It seems to me that your plan looked fine.
With 50. Kxa6! Ke5 51.Kb6! Kd4 52.Kb5, you could have won in the game concretely as it was.
Black maybe could have tried closing off the queenside with ...a5 and hoping that he will also be able to close the kingside so no king could penetrate ... but possibly this second hope (of being able to close also the kingside) would not have come true.
Thanks for the reply. This is the hardest loss i've had in awhile as I was so sure it was won, and that I knew how to win, so I really want to understand this as fully as possible.
With the line you gave, we end up in the same ending position if black plays 52..f4, reversing the zugszwang, don't we? I thought I calculated that 50. Kb6 and 50. Kxa6 worked out the same if black responded with ..a5, but a tempo better for me if he went straight for g2.
Ah, yes, you are right! So my suggested line does not work --- I fell into just the same trap as the one in the game :-)
--- Trying to break through on the queenside while black is the one who has more waiting moves left on the kingside has turned out to be a losing idea.
Next idea to try could be:
For example on move 43, play 43.g4!? for white, as you said.
Concretely with the ideas:
* Do not allow black to block the kingside. (When the smoke has cleared after g6, the white king will be able to get in via d3-e4-f5 towards capturing on g5)
* Do not allow black to keep any pawn waiting moves on the kingside.
* Then try to walk over to the queenside intending a4-a5 and breakthrough as in the game, but with the difference that black no longer saves himself with the waiting moves on the kingside
* If black prevents that by playing a6-a5 himself, use the king to walk over to e4-f5xg5
tricky questions for concrete calculation:
* will black at some point of this plan be able to break through on the queenside with a pawn break (b6-b5, maybe with black's king on d6)?
- When that happens, can white's king still be near enough (while on his journey to capturing g5) to get back to stop the resulting passed pawn? Or can the white pawns (d5 and b5, or the g4-pawn after Kxg5) be faster or equally fast in their race to queening themselves?
If this works out for white in very concrete lines of calculation, this may end up being a win for white.
If not, then white may just have to refrain from traing to invade, and it would be a draw.
Probably you have considered many of these ideas yourself. Now the interesting question is: Can we try concrete variations (they may be rather long :-) to see which of the two outcomes will be there at best play?)
Which makes us curious about the concrete lines you had looked at :-)
Interesting! In these lines, it turns out that white does not even really need the option to break through on the queenside, but white is much faster if he directly goes for taking g5.
Trying counterplay for black, my try would be a thought such as
43.g4 hxg4 44.hxg4 fxg4 45.fxg4 Kd6 46.Ke4 a5 (or b5 directly?) 47.Kf5 b5,
and the question is if black can get a passed pawn dangerous enough that it can end up managing to change the outcome here?
Is that inspiration for you for another concrete line to continue?
Lol @ 1.d4 :-)
Wow, 48.a5! in your main line is nice! While d5 is still defended, white has enough time for two nicely separated passed pawns on a6! and d5, which turn out to be much stronger than their black b- and c- counterparts.
In your sideline with 46...a5 47.Kf5 b5 48.cxb5,
I think black can make it a lot more interesting by trying
48...Kxd5!?
with the aim of making the black c-pawn queen faster than (or equally fast as?) white can queen his b-pawn or g-pawn, while maybe being able to block white's king from coming back in lines such as
49.Kxg5 c4 50.Kf4 Kd4!?
Maybe this can be about the most interesting place to calculate concretely, as far as we have found out so far?
Wow, nice!
This kind of feels like the typical calculation depth that can be necessary to really see through calculating a pawn endgame :-)
So with all we have found so far, this seems to be winning for white.
If we'd like to really be sure, we might next check whether there might be any improvements for black, such as finding a pawn setup on the kingside that is least vulnerable, so that white's task of striking it to fix the pawns there, and make them accessible for the white king to capture, gets harder?
At some point we should step back to when I decided to enter the k+p endgame, as that was the real point of no return, and see if it was an appropriate decision. I bring this up because the black pawn setup choice was made prior to my 43.Kd3? move. So, the endgame transitioned from pieces to king and pawn at move 33. I played Nd2 and allowed what happened next because I evaluated the pawn endgame as won due to the protected passed pawn being a huge advantage. How do you evaluate the endgame at this point, and what goes into your evaluation?
I guess you asked about black's pawn setup...I thought he did about the only thing he could do to make it hard for me, but I didn't cooperate. Do you have a better idea of what he could have tried?
Your feeling of "the farther he advances his king side pawns, the more they might tend to be vulnerable/quickly attackable" may point black towards a good direction.
(Some mention the saying "pawns cannot go backward, so think twice before each pawn move, it might be a weakening". Usually applies to middlegames or so, but maybe also here to the endgame?)
My try would be to suggest keeping black's pawns far behind (such as on f5, g6 and h7, and if necessary maybe advance a slight bit towards h7-h6 and maybe g6-g5).
The idea being that the longer the path for the white king towards capturing the decisive black king side pawn, the easier it may be for black to be in time with his queenside counter break chances of creating a fast passer there as soon as the white king has moved one step too far towards the king side to get back to the queenside in time to stop the passed pawn (and possibly still being too slow to just keep walking to the kingside, capture a pawn, and promote one of white's king side pawns quickest).
Maybe one could try to find concrete lines for this, too, trying to see whether white could break the black defense/counterplay if played about according to such ideas maybe?
Such as beginning with black playing a5/Ke7-d6 first, and then being ready for looking for a good point of time for a queenside pawn break, while trying to keep black's king side pawns as far back as possible while trying to keep the white king from getting in there on the king side too easily and quickly...
This is really interesting conversation. I love the in-depth analyses. In fact, I am kicking myself now for the foolish rook exchange and losing tempi. The original plan was to keep the king on d7 and move the rook to b8 adding pressure on the b-file and removing the defender of the d pawn. However, I ultimately chose to blockade king side and push queen side.
Going back, however, to some of the earlier analyses (with 43. g4), it might be helpful to note that my plan was not 45 ...Kd6. In fact, I was planning 45. ...Kf7. That way, if the white king ever moved to e4, I follow with ...Kf6! and take the opposition. It doesn't lead to much of a win, but keeps the game drawish and indirectly defends the passed pawn.
Yes, usually it is a very good idea to try to calculate a LOT before deciding to go into a pawn endgame --- just because it tends to become much more possible to calculate a LOT in pawn endgames as opposed to earlier positions with pieces still on the board.
Keeping pieces on the board instead usually requires more patience, no matter if it is objectively better or worse :-)
The idea of ...Kd6 in our discussed lines was to create maximum counterthread speed on the queenside.
If something like ...Kf7 instead, white may profit from that if he manages to create a target (weak pawn) on the kingside (that can be captured by the white king), then play d6... to force the black king off the kingside so that white can push through there (again hopefully faster than any possible breakthough passed pawn of black's on the queenside).
Of course that is mainly general thoughts and possibly first intuition ... which certainly will need to be checked by calculating concrete lines again :-)
http://www.chess.com/echess/game?id=103349262
I'm sure this is won for white at move 39 due to the protected passed pawn. I think the whole plan of 43. Kd3 moving to the queenside was where I went wrong. 43. g4 was the other plan I was considering as black can't hold the blockade, but both looked equally winning to me, so I picked one. I had counted on black's king running to g2, not sticking around to defend c5, and when I did evaluate that, while I saw the trebuchet, I completely overlooked that black had one more tempo he could waste with ..f4, and that was that.
Thoughts welcome, and I can provide lines I was looking at if anyone is curious.