It seems that I did not employ databases because I made an inaccurate move while still within the theory on this Volga Gambit (aka Benko). I had Black, and I've played the Volga off and on long enough that I lost a memorable game with it in which the moves were transmitted by postcard.
The game took a unique departure from prior practice with White's fourteenth move, and the critical decision for Black to give up his queen came as a result of White's fifteenth move.
Anonymous Opponent (1805) - Stripes,J (1869) [A57]
1.d4 Nf6 2.Nf3 g6 3.c4 c5 4.d5 b5 5.cxb5 a6 6.b6 Qxb6 7.Nc3 d6 8.Nd2 Nbd7 9.e4 Bg7 10.Be2 0–0 11.Nc4 Qc7 12.Bf4 Nb6 13.Ne3 Bb7?! 14.Rc1 Rad8 15.b4
Black to move
Forcing lines are easier to calculate, while passive defense can be uncomfortable. Moving one of the knights to d7 to secure c5 is worthy of consideration. Sliding the rook over to c8 is as well. Why did the rook move to d8, instead of c8, in the first place? I had been looking at an e-pawn thrust, and d6 needed to be more secure to avoid loss of material. Perhaps it was also a waiting move.
Now, my opponent has attacked. Moreover, he has done so while his king is in the middle, while mine is safely in his castle. The forcing line was attractive.
Since this game ended a couple of days ago, several chess engines have had a go at this position. I let the engine calculate the top three lines until it reaches a search depth of 18 ply.
Houdini 1.5 prefers 15...Rc8 and gives Black a 0.34 pawn advantage.
Stockfish 1.7 also prefers 15...Rc8 and evaluates the position as even. My 15...cxb4 is Stockfish's second choice, but it gives White an advantage of 0.20.
Rybka 4 has the same first two choices as Stockfish, but favors Black at 0.30 for Rc8 and 0.26 for the move played in the game.
Hiarcs 12 prefers 15...Rc8, but favors White at 0.45. Its second choice of 15...Nbd7 favors White at 0.56.
Fritz 11 prefers the move played in the game and sees the position as tilting in Black's favor by nearly one-half pawn: 0.43.
My decision to capture the b-pawn embraced an imbalance. I gave up my queen for a rook and knight, and at least temporarily a pawn. If I hang on to the pawn, the most common point count system has the material equal. Such an imbalance, however, is rarely equal.
After the move played, the next point with branching possibilities comes at White's move 18.
16.Na4 Qxc1 17.Qxc1 Nxa4
White to move
18.Qc7 is the principal alternative.
I do not recall looking at any alternatives. My engines. on the other hand, tend to prefers moves like 18...Nc5, Rb8, Nd7. For some reason, they do not see White then snatching the undefended b-pawn. Rather, 19.e5 is a leading choice. This center strike is a common idea in the Volga Gambit. Without it, the bishop looks funny on f4.
Normally, in the Volga, Black gives up the queenside pawns in order to gain some play with the heavy pieces. In this game, Black has a queenside pawn majority instead.
In my initial post-game analysis with Hiarcs 12, I gave my last move a ! (good move). Hiarcs prefers other moves until after this move is played. Once on the board, and when Hiarcs reaches a search depth of 14 ply, the evaluation rapidly drops from slightly favorable for White to even, and then to favorable for Black. Later evaluation to greater depth generates an assessment that the position is balanced.
White to move
As a consequence of this move, Black gets a strong initiative on the queenside. That White attacked with the b-pawn thrust before castling now matters. Over the next few moves and to the end of the game, I felt good about the coordination of my bishops and rooks.
White should have tried 21.Qb6. Even then, however, Black gets activity for his pieces and a slight advantage. But where else did White go wrong? Was 15.b4 an error?
In my analysis before playing Rd7, I considered the move played as the main line. That 21.Qxa5 was an error did not occur to me. The move that was hard for me to see as meritorious was my 22...Nxe2. When a queen battles several weaker pieces, she needs targets and an open board. Once all the queenside pawns are liquidated, her opportunities become much reduced.
21...Ra8 22.Qxb4 Nxe2! 23.Kxe2 Rxa2+ 24.Ke1
Black to move
Part of the difficulty of conducting my attack is finding the right time for the necessary Rd7-c7. Rd7 was played to drive away the queen, and to protect the bishop. But the rook needs an open file. I could spend more time with the engines exploring move order alternatives. Whether the sequence of my moves was optimal or not, I believe that for the most part, I found the correct moves. They were not particularly difficult to find.
I recall expecting 25.Bg3. It is a common miscalculation on my part to think that an attacked piece must retreat.
I spent some time trying to calculate some maddening variations. I decided to play my move, let my opponent show me where his queen would move, and thereby reduce the work.
White to move
That Rybka prefers 26.Qb1 reveals that Black has a decisive advantage.
26...Bc3+ 27.Kf1 Ba6+ 28.Kg1 Ra1+ 29.Nd1
White moves into a forced checkmate. Even so, after 29.Kf2 Rxh1 Black has an easy win.
Black to move
29...Bd4+ 30.Be3 Bxe3+ 31.Qxe3 Rxd1+ 32.Kf2 Rc2+ 0–1
I recently returned to Red Hot Pawn after several years away. After this game, and a much easier one against a weaker opponent, my rating has climbed to a new peak on that site. When I quit playing there, my RHP rating was nearly 200 higher than my USCF rating. Now it is ~50 lower.