How Not To Cope...

I was aware of the fact, Texaspete (Silman's Complete Endgame Course! ) but thank you for the demonstration anyway! 

After 51...Rxg4, there are 7 pieces (including the kings), so the position isn't solved by that site, and It only becomes a solvable position after the blunder 55...Rf4?. I didn't see any special need to talk about a specific blunder or check its effects that late in the game in my post. Maybe that was a bit neglegent on my part, since I tend to give much more complete analysis. I simply didn't look at the position for many moves after that. Sorry about that. (not sure I would have seen a 6 piece tablebase draw (well...I wouldn't probably. Definitely not without very extensive analysis for sure), but at least I would have seen it becomes more complicated)

I'll be very surprised if after 51...Rxg4, white has a prayer. 55...Rf4 was just... a blunder.

I hold to my assertion that the position is won for black after 51...Rxg4.

Black's plan for winning that position seems to be a 3 step process:

1) Win one of the white pawns. Easily done with 55...Ke4 instead of 55...Rf4?, where on everything white plays, his d pawn is toast.

say, 56.Bb8 Rxd5 57.Kb6 (I don't see anything better for white).

And now it comes the time for phase 2:

2) Win the other pawn by bringing the black king to d5 and overloading the white bishop with its duties of protecting the c pawn and blockading the h pawn. (There may be zugzwang possibilities here too, with white having to move either his king or bishop from the critical defending squares, or the pawn itself, which in many positions could lead to its loss. Continuation:

That phase would go something like that:

57. ...Rd8 (problem is the immediate Rd2 c6 leads to an immediate draw, so need to harass the bishop a bit).

If white blocks black's rook from returning (which is a big part of running the king and bishop out of moves, since the rook has more scope from behind the pawn to make threats while supporting the h pawn if necessary) with 58.Bd6

just 58...Kd5 seems sufficient to me for this phase. Rook will have time to return to its proper place behind the lines anyway, because c6 is never possible without losing: 59.c6 Rxd6 is easy, but it also is easy after 59.Kc7 Rg8 .Again here 60.c6? Rg7+ 0-1.

I am willing to give more comprehensive analysis to support (or disprove) my judgement of this endgame, if required, but this post is already too long.

I can see no way to hold on to the c pawn for long for white.

After stage 2 is complete, stage 3 is trivial: win the white bishop for the blakc h pawn. over.

That would be my approach to this endgame, and I believe that's the only way humans can tackle such seemingly complicated edngames: by training themselves to find constructive plans, even they are take a while to accomplish.

That's for humans...now, IF I were a tablebase...

As I hinted at the end of the previous message, after black plays 55...Ke4 (instead  of 55...Rf4?) and wins the first of the 2 white pawns, you get a 6 piece tablebase, which just states a win (mate)  for white in 18 moves after that, just in case you haven't noticed:)

Humans aren't tablebases. but the more practical plan based approach I outlined in my previous post can work wonders in seemingly very complicated positions.

51...Rxg4 just had to be losing. I knew that when looking at that for a few seconds (a bit more perhaps...). The specifics of how to win do require not to make big errors, and to keep making rational decisions all the way through.

That's true for almost any position. Most wining ones included.