9.Qd3. I did spend a lot of time considering 9.0-0, which is the more critical line.
10.d5?! Again, castling was better.
12…Ng6?? I missed 13.Nxd5!
14…Bf5?? This move could have been played two moves ago, when I would have responded Qe3 with an equal position. Of course, I’d been waiting for this move, and it’s surprise value was history by the time he played it. Simply 14…0-0 was best.
15…Ng3?? Doubling-down on a poor position. I was expecting to see 15…Be6 +-. 15…Ne7? fails to 16.Nxf7 KxN, 17.QxNe4 (works due to the double pin).
18…fxe6. It looked a bit more complicated to see 18…f6 when I was going to respond with 19.Bxd Qe7, and then Nxh7 wins another pawn. The last part is something I had seen, but OTB it’s difficult to remember and keep everything in order. When he got out of the restroom he played this move instantly, and then resigned within a minute or two seeing the continuation that I had seen, namely 19.Qxe6+ Kh8, 20.Bxd5 Qe7 (else …Rc8, 21.Qg8+ RxQ, 22.Nf7 is smothered mate), 21.BxRa8. He pointed out this continuation excitedly, but I refrained from playing it more quickly because it was a committal and decisive sequence, and he had been showing a lot of will to win in his moves.
7.fxe 7.0-0 is better, allowing …exd4, Nxd4.
10…Ne5?! 10…0-0 is the straightforward way to play.
11…NxNf3+? 11….Ng6? fails to 12.e5, but …Nfg4! and if 12.Qxg Ng6 threatens to trap the queen with …Bf8 (and Black can take on h2), so Be3-g1 (to give the queen the e3 square) +=.
14.Qg3 Simply not ambitious enough. 14.Qe3! was the way to avoid the queen trade, while Black is still seeking shelter for his king.
11.Qg5! After a long think, I found this idea. It was not easy to spot this move. Black’s idea was to take on f3, then h2 and control that dark diagonal with his bishop, which could rest on f4, hence my concern to find a strong reply to this idea.
27…f6. After great endgame play by Black, I felt that this move gave me a new lease on trying to play for a win. The computer likes this move, but now he has to play this endgame very well, whereas there was no real way for White to make progress here I showed him in the endgame, and he proved it post-mortem. Well, …f6 happens sooner or later, according to Houdini and I can see why after some analysis with it.
In the endgame, I was worried that Black would turn it into a race, when my slower extra pawn could turn it into a draw. As Neal started moving his king further toward my pawns, which was the start of the wrong idea for him, my expectations for the outcome of this game began to soar. He was guarding a backwards pawn on the queenside and had no space over there, and was down a pawn there.
35…Rd2 This is where he started to go wrong. I expected 35.Ke7, and I wanted to play 35…Rf5 instead of 35…Rb8 because he appeared to have had the fast outside passer. White actually wins this endgame by a single tempo because the a8 queen can take the h1 queen; e.g.; 35… Ke7 36. Rb8 Rxf3 37. Rxb7+ Kf8 38. a4 Rg3 39. Ra7 Rxg4 40. Rxa6 h5 41. Ra8+ Kf7
42. a5 h4 43. Rh8 g5 44. a6 Kg7 45. Rb8 h3 46. a7 h2
Once you see that White is winning in that line above, I want to backtrack one move. When I played 35…Rf8 it was because I had looked at this (best) line: 35. Rb8 b5 36.
Rb6+ Ke5 37. Rxa6 Kf4 38. Rg6 Kxf3 39. Rxg7 Rg2 40. Rh7 Rh2 41. g5 h5 42. g6, and after move 37.Kf4 I hadn’t bothered to look any further, feeling it was unclear, but it’s clear as day once you follow that entire line or up to the simple idea behind 41.g5 that White is winning.
47…Re5+? I was expecting 47…Ke6, but I knew when I played 47.Rb8 that 48.Rc8 followed by 49.Rc7 would be decisive. Also, I sensed that 47.Kd5 was best and was tempted to play it, but 47.Rb8 was more immediately clear-cut with less to consider, and I had maybe 3 minutes left, so I didn’t want to have the chance to confuse myself.
50.Rc7 50.c5+ also works because after the rook trade …Ke6, 51.Re8+ Kd5, 52.RxR I can play the c6 pawn-promotion sac, but I hadn’t seen this idea yet (51.c6 bxc6 52.Rxc6 Rb5+, 53.Ka4 will also win).
Going over this game has given me the sense of how difficult it is to defend against a += position. In fact, it is likely to be one of the hardest things to do in chess between two strong players. He was defending better than I realized at points, but his task was also a lot larger than I had assumed.
13.c5 I spent a long time debating between this move and 14.b3, but for better drawing chances, I wanted to play a more forcing line, which wasn’t a good reason to do it.
20.a4 f4 is a stronger idea, because it will give me a possibility of playing g5 in response to the …Bf6, …Nxd4 idea. I wanted to play Bd3, but was concerned about …Bf6, threatening to take the d4 pawn.
23.Rhf1? Loses the exchange. I thought about this game the next day, as I was getting ready to take a shower and thought I should have played the other rook to this square, as it was unfortunate that I didn’t have a check after Be2, in order to get out of losing the exchange.
24.Bxf4 Houdini gives Rxf4 as much better, and it’s easy to see why, keeping the bishop pair on the board. There are a number of errors that we both made after this point, me making the most, but despite the differences in evals between moves this ending should be won for Black.