Continuing our endgame analysis of my recent game against my personal kryptonite (and no, Joe, I won't reveal who it is because uttering their name summons them like a dark spirit)...
Here it's White to play, and as noted below White faces a grim defense, being one pawn down and having isolated doubled pawns. I would go so far as to say that if the rooks are taken off the board, White is lost. White must keep and activate at least one rook, trying to create (in combination with the bishop) weaknesses in Black's queenside pawns. In White's favor is Black's kingside pawn structure, which is wretched - if the black king drifts from the kingside then a White rook can bring serious pressure from the g-file. Or White can try to exchange kingside pawns and perhaps open up space for a rook to penetrate.
20. Rab1 Nc5 The knight takes up its perfect outpost
21. Rf4?! Re8 Black sets up to dominate the e-file, whereas White begins to drift without a plan
22. Bc4+ Kf8
23. Rf3?! Re7
24. Rh3 Rae8 White has wasted 3 of the last 4 moves with 'mysterious' rook maneuvers, though to be honest I think this position is almost lost at any rate. I think White may have been planning to break up the Black pawns with a g-pawn sacrifice, but this plan is too slow if also dubious.
25. Rf1 Re4
26. Bd5 R4e5
27. Bb3 NxB? A mistake of overcaution - I didn't like the way the Bishop controlled the vulnerable light squares of Black's K-side, but those square are empty and unneeded. Better is 27....a5 starting the Q-side pawns rolling and gaining tempo on the immobile bishop. I thought the Rook endgame was simply won for Black, and while it is won it is not entirely simple.
28. cxN Re2
29. Rf2 Re1+
30. Rf1 R8c2?! The priniciple of two weaknesses is handy to remember - a single weakness is defensible in the right circumstance, but two weaknesses are almost impossible to defend against a vigorous attack. White's weaknesses are 1) A 1-pawn deficit, 2) Backwards Q-side pawns. My last move was meant to provoke a rook exchange but White does not need to oblige and can take steps toward resolving their second weakness with 31. a4! Better for black was 30...b5! fixing White's weakness and developing a passed pawn.
31. RxR? RxR+ Missing 31. a4, and now the endgame is again lost. Always look for the second resource!
32. Kf2 Ra1
33. a4 Ra2+!
34. Kf1 a6?? Black now needs to capitalize on the progress that has been made. I wanted to play ....a6, ...b5 and ...bxa4, but at any time White can prevent this plan, most easily with 35. c4 and it is not clear how Black can improve his position. With Black's king stuck on the K-side because of the weak g-pawn, White has real chances to draw. The correct plan, which I saw but misevaluated, is to play 34....d5 35. Rd3 c6 36. c4 dxc4 where I feared the penetration of White's rook onto the 7th and a pawn massacre. A closer look reveals of course that this position is winning for Black but I did not look deeply enough and chose a 'safer' continuation, one which should safely draw as it turns out.
35. Re3 (c4!) b5? (d5 can still be played)
36. Re4? Rc2! White's last chance was the simple axb when the 3 vs 2 pawn structure on the Q-side may yet be good enough for a draw with best play. Now the loss can be forced.
37. Re3?! PxP
38. PxP Kf7?! A delay, on the next move I find a winning rook maneuver
39. h3 Rd2! The rook will go to c5 where White can no longer defend all their pawns, and with a two pawn deficit the game cannot be saved.
40. Kg1 Rd5
41. Rf3 Rc5
On the next move Rc4 will win either the a-pawn or c-pawn, all can't be defended. The game continued to the 56th move as my opponent played for a trap - What should Black really not do in this position? Fortunately I wasn't inebriated at this point.