?, as well as the move played'. Let's see what the Fish thinks:

34.Qf4 Kc8 35.Re4 (after the immediate 35.Qxh6 Rfd8 Black has no serious problems) 35...Kb7 36.Qxh6 is very good for White. In fact, it is objectively better than the text-move, 34.Qf3.

But 34.Qe5 is not good! 34...Kc8 35.[something, such as Bb3] Rfd8 controls everything. No Kingside break could be achieved, as 36.g5 hxg5 37.hxg5 Rd2 is equal.>

34...Kc8

<You state here that the alternative 34...Qc5 is also lost. I left the Fish to think about it for hours, and it came up with the strage-looking 35.Qb7 Qc7 36.Qf3 Qc5 37.Bb3 f5 38.Bc2 Qc6 39.Qf4, with a definite advantage for White.>

35.Re7

<You state here that 35.g5 h5 36.Re4 leads to a very large advantage. The Fish thinks differently: 35...hxg5 36.hxg5 Kb8 37.Re4 a5 38.a3 Rfd8 39.Kb1 Rd3 40.Qg2 offers a negligible pull for White. For 35...h5 it suggests, not to my most massive surprise, 36.Re7 Rd1+ 37.Kxd1 Qxe7 38.Qa8+ Kc7 39.Qa7+ Kd6 40.Qxa6+ Ke5 41.Qb5+ Kf4 42.Kc1 Rc8 43.Bd1 with a very serious edge. After your original 35.g5 h5 36.Re4 it displays at depth 50 the 36...Kb8 37.a3 Rb6 38.Be8 variation with a serious advantage; however, your 35.Re4 Kb8 36.Rb4+ Rb6 37.Rxb6+ Qxb6 38.Qf4+ Kc8 39.Bb3 Rd8 fails to 39...f5, where Black has to hold without horrible difficulties.>

35...Rd1+

<From this point there are no questions.>