I got down to under half a minute when I played 37…b4?! This is the move that is supposed to keep winning and drawing chances alive, but it is only digging the hole deeper.
39…g6 39…fxe4 is also drawing here.
40…Kc6?? With 15 seconds on my clock, I played this move thinking that everything is losing, but I at least should have gone into the unknown with 40…fxe4!, which no one could know draws at the speed we were playing it, but at least you can play it knowing that you don’t know yet whether it’s losing or not. 40…fxe4, 41.fxe4 Kd6, 42.Kd4 e5+!, 43.fxe Ke6, 44.Kd3 Kxe5, 45.Ke3 Ke6! draws.
41…Kd6?? Another surprisingly poor choice, since 41…fxe would at least give White the chance of going wrong with 42.Kxe?? which only draws, and 41…Kb5 would at least get into a pawn race where I queen one tempo short and White is required to show some technique. He had plenty of time, over half an hour still on his clock, so there wasn’t much reason for him to falter, but it shows how I am just making guess moves in time-pressure.
I spent around two hours going over pawn endings stemming from this game. Since I knew this game was heading for an endgame, I should have saved more time on the clock for the ending, and even two more minutes more would have given me a decent chance to find the draw, but I should allow the game to end with more time just because – it’s not like I was calculating checkmates against me, which would be different. It was really poor to get into time-pressure since I was better nearly the whole game. I was trying too hard to win, and the endgame was only equal.