I knew that I could have played 16…BxN, but had decided that he surely wouldn’t agree to a draw, so I’ll play for a win. Unfortunately, I made this decision in time-pressure.
When he played 21.Nh4, my original idea had been to counter this move with 21…Bg5 (with ..f6 to follow) but somehow played ..Qd7 (because my mind was on where to put the queen) first and was then stuck with the chore of defending ..f7.
I flagged just barely, but he had achieved a dominating winning position by then.
I didn’t notice he had the Qf6 move winning a piece until move 23, when I would have replied 23…Rh8-c8, sacking the bishop. I didn’t see Qf6 a move earlier, but now I am in time-pressure and he is playing quickly, and is looking at my clock now and then during the game. If I had played ..Bg5 when I had meant to, then his g4 could have been met by ..Bf4. I had intended to sac my h-pawn for some counterplay, but had bungled the whole affair starting with my ..Qd7 move.
Paul showed why he is such a great endgame player, just keeps building the position and doesn’t let out the tension with a NxB trade. He had over 50 minutes at the end of the game. This is one thing I find unfortunate about G/90, the chance of one endgame mistake (not playing ..Bg5 at the crucial moment) and it becomes an entirely different endgame, not even counting the piece he could have taken.
The big mistake I made in this game was a psychological and time-pressure one. In time-pressure, I decided I liked my position enough to play for a win instead of a draw, even though I knew as soon he had played d5, that strategically it would be best for me to trade my bad bishop for a knight, but I tried for something tactically instead when I had a great big nothing of a position, and a draw would have been a great result against him as Black from this opening position.