Material: Up a P, “but all R endgames are drawn”. The Black K has 1 legal move, so mate threats are in the air, if White plugs the flight square at g5. Primary candidates should therefore plug g5. An examination of checks reveals a sacrificial candidate, 56.g5+, which calculation refutes.

Candidates (56.): g5+, Kf4

56.Kf4(threatening 57.Rh3#)

56…g4+ 57.Kf5 (again threatening 58.Rh3#)

57…Ra5+ [or Rc5+] 58.e5 (still threatening 59.Rh3#)

58…Rxe5+ 59.Kxf6

But I liked 56 Re6?! over the board, so I let the engine (Rybka_v2.1c.demo.w32:) play the position quickly a few times.

It always produced a two-pawn advantage for white. White also was able to avoid doubled pawns, so it was a table base win every time when six pieces were left on the board.

The right idea is 56.Kf4, which is very attractive since it immediately threatens Rh3#, so 56...g5+ is practically forced.

Then, what else but 57.Kf5, getting out of check and keeping the mate threat alive? Again, checking with the rook (at a4 or c5, but ...Ra5+ is probably better) is practically forced.

White has nothing besides 58.e5 to block check (58.Kxf6 just exposes our king to checks, like 58...Rf8+, repelling the attack), and black, needing to keep us in check, yet again has no choice but 58...Rxe5+.

The key to the position seems fairly clear. White is a pawn up and the black king is semi-trapped on the h file. A mating attack seems to be called for.

I spent an inordinate amount of time trying to make 56. g5+ work.

Then I dived into the murky depths of 56. Rbb7.

Finally I gave up and peeked at the game continuation.

Of course, 56. Kf4! It is actually the most forcing move on the board because it is the only one that threatens mate on the move. It simultaneously stalemates the black king and clears a path for the rook check on the h file.

<Slurpeeman>, 56. g5?? is a bad move that leads to a draw. It isn't a winning move, <Slurpeeman>. Just take a look at Zooter's and Oliveira's correct analysis and commentary of 56. g5+??.

56. Kf4!! is the only move for White's win and beautifully sets up the mating net. It can lead to the checkmate or winning the Black rook.

In my opinion, 56. g5+?? could help Black escape from the disaster.

I completely agree with Zooter's, Kevin86's, Dzechiel's and Patzer2's commentary. The simple move for White is 56. Kf4!! .

Not giving myself an adjustment of my week's score. I want a fair analysis of the line I'm about to give, and although it doesn't win as easily as the played line (which I badly enough analised up to 58....Rxe5†..., but I couldn't find the followup and went back), I think it wins.

-ML-
<56.Kf4> Is forcing almost all the way. White is threatening Rh3‡ all the way, and Black has solarity moves in either the mainline, or giving away a Rook. Those I can leave out.

<56....g5†> Rxe4† just loses a Rook, anything else meets Rh3‡.

<57.Kf5> Keeping the threat, and Black needs to check. Using the backrank Rook loses to:

/A\
<57....Rc5† 58.Kxf6 Ra6† (Rc6† 59.Kf7) 59.Kf7> Either a Rook has to fall or mate on h3, as was the last time. But now, there's no other move.

/B\
<57....Ra5† 58.Kxf6> Now I know this is not a terminator, but I think it wins easy.

<58....Rf8†> Only move, Ra6† meets Kf7 again as in variation AA. Now White seems to lose both the initiative and the mate threat, but I think not.

<59.Rf7!> Surely, the game's end is way prettier than mine, but I think White can wrap this up simply: the mate threat is STILL there, and the White King has to be moved by Black to avoid it, once again with check. Trading now leads to the line below, so it's unnecessary giving it standalone.

<59....Ra6† 60.Ke7> Trying not to trade but win the f7-Rook, but White can simply step aside and continue to defend it. Now, Rb6‡, Kxf8 and Rxf8 are threatened, and Black must suffice with a trade.

<60....Rxf7† 61.Kxf7> Still the mate threat, even now...

<61....Ra7† 62.Kf6 Ra6† 63.Ke5> All forced, and White can, and SHOULD win this. However, I agree it's not as easy as game line.

to