click for larger view

Komodo-9.42-64bit:

Black threatens 31... exd3 and the lsb is in danger.

The a1-h8 diagonal and the g-file are open. This leads to consider 31.Rg1:

A) 31... exd3 32.Rxg7+ Kh8 33.Rxd7+

A.1) 33... Kg8 34.Rg1+ Qg5 35.Rg7+ (probably better than 35.Rxg5+) 35... Qxg7 (35... Kh8 36.Rxg5+ wins) 36.Rxg7+ Kh8 37.Rg5+ Kh7 38.Bxf5+ Rxf5 39.Rxf5 + - [B+P].

A.2) 33... Rf6

A.2.a) 34.Bxf5 Rf7 35.Rh7+ Kg8 36.Rg1+ Qg5 37.Rxg5+ hxg5 38.Bxf6 Rxf6 39.Bxd3 Nxb4 40.Bb1 Rxf2 41.Rxb7 Nxd5 42.cxd5 Rxd2 43.Bf5 + - [B] (43... Rxd5 44.Be6+).

A.2.b) 34.Rf7 Qxh3 35.Bxf6+ Kg8 36.Rg7+ Kf8 37.Rfg1 and the threat Bc3, Rg8+, R1g7# seems to win.

B) 31... Rf7 32.Qg3 Qxg3 33.Rxg3 Kf8 (due to 34.Rdg1) 34.Bxg7+ Rxg7 35.Rxg7 Kxg7 36.Bxf5 Nf6 37.Ra1 with an extra pawn and the better ending.

C) 31... Ne5 32.Qg3 Qxg3 (32... Qf6 33.Bxf5 wins a pawn) 33.Rxg3 with some edge for White.

However, if Black declines the capture of the poisoned Queen with the strongest reply 31...Ne5 then the win becomes much more difficult for White as indicated in <RV>'s 37 depth analysis with Komodo 9.42 x 64.

In the game continuation, after 31. Rg1 Nf6, the computers indicate a stronger move than 32. Qg3 (+2.19 @ 27 depth, Komodo 9.02) is the swashbuckling 32. Rxg7+! (diagram below)

click for larger view

when play might continue 32...Kxg7 33. Rg1+ Kh8 (33... Kf7 34. Qd4 Nc7 35. Bxf5 Ke8 36. Qb6 ) 34. Qd4 Nxb4 35. Rg3 f4 36. Qxe4 (+6.31 @ 18 depth, Deep Fritz 15).

<31.Rg1> develops a R on the open file facing black's K, while tactically buying time. Black can't take the Q: 31...exd3? 32.Rxg7+ Kh8 33.Rxd7+ (not 33.Rg4+? Qf6) Rf7 (not 33...Kg8? Rdg1!+) 34.Rf7 Qxh3 35.Rg1 (threat Bxf6#) and black has no good answer.

That's about as far as I got. I figured black would reply with <31...Ne5>, blocking the DSB's attack on g7 while developing the N on a strong square, protected by a P.

After this, no great tactics jumped out at me. White has to save the Q (Qe3 or Qf1 to maintain defense of Bh3). I sort of prefer <32.Qf1> where my Q seems less exposed to remove-defender tactics, which left me with this position:

click for larger view

I think white is better, mostly due to a safer K and better piece coordination, but I'm not seeing some definite winning shot.

i think andrew soltis documented that most very strong players claim they never really calculate more than 3 or 4 moves ahead, and along only a few lines, in any position.

from any problem position, if more than 4 moves are required and unless the whole sequence of moves is forced, you don't have a "puzzle" to be solved, just a real game to analyze.

Good question. I believe that the objective of these puzzles is "find the best move".

Of course, the idea is that you not only find it, but provide justification for why it is best.

In this case, I think if you just find 31.Rg1!, and see why it is superior to other moves (really, just 31.Qe3), then I think you've done it. The game as played doesn't reflect "best play", so you shouldn't be obligated to expect or analyze that.

I saw a lot of whites moves because I've been studying a lot lately. Missed 38.Bxb4 because I wrongfully assumed the bishop must stay on a1-h8. But OTB I may have thought, "I'm screwed if I don't take the pawn." Never know.