Fritz indicated the following evaluation: 25.h4 Ng4+ 26.Nxg4 hxg4 27.f4: (-.95) (21 ply) 27...Bg6 28.Qe1 b6, or (-.93) (21 ply) 27...Re8 28.Nb5 Bf8. White may be able to hold this position as I did not find a line that increased Black's advantage.

At move 25, White had a number of moves that give him a good or reasonable chances to hold the position.

Fritz strongly preferred three choices that were all evaluated as near equal or with a very minimal advantage for Black: 25.Qh1, 25.Nb5, or 25.Qf3.

In my last post, I did not correctly note Alekhine's opinion of 25.Qh1. In his notes Alekhine stated that 25.Qh1, <... would have been ineffective if Black had been merely content to protect his pawn simply by means of 25...Rd8.> As I noted in my last post, Fritz indicated 25.Qh1 Rd8 would lead to an equal position.

Alekhine then stated that Black's next move 25...h4, was more energetic and to the point. Kasparov gave 25...h4 an exclamation mark. However, Fritz verified that the game continuation 25...h4 26.Nxd5 hxg3+ 27.fxg3 Nxd5 28.Bxd5 Bf6, leads to equality, which White could have maintained by 29.Qf3!.

I believe you have found an improvement. I note that none of the commentators, Alekhine, Kasparov or Soltis, made any comment regarding 32...Rh5.

Fritz indicated White's best reply to 32...Rh5 was either 33.Nc6 or 33.e3; 33.g4 was Fritz's 3rd choice.

In your variation, 32...Rh5 33.g4 Qd6+ 34.Kg2 Re5 35.e4 Re8, Fritz indicates Black is winning.

After either 33.Nc6 or 33.e3 White still has a very difficult position. I believe Black may be winning in both lines.

Here is Fritz's analysis: (-76) (25 ply) 33.Nc6 Qe8 34.Qc8 Qxc8 35.Bxc8 Bxb2 36.Nxa5, (36.bxa5 Rc5 37.Bb7 Rc2 38.a6 almost saved the day for White, but Black can stop the a-pawn) 36...Kf8.

In the second variation: (-.83) (25 ply) 33.e3 axb4 34.Bc4 Be5 35.Kg2 Bxg3 36.Ne6 Qh4 37.Qc8+ Kh7 38.Nf8+ Kh6 39.Nxg6 Kxg6.

I have found it. The variation is 25.h4 Ng4+ 26.Nxg4 hxg4 27.f4 Bg6 28.Qe1 b6, or 27...Re8 28.Nb5 Bf8.

There is an alternative to both 28 Qe1( after 27...Bg6) and 28 Nb5 ( after 27...Re8) and that is the central advance 28 e4.

On 27...Bg6 28 e4 de 29 de f6 30 Qe1 b6 31 e5 fe 31 fe the e pawn may become a weapon.

e3 and Nc6 both lead to winning positions for white. Here's a dump of my findings including 33. g4.

32...Rh5 33. e3

(33. g4 Qd6+ 34. Kg2 Re5 35. e4 Re8 36. Qb5 Rb8 37. Qc6 Rd8 38. bxa5 Bxd4 39. Qxd6 Rxd6 40. Bxd4 Rxa6 41. Bb6)

<(33. Nc6 Qe8 34. Qc8 Qxc8 35. Bxc8 Bxb2 36. Nxa5 Kf8 37. Bg4

(37. Nc6 f6 38. Bg4 Rg5 39. b5

(39. Bf3 Be8 40. Kg2 Rf5 41. Be4 Bxc6 42. Bxc6)

39...Be8 40. d4

(40. Na5 f5 41. Nc4 fxg4 42. Nxb2 Rf5 43. hxg4 Rf2+ 44. Kh3 Rxe2)

(40. Kg2 Bxc6+ 41. bxc6 Rc5 42. Bd7 Ke7 43. Kf2 Kd6)

40...f5 41. Bf3 Bxc6 42. bxc6 Ke7 43. h4

(43. e3 Bc1 44. d5 Kd6 45. e4 fxe4 46. Bxe4 Re5 47. Bf3)

43...Rg6 44. h5 Rg5 45. e3 Bc3 46. Kg2 Be1 47. Kf1 Rxg3 48. Ke2 Kd6)

37...Rh6 38. b5 f5 39. Bf3 Bc3 40. Nc4 Bf7 41. Kg2 Ke7 42. Ne3

(42. g4 fxg4 43. hxg4 Bxc4 44. dxc4 Kd6 45. Bd5

(45. b6 Re6 46. b7 Kc7))

42...Be6 43. Nc4 g5 44. Bc6 Bxc4 45. dxc4 Kd6 46. e4 fxe4 47. Bxe4 Kc5 48. Bd5 Be1 49. g4 Bd2)>

In the variations I have given, White plays e5 and not f5.

In the first variation above, White plays ( after 25.h4 Ng4+ 26.Nxg4 hxg4 27.f4 Bg6 28.e4 dxe4 29.dxe4 (-1.38) (23 ply) 29...Bc5 ) 30 e5 and not 30 f5.

Alekhine writes,

"With correct positional judgment White seeks his salvation in this sacrifice by which he can dispose of one of Black's center pawns. 17 Qa2, for instance, would have been apparently less profitable on account of 17 ... Na6.

Rybka does not agree with Alekhine:

Analysis by Rybka 2.2n2 mp 32-bit :

1. = (-0.18): 1.N3d2 Na6 2.Rd1 Nb4 3.Rcc1 Bc5 4.Ne3 Bxe3 5.fxe3 Ng4 6.Nf1 Qd6 7.h3 Nf6

2. = (-0.22): 1.Re1 e4 2.Bxf6 gxf6 3.Ne3 exf3 4.Nxd5 Qe5 5.Nxf6+ Kh8 6.Nxe8 Qxa1 7.Rxa1 Rxe8

3. = (-0.23): 1.Qa2 Na6 2.Qa1 d4 3.N1d2 Nb4 4.Rc4 Bf5 5.Ba3 Rad8 6.Qb1 Bg4 7.h3 Be6

4. = (-0.24): 1.Rd1 Rac8 2.Ne3 Nxb3 3.Qb1 Nc5 4.Bh3 Rc6 5.Qa2 Qc7 6.Rdc1 Ne6 7.Qb3 d4

5. ³ (-0.27): 1.N1d2 e4 2.Ne1 exd3 3.exd3 Nxd3 4.Nxd3 Bxd3 5.Rc3 Bf5 6.Re3 Be6 7.Nf3 Bc5

6. ³ (-0.32): 1.Qb1 Nxb3 2.Rd1 d4 3.Qa2 Nc5 4.N1d2 Na6 5.Qb3 Nb4 6.Rcc1 Bf5 7.Nc4 Rec8

7. ³ (-0.33): 1.Rxc5 Bxc5 2.Nxe5 Bd6 3.d4 Qe6 4.Ne3 Rec8 5.Rxc8+ Rxc8 6.Qf1 Rc7 7.Bh3 Qe7

8. ³ (-0.47): 1.Bxe5 Bxe5 2.Qxe5 Qxe5 3.Nxe5 Nxb3 4.f4 Nxc1 5.Rxc1 Bf5 6.Ne3 Be6 7.Rb1 Rac8

9. µ (-0.78): 1.Rc3 d4 2.Rxc5 Bxc5 3.N1d2 Rad8 4.Bf1 Rd5 5.Nc4 Qd8 6.Rd1 b6 7.Bh3 e4

What is the depth of the search? I assume that for such a long list, the depth wasn't much more than 15.

Thank you for the prompt reply.

One clue to the value of these numbers is the depth at the last move. Given a depth of 20, with 14 ply lines the depth is 6 at the last move. If you let the engine sit on the last move for a while, you will usually find a different result.

This is the main reason that infinite analysis has little value in The World games. Even RV's monster searches at depths of 30 or more are used only as baselines for more thorough analysis.

I understand the temptation to post these numbers. If you look earlier in this page, you'll see me dumping a great load of unannotated numbers.

However I also posted some commented lines. For example, here's one of my posts from 7/22/08: <20...Bd6 (0.26/21) looks bad or at least pointless. Better is 20...Qd6 21. Kh2 (21. d4 Bb4 22. Rxc8 Rxc8 23. Qd1 Ne4 (-1.20/20)) Bxe3 22. Rxc8 Rxc8 23. fxe3 (-0.81/20)>

Notice that the eval and depth are given after the move where the analysis took place. Still I think I should have explained why Bd6 was pointless and not just given the numbers as "proof".

Here is an example of where I added commentary to Rybka evaluations:

Keres vs Petrosian, 1959

Note that <JG27Pyth> asked, "what does Alekhine say," and "what does a search engine show?"

I thought it was succinct to answer those two questions.