Knight v Bishop (part 45)

White has just captured on e5 in Andrew C Waters (154) - Spanton (170), Kent League 140 Average 2020. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

Unbalanced pawn-majorities favour a bishop over a knight. In addition Black has an isolated d pawn. True, it is passed, but then so is White's protectable e pawn. The analysis engines Stockfish12 and Komodo12.1.1 give White a slight edge.

24...Nc4 25.f4!?

The engines prefer the less-committal 25.Kd3. My engines at the time of this game, Stockfish10 and Komodo10, preferred 25.e6!?, meeting 25....Kf8 with 26.Bg5!? My more-modern engines reckon the continuation 26...Nxb2 27.Ke3 h6 28.Bf6 gives White compensation for the sacrificed pawn, but no more.

25...Kf7

Stockfish12 strongly prefers 25...Na5!?

26.Kd3 Ke6 27.Kd4 Nb6 28.b3 Nc8 29.Ba3 Nb6

The engines prefer 29...b6!? but reckon White is on top.

30.Bb4 Nd7 31.Bd6 Nb6 32.a4 Nc8 33.Ba3 b6 34.Bb4 a5 35.Ba3?

The engines give White a large advantage after 35.Bd2!?, the point being to target the black queenside pawns fixed on dark squares, eg 35...Ne7 36.Be3 Nc8 (36...Nc6? 37.Kd3 leaves Black unable to defend b6) 37.Kc3 Kd7 38.Bf2 Ke6 39.b4 axb4+ 40.Kxb4 Kd7 41.Kb5 Kc7 42.Ka6 Kc6 43.e6! Kc7 44.Bd4 Kc6 45.Be5 h6 46.Bd4 Kc7 47.Bc3 Kc6 48.Be5 g5 49.fxg5 hxg5 50.Bf6 g4 51.h4 etc. This is a long line, but Black is very restricted throughout and must succumb eventually.

Clearly the queenside pawns are the direct cause of Black's troubles, but the moves 33...b6 and 34...a5 are quite liked by the engines, showing Black has major problems whatever he plays. It seems the diagrammed position is much more favourable for White in the long term than the engines realise.

35...Na7

Black equalises because it is possible to play ...Nc6+, driving away the white king, without dropping the b6 pawn.

36.Bd6

Or 31.Bc1 Nc6+ 32.Kc3 Kd7 33.Be3 Kc7, and if 34.e6!? then 34...Kd6 35.Bxb6 Kxe6 with equality.

36...Nc6+ 37.Kc3 Kd7

*****

*****

*****

*****

*****

38.g4?

This turns a draw into a loss by allowing Black to create a passed h pawn.

38...fxg4 39.hxg4 h5 40.gxh5 gxh5 41.f5

Also no good is 41.Kd2 h4 42.Ke1 d4 43.Ba3 d3 44.Bc1 h3 45.Kf2 Nd4 etc.

41...h4 42.Bf8

*****

*****

*****

*****

*****

42...h3?

Turning a win into a draw. The patient 42...Nxe5 and the aggressive 42...d4+ win easily.

43.e6+

*****

*****

*****

*****

*****

43...Kc7!

Other moves lose, eg 43...Ke8 44.Bd6 Nd8 45.Kd4 etc.

44.Bh6

*****

*****

*****

*****

*****

44...h2?

Turning a draw into a loss.

The game finished:

45.Bf4+ Kd8 46.Bxh2 Ke7 47.Bg3 Kf6 48.Bh4+ Kxf5 49.e7 Nxe7 50.Bxe7 Ke5 1-0

'Secret' Gambit Repertoire For White (part four)

BELOW is an article I posted in April, which fits into the current 'Secret' Gambit series as it covers a gambit against the French Defence that is very little-known.

HERE is another way to spice up a repertoire, and, believe it or not, it can be played against both the French Defence and the main way of declining the Queen's Gambit.
The same position is reached after 1.e4 e6 2.d4 d5 3.c4!? and 1.d4 d5 2.c4 e6 3.e4!?
As far as I can recall, I came up with the idea during an idle moment at a tournament in the Czech Republic four years ago
After the natural continuation 3...dxe4 4.Nc3 Nf6, the position resembles a Blackmar-Diemer Gambit (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6).

Position after 4...Nf6 in the 'New BDG'

Position after 3...Nf6 in the regular BDG

The extra moves in the first diagram - c4 for White, ...e6 for Black - have interesting effects.
Having a pawn on c4 gives White extra control in the centre. It keeps a black piece off the d5 square, and makes it easier, and probably more advantageous, to meet ...c5 with d5.
This costs a tempo, which Black has used to open a diagonal for his dark-square bishop. But playing ...e6 rules out developing the light-square bishop to f5 or g4, which is a common occurrence in the regular Blackmar-Diemer Gambit.
All in all, it could be argued this is an improved BDG, and certainly one in which Black cannot rely on having memorised a recommended antidote.
But perhaps there really is nothing new under the sun, for on checking in ChessBase's 2020 Mega database I found long-time US champion Frank Marshall played 1.e4 e6 2.d4 d5 3.c4!? four times in 1911 alone, while the move-order 1.d4 d5 2.c4 e6 3.e4!? is known from a game in the 1983 Armenian U18 Championship.
Here is Emil Diemer, of Blackmar-Diemer fame, using yet another move-order:
Diemer - Ewald Habermann
Zemgalis Farewell Tournament (Stuttgart) 1951
1.d4 d5 2.e4 e6 3.c4!?
Many BDG enthusiasts, when deprived of their favourite opening by 2...e6, play 3.Be3!?, which is known as the Alapin French.
3...dxe4 4.Nc3 Nf6 5.f3!? exf3
Black does not have to accept the gambit, as grandmaster Alexei Barsov showed in a 2004 game when he played 5...Bb4!? His 2067-rated opponent replied 6.Bg5, but could not find a decent answer to Barsov's continuation 6...c5. Better tries would seem to be 6.a3, as Marshall played in a draw in his 1911 match against Boris Kostić, and 6.Be3. Indeed, Be3 might be better played at move five, as I will show in the next game.
6.Nxf3 Bb4!?
As Barsov was later to do, Habermann makes full use of having opened the f8-a3 diagonal. More popular, however, is 6...Be7 7.Bd3 0-0, although the engines reckon White has fairly decent compensation for his pawn-minus (Stockfish10 reckons the position is level; Komodo10 gives Black a slight edge).
7.Bd3 Bxc3+!?
This is somewhat reminiscent of Black capturing on c3 in lines of the Göring Gambit Accepted, but the engines much prefer 7...c5 or 7...0-0.
8.bxc3 0-0
Here the engines give 8...Nbd7 or 8...b6, but with advantage for White.
9.0-0 c5 10.Bg5 Nbd7 11.Ne5 cxd4?
This allows a winning attack against Black's king, so the engines prefer 11...h6 12.Bh4 g5, but this also gives White a very dangerous attack.
 12.Nxd7 Bxd7 13.Rxf6!?
Also winning is the simple 13.Bxf6 gxf6 14.Bxh7+ Kxh7 15.Qh5+ Kg8 16.Qg4+ Kh8 17.Rf3 etc.
13...gxf6?
Best-play, according to the engines, is 13...dxc3 14.Qh5 g6 15.Qh4 Qb6+ 16.Kh1 Qd6 17.Bc2, when White has a huge attack as well as a bishop for three pawns.
14.Qh5 1-0
**********
Alexander Ianovsky (2202) - Vladimir Tchoubar (2390)
Kiev Independence Cup 2002
1.e4 e6 2.d4 d5 3.c4!? dxe4 4.Nc3 Nf6 5.Be3!?
This can be played with the idea of following up with f3 only after Black has committed to ...Be7, thus meaning ...Bb4 would come with loss of tempo. But more often the idea is to dispense with f3, certainly in the short term, and instead prepare long castling, a plan which should appeal to the attack-minded.
5...Be7
Black played ...Bb4 anyway in Spanton (1923) -  Ivo Škvír (1599), Olomouc Open 2016. There followed 6.Qb3 Bxc3+ 7.Qxc3 0-0, when 8.Ne2, intending Ng3, was the way to go, according to the engines, although they prefer Black.
6.Qc2 c5 7.d5!?
As I pointed out in the introduction, one of the advantages of having played c4 is that it makes it easier to reply to ...c5 with d5.
7...exd5 8.cxd5 Bf5
Not 8...Nxd5?? as 9.0-0-0 Be6 10.Nxd5 Bxd5 11.Bc4 wins for White.
9.0-0-0

How should Black proceed?

*****
*****
*****
*****
*****
9...0-0??
A remarkable mistake for a 2390 to make.
Blockading the passed pawn with 9...Bd6 makes sense. The engines then reckon Black has a small edge, but, with Black almost certainly castling short, the position is unclear.
10.d6 Bxd6 11.Nb5 Ne8 12.Qd2?!
Stronger is the simple 12.Nxd6 Nxd6 13.Qxc5.
12...Nc6 13.Nxd6 Nxd6 14.Qxd6 Qa5!?
Black's best try, according to the engines.
15.Qxc5 Qxa2 16.Qa3 (1-0, 29 moves)
**********
Here is a look at what can happen when Black tries to spoil White's fun by refusing to capture on e4, typically by playing 3...c6!?
Richard K Delaune (2365) - Alexander Passov (2290)
Eastern Open (Washington DC) 1997

Position after 3...c6!?

(The game reached the diagram position via the move-order 1.c4 c6 2.e4 e6 3.d4 d5)
4.e5!?
Komodo10's choice, 4.Nc3, is marginally more popular in ChessBase's 2020 Mega database, while Stockfish10 likes 4.exd5. Grandmasters have played all three moves, and 4.cxd5. Suffice it to say there is no consensus on how either player should proceed.
4...dxc4!?
This has been played by Alexander Shabalov, but the most-popular move, at least in Mega20, is 4...c5, whereupon Stockfish10 reckons White gets an edge with 5.Nf3 and 5.cxd5, but Komodo10 reckons both moves lead to equality.
5.Bxc4 Nbd7?!
The most-popular move, albeit from a small sample size, in Mega20. Black's idea is that this knight will hit White's light-square bishop with ...Nb6, while his other knight will occupy the outpost square d5. But it seems the plan is too slow, not least because it takes the queen's knight two tempi to reach b6, so no tempo is being gained on the white light-square bishop.
Shabalov has three times played 5...c5, when 6.d5!?, as first played in Igor Stohl (2530) - Shabalov (2465), Werfen (Austria) 1990, is very interesting. Shabalov replied 6...a6 (the engines prefer 6...exd5) and drew against Stohl's 7.Nc3. Shabalov again played 6...a6 when he had the same position 17 years later against Zviad Izoria. The Georgian grandmaster tried 7.Qf3, but lost in only 24 moves. However, the engines are adamant White has the upper hand after the simple 7.d6!? As with many lines in the 'New BDG', we are on fresh ground from an early stage, and it is very difficult to know where the truth lies.
6.Nf3 Nb6 7.Bd3
The bishop takes the chance to eye h7, as it would not seem to have many prospects on b3.
7...Ne7 8.Nc3 Ned5 9.Ne4 Be7 10.a3!?
White takes time-out to prevent Black's pieces from using the b4 square.
The engines reckon White is also much better after 10.0-0, and if 10...Nb4, then one line they give goes 11.Bb1 h6 12.a3 N4d5 13.Ne1!? with the idea of following up with Qf3 or Qg4.
10...0-0 11.h4!? f5
This comes to be Stockfish10's choice. The problem with Komodo10's 11...Kh8 is 12.Neg5, eg 12...h6 13.Bb1!, when Komodo10 reckons 13...g6 'only' gives White the upper hand, but as soon as the move is played on the board it comes up with 14.h5! with what the engines agree is a winning attack.
12.exf6 gxf6 13.Qe2 Qc7
Komodo10's choice, but Stockfish10 does not like it because of the reply ...
14.Neg5!
… whereupon Komodo10 agrees with the game's …
14...fxg5
… but Black seems to be busted after …
15.hxg5 Kg7
The engines reckon best is 15...Rf5 16.Ne5 Bd6, but then Stockfish10's 17.g6 Bxe5 18.Rxh7 seems to be winning, eg 18...Bg7 19.Bxf5 Qd6 20.Rxg7+! Kxg7 21.Qh5 Nf6 22.Qh6+ Kg8 23.Bd3, when White is a piece-for-two-pawns down but has a raging attack. The tactics are sharp and the line is quite long, but Black's position has been difficult to play from an early stage.
16.Ne5+ Ke8 17.Rxh7 Kd8
The only alternative, according to the engines, is 17...Nd7, but then 18.Qh5+ Kd8 19.Nf7+ Rxf7 20.Rxf7 is simply winning for White.
18.g6 Nf6 19.Bf4
Black's queen is trapped, despite being deep in its own half of  the board (1-0, 34 moves).
**********
CONCLUSIONS
The 'New BDG' is probably no better, but no worse, than the traditional Blackmar-Diemer Gambit.
Its main advantage for the practical player is that, over-the-board, Black will have to start thinking from move three, rather than following a memorised BDG antidote.
The main differences are that, in the New BDG, White has more central control while Black has a tempo more for development (but with a restricted light-square bishop).
This means, generally speaking, White is less likely to produce a quick knockout blow (not withstanding Diemer - Habermann), but can afford to, and should, build up more slowly.
Assuming Black captures on e4, White neither has to rush to make the gambit permanent with f3, nor rush to try to win the pawn back - after all, it is a doubled-pawn that may fall in due course anyway.
The position after 3...dxe4 has had no modern tests at the highest level, and possibly never will, but it will continue to appear from time-to-time in club chess.

Fundamentals (part 36)

Tobias Boje (1908) - Spanton (1995)
Gausdal Elo 2007

White has just captured on f5 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
The position is dead-equal, according to the analysis engines Komodo10 and Stockfish10.
28...Ke7 29.Kg4 c5?
Advancing and, if possible, centralising the king is usually a priority in pawn endings, and this one is no exception. Here 29...Kd6? loses to 30.Kh5, eg 30...Ke7 31.Kg6 Kf8 32.e7+! Kxe7 33.Kxg7 etc, so sensible was 29...Kf6, which maintains the balance.
30.f4?
White offered a draw.
But winning is 30.Kh5 Kf6 31.e7! Kxe7 32.Kg6 Kf8 33.f6, eg 33...gxf6 34.Kxf6 e4 35.e5 etc, or 33...e4 34.fxg7+ Kgg8 35.Kf5 Kxg7 36.Ke5 etc.
30...e4 31.Kg3 d4 32.Kf2 b5
*****
*****
*****
*****
*****
33.Ke2
White seems to have a draw with 33.b4! c4 34.c3! although Komodo10 reckons Black is winning. After 34...dxc3 35.Ke2 there seems no way for Black to convert his pawn advantage, although Komodo10 for a long time continues to believe Black is winning. However eventually Komodo10 agrees with Stockfish10's assessment of equality, one line running 35...Kd6 36.Ke3 h5 37.h4 Ke7 38.Ke2 when there is no way to make progress.
33...a5 34.Kd2
*****
*****
*****
*****
*****
34...e3+?
Black wins with 34...b4, eg 35.a4 h5 36.h4 Kf6 37.Ke2 c4! 38.Kd2 (if 38.bxc4 then 38...d3+ 39.cxd3 b3 etc) c3+ 39.Ke2 Ke7 40.Kd1 d3 41.cxd3 e3! 42.d4 Kd6 43.d5 Ke7, or 35.axb4 axb4 (35...cxb4 also wins) 36.Kd1 c4! 37.Kd2 (if 37.bxc4 then 37...d3 38.cxd3 e3! etc) c3+ 38.Ke2 h5 39.h4 Kf6 40.Ke1 d3 42.cxd3 e3! etc.
35.Ke2 c4 36.b4 a4 37.h4 h5 38.Kf3 Kf6 39.Ke2 c3 40.Kd3 Ke7 ½–½

Fundamentals (part 35)

Spanton (169) - Jim Howson (174)
Southern Counties Chess Union 2007

White has just captured on d1 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
Black is slightly better, according to the analysis engines Stockfish10 and Komodo10.
25...h4
This isolates the g5 pawn, but it is far from clear Black will be able to win it. The engines suggest a move such as 25...a5, 25...Ke5, 25...c5, 25...b6, etc..
26.Ke2 c5 27.Kd3 b5 28.a3
*****
*****
*****
*****
*****
28...f5?!
This position is trickier than it may seem. My notes show Junior 10.1 gave 28...a5? but that loses to 29.a4! bxa4 (if 29...b4 then 30.Kc4 Kd6 31.Kb5 etc, and if 29...c4+ then simply 30.Kd4) 30.Kc4 Kd6 31.Kb5. However the natural 28...Ke5 is among those moves that hold the balance.
*****
*****
*****
*****
*****
29.gxf6
The engines' 29.b3! puts great pressure on Black's position, eg 29...g6 30.b4 (30.exf5+ and probably 30.c3 also win) cxb4 31.axb4 Ke5 32.exf5 Kxf5 33.Kd4 Kxg5 34.Kf5 Kh6 35.Kxf4 Kh5 36.Ke5 Kg5 37.f4+ Kh6 38.Kf6 Kh5 39.Ke6 g5 40.Kf5 gxf4 41.Kxf4 followed by Kg4 and Kxh4 - a long line but a fairly forcing one. If 29...Ke5 then 30.exf5 Kxf5 31.c4 etc, and if 29...a5 then 30.a4.
*****
*****
*****
*****
*****
29...Kxf6?
Black draws with 29...gxf6, eg 30.c4 Kd6 31.cxb5 axb5 32.b3 Kc6 33.a4 c4+ (or 33...Kb6 but not 33...bxa4??) and now White has to find the only drawing move, but it is the obvious 34.Kc3.
*****
*****
*****
*****
*****
30.b4! cxb4
Or 30...c4+ 31.Kd4 Ke6 32.e5 (but losing is 32.Kc5? Ke5 33.Kb6 g5 34.Kxa6 g4 etc) g6 33.Ke4 g5 34.c3.
31.axb4 Ke5 32.Kc3 1-0
Black lost on time. The main line continues 32...g6 33.Kd3 g5 34.Ke2 (only drawing is 34.c3? Ke6 35.Kd4 Kd6 36.e5+ Ke7 [or 36...Kd7 but not 36...Ke6? 37.Ke4 etc] 37.Ke4 [not 37.Kd5? g4!] Ke6) Kd4 35.Kf2 Ke5 (not 35...Kc3 36.e5) 36.Ke1 Kd4 37.Ke2 Ke5 38.Kd3, which reaches the same position as after 33...g5 but with Black to move, which puts Black in zugzwang. There are alternatives along the way but the engines show White is winning.

Fundamentals (part 34)

David Barron (1966) - Spanton (2051)
Hastings 2006-07

Black has just captured on g6 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
No one, at least according to Stockfish10 and Komodo10. They reckon the position is dead-equal, but clearly one slip could radically change the assessment.
44.Ke5 Kf7
My original notes show Fritz9 and Junior10 wanted Black to play 44...Kh5. There seems nothing wrong with their suggestion, but Stockfish10 and Komodo10 both give the text.
45.b4 Ke7 46.Kf5 Kd6
*****
*****
*****
*****
*****
47.Kg6?
It seems 47.a3, 47.Ke4 and 47.Kf4 hold the balance.
47...Ke5
White is lost.
48.Kxg7 Kd4 49.c5 Kc4 50.a3
*****
*****
*****
*****
*****
50...a6
But White wins after 50...Kb3?? 51.b5.
51.Kf6 Kb3 52.Ke6 Kxa3 53.Kd6 Kxb4 54.c6 bxc6 55.Kxc6 a5 0-1

Fundamentals (part 32)

Spanton (2032) - Foppe-Jan Montsma (1956)
Amsterdam B 2005

Black has just captured on d5 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
26.Ke2
White has a large advantage, although the text is the only way to preserve it, according to the analysis engines Stockfish10 and Komodo10.
26...h5 27.Kd3 h4 28.g3 h3 29.f4 a6 30.a3 a5 31.a4 c5 32.g4
White's threat to create a passer on the kingside will, if all else fails, force Black to abandon the queenside.
32...c6 33.g5 Kd6
Or 33...Ke6 34.Ke4 etc.
34.Kc4
Black is in zugzwang.
34...Ke6 35.Kxc5 Kf5 36.b4 1-0

Fundamentals (part 31)

Spanton (2014) - A Seidl (1588)
Liechtenstein 2003

White has just captured on d3 - how would you assess the position?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
Black is slightly better.
Don't believe it? Neither do I, but that is the initial verdict of the analysis engine Stockfish10. It quickly changes to equal, and eventually settles on plus-over-minus, ie White has the upper hand. Komodo10 starts off rating the position as equal but comes to slightly prefer White.
Most human players of club standard and above would, I suspect, A) guess the ending is likely to have arisen from the Exchange Variation of the Spanish (it did), B) assess the position as much better, probably winning, for White.
27...Ke5
This is the engines' choice.
28.Ke3 c6
The engines suggest 28...Kd6 but that does not seem to be an improvement.
29.f4+ Ke6 30.g4 g6 31.a4 b5 32.b3 f5?!
I believe Black is lost anyway but this makes it too easy for White. The engines suggest 32...Kf7, eg 33.h4 Ke6 34.f5+ gxf5 35.exf5+ Ke5 36.Kf3 h6 37.a5 b4 38.Ke3 Kd6 39.Kf4 etc.
The game finished:
33.exf5+ gxf5 34.gxf5+ Kxf5 35.Kf3 h6 36.h3 h5 37.Ke3 Kf6 38.Ke4 Ke6 39.f5+ Kf7 40.h4 bxa4 41.bxa4 a5 42.Ke5 Ke7 43.f6+ Kf8 44.Kf5 Kf7 45.Kg5 Kf8 46.Kxh5 1-0

Fundamentals (part 30)

J Coleman (2052) - Spanton (2005)
Hastings Challengers 2002-03

White has just captured on d4  - what should the result be?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
White is winning easily, according to the analysis engines Komodo10 and Stockfish10, but why is this? I am not sure, but I guess the main reason is White's central space advantage.
39...b5
The seems to be Black's best try, but White's advantage is large: approximately +10 according to Komodo10, almost +24 according to Stockfish10.
40.f5 a5
Hoping to create a queenside passed pawn.
41.Ke4 Kd7 42.d5 c5
*****
*****
*****
*****
*****
43.d6?
White keeps a winning advantage with a move such as 43.f6 or 43.e6+.
*****
*****
*****
*****
*****
43...a4?
The general rule that it is best to first advance the unopposed pawn of a majority is correct here (but note that White's 43.d6? incorrectly conformed to that rule). After 43...c4 our postmortem analysis went 44.Kd4 Kc6 45.g6 hxg6 46.e6 Kxd6 47.exf7 Kxf7 48.fxg6, which is drawn after 48...a4. However the engines show White wins with the intricate line 45.a4!, eg 45...b4 46.Kxc4 Kd7 47.Kd4 Kc6 48.e6 Kxd6 49.exf7 Ke7 50.g6 hxg6 51.fxg6 Kf8 52.Ke5 Ke7 53.Ke4 Kf8 54.Kf5 Ke7 55.Kg5 b3 56.Kh6 b2 57.Kg7 b1=Q 58.f8=Q+ etc.
But this is academic as, going back to the position after 44.Kd4, the engines find 44...a4, which seems to draw, eg 45.e6+ Kxd6 46.exf7 Ke7 47.g6 hxg6 48.fxg6 Kf8 49.Ke5 Ke7 50.Ke4 and now, rather than 50...Kf8?, which loses in a similar way to the previous line thanks to 51.Kf5 etc, Black draws with 50...c3 or 50...b4.
The game finished:
44.Kd5 b4 45.e6+ fxe6 46.fxe6+ Kd8 47.Kc6 b3 48.e7+ Ke8 49.d7+ Kxe7 50.Kc7 b2 51.d8=Q+ Kf7 52.Qf6+ 1-0

Fundamentals (part 28)

Leo Keely (174) - Spanton (173)
Hastings Challengers 1998-99

White to make his 33rd move

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
33.a3?
This loses, as does 33.g3?, but White has two ways to make a draw: 33.c4+ and 33.a4! The latter is better because it sets a trap.
After 33.c4+ bxc4 34.Ke3 c5 (if 34 ...a4 then 35.a3) 35.dxc5 Kxc5 36.c3 Black cannot make progress, eg 36...Kb5 37.Kd2 Ka4 38.Kc2 Ka3 39.Kb1 etc.
After 33.a4! the impulsive 33.bxa4? loses to 34.c4+ Kd6 35.c5+! Kd7 36.Kc3 Kc8 37.Kb2 Kb7 38.Ka3 Ka7 39.Kxa4 Ka6 40.c3 Kb7 41.Kxa5 Ka7 42.c4 Kb7 43.d5 etc, eg 43...Kc8 44.d6 Kd7 45.dxc7 Kxc7 46.Ka6 and Black will lose the c6 pawn.
However after 33.a4! the analysis engines Stockfish10 and Komodo10 show Black draws with 33...Kd6 and the transpositional 33...h6, eg 33...Kd6 34.c4 (but 34.g3? loses to 34...bxa4, eg 35.c4 a3 36.c5+ Kd7 37.Kc3 a4 when White is in zugzwang) b4 when neither side can make progress.
After the text it is Black to play and win.
*****
*****
*****
*****
*****
33...a4?
33...c5 wins easily, eg 34.dxc5 Kxc5 35.Ke3 Kc4 36.Kd2 a4 etc.
After the text it is White to play and draw.
*****
*****
*****
*****
*****
34.g3
The engines show 34.c4+! draws after 34...bxc4+ 35.Ke3! c5 36.dxc5 Kxc5 37.c3.
34...c5 35.dxc5 c6?
Over-finessing - 35...Kxc5 wins.
After the text it is again White to play and draw.
*****
*****
*****
*****
*****
36.Ke3?
Drawing is 36.c4+ bxc4+ 37.Ke3.
The game finished:
36...Kc4 37.Kd2 h6 38.Kd1 Kxc3 39.Kc1 Kc4 40.Kb2 Kd4 41.Kc1 Kxc5 42.c3 Kc4 0-1

Fundamentals (part 26)

Spanton (163) - Jovanka Houska (173)
Barking (London) 1995

Black has just captured on f6 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
The position is dead-equal, according to Stockfish10 and Komodo10, but that does not mean everything is simple.
30.f4
*****
*****
*****
*****
*****
30...h5?
This loses, but can you see why?
*****
*****
*****
*****
*****
31.h3?
Winning is 31.f5+, eg 31...gxf5 32.gxh5 when White's outside passed pawns make for an easy full point. Similarly 31...Kf7 32.fxg6+ Kxg6 33.gxh5+ Kxh5 34.Kf4 is a simple win.
The game finished uneventfully:
31...hxg4 32.hxg4 b6 33.b4 Kd6 34.Kd3 c5 35.dxc5+ bxc5 36.a4 g5 37.fxg5 fxg5 38.Kd2 Kc6 39.Kd3 Kb6 40.Ke3 d4+ 41.Kd3 dxc3 42.Kxc3 cxb4+ 43.Kxb4 a5+ 44.Kc4 Kc6 45.Kd4 Kd6 46.Ke4 Ke6 46.Kd4 ½–½

Fundamentals (part 22)

Spanton (156) - L Kuiken (152)
Hastings 3rd Week 1992

Black has just captured on a5 - what should the result be?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
31.a4! 1-0
The point is 31.a4 effectively gives the c pawn the power of being a protected passer, which means the white king is free to pick up the black pawns and then return to the queenside. However, there are pitfalls along the way, eg here is the ending played out against Stockfish10: 31...f5 32.Kd2 Ka6 33.Ke3 Kb7 34.Kd4 (not 34.Kf4? d4) Kc6 35.a5 g5 (35...Kb4 36.Kxd5 Kxa5 37.c6 etc) 36.a6 g4 37.hxg4 fxg4 38.a7 (but 38.g3? h5 is a draw) Kb7 39.Kxd5 Kxa7 40.Kd6 etc.

Fundamentals (part 18)

Spanton (147) - Arthur F Reed (100)
Barbican (London) Rapidplay 1991

White has just captured on c4 - who stands better, and by how much?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
White has the more-centralised king and the better pawn-majority, so has an advantage. The opening, as you may well have guessed, was a Spanish Exchange, and a position like the one in the diagram is supposed to be White's dream ending. The supposition is probably correct, but Komodo10 reckons White has only a slight edge, while Stockfish10 is not that much more bullish about White's chances.
23...b6
The engines reckon Black should play 23...a5, although 24.f4, as in the game, and 24.b4, to get rid of White's doubled pawn (24...a4? is met by 25.b5), appear good answers.
24.f4 Kf7 25.e5!?
The engines strongly dislike this move, but it may be OK. They want White to gain queenside space with 35.b4.
25...fxe5
The engines switch their preferences between the text, 25...f5 and 25...Ke6, but it is not clear that any move solves Black's problems.
26.fxe5 Ke6 27.Kd4 c5+ 28.Ke4 b5
The engines prefer 28...g6, which covers the White king's entry square at f5, but they reckon White is much better.
29.c4 g5
Again the engines prefer ....g6, but with a large advantage for White.
30.g3 c6

White to play and win

*****
*****
*****
*****
*****
31.h4?
The only winning move is 31.h3, when Black will end up in zugzwang, eg if, as in the game, 31...h6 then 32.h4 gxh4 (32...b4 33.h5 a5 34.g4 etc, or 33...g4 34.Kf4 a5 35.Ke4 etc) 33.gxh4 and, once Black runs out of pawn moves, his king will have to retreat.
31...h6 32.h5 a5 33.cxb5?
This loses. White's easiest route to a draw is 33.g4 b4 34.Kf3 (or 34.Kd3) Kxe5 35.Ke3.
33...cxb5 34.g4 b4 35.Kd3 Kxe5 36.Ke3 Kd5 37.Kd3 a4?
This allows a draw. Black's most-direct win is 37...Ke5 38.Ke3 c4! 39.bxc4 a4 40.Kd3 a3 41.bxa3 bxa3 42.Kc3 a2 43.Kb2 Kd4 44.Kxa2 Kxc4 when Black wins both White's remaining pawns.
38.bxa4 c4+ 39.Kc2 Kc5

White to play and draw

*****
*****
*****
*****
*****
40.Kd2?
The drawing line is 40.b3 c3 (not 40...cxb3+? 41.Kxb3 and 42.Kxb4) when the mutual protected passed pawns mean neither side can make progress.
40...Kb6 41.Ke3
Black offered a draw.
41...Ka5 42.Kd4
I should have taken the draw as now it is Black to play and win.
*****
*****
*****
*****
*****
42....Kxa4?
This turns a win into a loss. Correct is 42...c3! 43.bxc3 b3 44.Kd3 Kxa4 etc.
43.Kxc4 b3
If 43...Ka5 then 44.b3.
44.Kc3
The game finished:
44...Kb5 45.Kxb3 Kc6 46.Kc4 Kb6 47.b4 Kc6 48.b5+ Kb6 49.Kb4 Kb7 50.Kc5 Kc7 51.b6+ Kb7 52.Kb5 Kb8 53.Kc6 Kc8 54.Kd6 Kb7 55.Ke6 Kxb6 56.Kf6 Kc5 57.Kg6 Kd5 58.Kxh6 1-0

Fundamentals (part 17)

Spanton (147) - J Marley (153)
Sutton (London) Premier 1991

Black has just captured on d5 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
The position is dead-equal, according to the analysis engines Stockfish10 and Komodo10.
41.Kc3 Kc5 42.g4 g5?
42.f5 holds. The problem with the text is the white king will get access to the e4 square.
43.f5
This is probably good enough, but even stronger is 43.fxg5 fxg5 44.b4+! axb4+ 45.Kd3 Kd5 46.b3 h6 47.h3 when Black is in zugzwang. Note that if 42...f5 had been played, this idea does not work, eg 43.g5 Kd5 44.b4 axb4+ 45.Kd3 Kc5 - White's king cannot advance because the e4 square is covered. If in this line White plays 45.Kxb4 then 45...Ke4 leads to an equal queen-and-pawn ending.
43...Kb5  44.b4?
Winning seems to be 44.Kd4 Kb4 45.Kd5 Kxb3 46.Ke6 Kxb2 47.Kxf6 a4 48.Kxg5 - both players will queen but White will be a pawn up with what looks like sufficient shelter for his king as well as threats against Black's remaining pawn.
44...a4 45.Kd2 Kb4 46.h4!? h6
46...gxh4 also seems to draw.
47.hxg5 hxg5 48.Kc2
I offered  a draw.
48...Kc4

White to play and draw

*****
*****
*****
*****
*****
49.Kd2?
White holds with 49.Kb1 as 49...Kd4 50.Ka2 Ke3 51.Ka3 Kf3 52.Kxa4 Kxg4 is a draw, according to the Nalimov endgame tablebase.
The game finished:
49...Kd4 50.Kc2 Ke4 51.Kc3 Kf4 52.Kb4 Kxg4 53.Kxa4 Kxf5 54.b4 g4 55.b5 g3 56.b6 g2 57.b7 g1=R 0-1

Fundamentals (part 16)

J Clifford (108) - Spanton (147)
Barbican (London) rapidplay 1991

White has just captured on c5 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
Black, thanks to having the more-centralised king and the better pawn-structure, is winning, according to Stockfish10, or at least has the upper hand (Komodo10).
30...Ke5
But it seems that necessary to prove Black's advantage is 30...d4! The point is that after 31.cxd4 Ke6 32.Ke3 Kd5 Black has more reserve tempi with his pawns and so d4 will eventually fall, which means c5 will also fall. So the engines suggest 31.c4!? when they much prefer Black but never manage to make progress, at least in the lines I looked at with them.
31.Ke3
Better is 31.Kf3! but can you see why?
*****
*****
*****
*****
*****
31...g6
Winning is 31...d4+ 32.cxd4+ (if 32.Kd2 then 32...Kd5!) Kd5 and again Black's extra reserve-tempi mean the d4 and c5 pawns will eventually fall. Note that after 31.Kf3! White can meet 31...d4 with the drawing 32.c4.
After I played the text White offered a draw for the third time in the game.
32.g3?
Drawing is 32.Kf3! and, probably, 32.d4+ and 32.Kf2!?
32...g5
32...d4+! wins.
33.h3?
The engines reckon 33.Kf3! and 33.d4+ draw.
33...h5
Once again ...d4+! wins.
34.d4+
The engines slightly prefer 34.Kf3! but it seems the text also draws.
34...Kf5
The engines reckon Black's best shot is 34...Ke6!? Their main line runs 35.Kd3 Kd7!? 36.h4! (an only-move, apparently, eg 36.Ke3 runs into 36...Kc6, and if 37.Kd3 then 37...Kb5 seems to win, one variation running 38.h4 gxh4 39.gxh4 a5 40.Kc2 Kc4 41.Kd2 b5 42.cxb6 cxb6 43.Kc2 b5 44.Kd2 b4 45.cxb4 axb4 46.Ke3 Kc3 etc) gxh4 37.gxh4 Ke6 38.c4 a6!? 39.cxd5+ Kxd5 40.a3 a5 41.a4 c6 42.Kc3 Ke4 43.Kc4 and White holds, but there are plenty of alternatives along the way.
35.Kf3?!
It may be White had to play 35.Kd3.
35...a5
The engines give 35...Ke6!, which reaches the same position as in the previous note but with the white king one further square away from the queenside. Their main line continues 36.Ke3 Kd7 and now 37.h4 does not work as after 37...gxh4 38.gxh4 Black has 38...Ke6!, eg 39.Kf4 Kf6 40.a3 a6 41.a4 a5 42.Ke3 Kf5 43.Kf3 c6 and the white king has to give way.
36.g4+ hxg4+ 37.hxg4+ Ke6 38.Ke3 Kd7 39.Kd3 Kc6 40.a4 b5 41.cxb6 cxb6??
41...Kxb6 draws easily, eg 42.c4 c6 but presumably I was still trying for a win.
42.c4 Kd6
White offered a draw.
43.Kc3 dxc4??
Other moves also lose, eg 43...Kc6 44.cxd5+ Kxd5 45.Kd3 etc, but the double question-mark is for not taking the draw.
44.Kxc4 Kc6 45.d5+ Kd6 46.Kb5 Kxd5 47.Kxb6 Kc4 48.Kxa5 Kc5 ½–½
My notes do not show, but I think that after I played 48...Kc5 White made his fifth draw-offer of the game, and even I by now could see I had to accept.

Fundamentals (part 15)

D Scott (166) - Spanton (147)
Hastings Main A 1990

Several candidate moves come to mind, including ...c5, ...b5, ...a6 and ...Ke7, but can you tell which one of them loses?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
32...c5?
This loses, as will be seen. The other three moves draw, but it is easier to understand why by first seeing how the text loses.
33.c3 c4
White offered a draw.
34.a4 ½–½
White should not have offered a draw. After 34...a5 (34...a6 is met in the same way), White has 35.g5 fxg5+ 36.Kxg5 and Black will lose the d5 pawn, eg 36...Ke7 37.Kf5 Kd6 38.Kf6 Kd7 39.Ke5 Kc6 40.Ke6 etc.
So the reason 32...c5? loses is that it unprotects the d5 pawn, after which the black king is shouldered out of the way by the white king The same is not true after the other three suggested moves, although Black has to beware zugzwang, eg:
A) 32...b5 33.b4 Ke7! (an only-move, as will be shown later) 34.g5 fxg5+ 35.Kxg5 Ke6 (another only-move, but this time easier to find) 36.Kg6 Ke7 (only-move) 37.Kf5 Kf7 (only-move) 38.Ke5 Ke7 (only-move) 39.c3 Kd7 (39...a6 also draws) 40.Kf6 Kd6 (another only-move, but now it can be seen White cannot make further progress as the d5 pawn is defended and he cannot get at the c6 pawn).
Going back to the position after 33.b4, the plausible 33...a6 loses to 34.g5 fxg5+ 35.Kxg5 because White has two reserve tempi, eg 35...Ke7 36.Kf5 Kf7 37.Ke5 Ke7 38.c3 (the first reserve tempo) Kd7 39.Kf6 Kd6 40.a3 (the second reserve tempo) Kd7 41.Kf7 Kd6 42.Ke8 Kc7 (or 42...Ke6 43.Kd8 Kd6 44.Kc8 etc) 43.Ke7 Kc8 44.Kd6 Kb7 45.Kd7 Kb6 46.Kc8 Ka7 47.Kc7 and the c6 pawn falls.
B) 32...a6 33.g5 fxg5+ 34.Kxg5 a5 35.a4 Ke7 (an only-move as Black needs to keep the d5 pawn protected from c6) 36.Kf5 Kf7 37.Ke5 Ke7 38.c3 (White uses a reserve tempo so he can shoulder the black king out of position, but here it is not enough to win, although Black has to continue to find only-moves) Kd7 39.Kf6 Kd6 40.b3 (not 40.b4? as 40...b5! wins since Black will get an unstoppable passer) Kd7!? (alos drawing is 40...b5 and, now that b3 has been played, even 40...c5, but the text sets a trap) 41.Kf7 Kd6 and White has to accept a draw as the analysis engines Stockfish10 and Komodo10 show 42.Ke8? loses to 42...c5!, eg 43.Kd8 c4 44.bxc4 dxc4 45.Kc8 Kc6 with ...b5 to come.
C) 32...Ke7 (perhaps surprisingly, at least at first glance, the engines reckon this retreat also draws) 33.Kf5 Kf7 34.a3 a6 35.c3 b5 36.b4 (Black has been run out of pawn moves and has to move his king, but White has no reserve tempi and so cannot win) Ke7 37.g5 fxg5 38.Kxg5 Kf7 (38...Kd7 also draws, but losing is 38...Ke6? 39.Kg6 Ke7 40.Kg7! Ke6 41.Kf8 as the c6 pawn eventually falls) 39.Kf5 Ke7 40.Ke5 Kd7 41.Kf6 Kd6 - Black draws as his king cannot be shouldered out of the way by the white king.

Fundamentals (part 14)

Spanton (147) - D Ewart (148)
Highbury (London) Rapidplay 1990

White has just captured on c4 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
No prizes for guessing this ending arose from the Exchange Variation of the Spanish.
The analysis engines Stockfish10 and Komodo10 are not particularly impressed by the white position, but eventually agree White is slightly better. I suspect most humans would put White's advantage as higher than that - certainly White is the only player who can realistically hope for a win.
26...g5!?
Stockfish10 marginally prefers 26...a5. Komodo10 at one point sticks with 26...f5 for quite some time, but then switches to the text.
27.fxg5!?
The engines much prefer 27.g3, continuing 27...a5 28.b4 axb4 29.h4!? gxh4 30.gxh4 f5!? 31.e5 Kf7 32.Kxb4 with what they reckon is a large advantage for White, one line running 32...Ke6 33.Kc4 Ke7 34.Kc5 Ke6 when they continue to much prefer White, but do not come up with a continuation that makes progress.
27...fxg5 28.Kd4 c5+! 29.Ke3
The white king has to fall back as 29.Kxc5? is met by 29...Ke5 when White is close to being lost (the engines reckon Black is winning, but best play seems to lead to a queen-and-pawn ending in which Black is a pawn up, and is winning according to the engines, but the Nalimov endgame tablebase shows the position to be drawn).
29...h5!?
This seems to be OK, although at first the engines do not like it, but more natural is 29...Ke5.
30.g3
A better try seems to be the engines' 30.h4!?, eg 30...gxh4 31.Kf4 Kf6 32.e5+ Kg6 33.e6 Kf6 34.e7 Kxe7 35.Kg5 h3! 36.gxh3 Ke6 37.Kxh5, but 37...Kf5 looks to hold the draw.
30...c6?
Black seems to have a simple draw after 30...Ke5, and if, as in the game, 31.h4 then 31...g4 as Black has more reserve tempi and so can stop the white king penetrating.
31.h4 gxh4
Now 31...g4 is hopeless, but the text is no improvement.
32.gxh4 Ke5
At last, but too late - Black's reserve tempi will not save him as the white king can oscillate between f3 and e3 until the black king is forced to withdraw.
33.c3 b5 34.Kf3 b4 35.c4 a5 36.Ke3
The game finished:
36...Ke6 37.Kf4 Kf6 38.e5+ Ke6 39.Ke4 Ke7 40.Kf5 Kf7 41.e6+ Ke7 42.Ke5 Ke8 43.Kd6 Kd8 44.e7+ Ke8 45.Ke6 a4 46.bxa4 b3 47.Kd6 Kf7 48.Kd7 1-0

Fundamentals (part 11)

Keith Jones (143) - Spanton (147)
London League Division 5 1990

White has just captured on f2 and offered a draw - should Black accept?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
The position is dead-equal, according to the analysis engines Stockfish10 and Komodo10, so accepting a draw or playing on is largely a matter of taste.
35...h4 36.Kf3 b5 37.e4 bxc4 38.bxc4 fxe4+ 39.Kxe4

Black to play and draw

*****
*****
*****
*****
*****
39...a6?
Three moves draw, but this is not one of them.
A 39...a5 40.a4 c5 (40...Kf6 also draws) 41.dxc5 (simpler than 41.d5+ Kf6 42.d6 cxd6 43.Kd5 Ke7 44.Kc6 Ke6 45.Kb5 Ke5 46.Kxa5 d5 47.cxd5 Kxd5 48.Kb5, when both players queen and the game is drawn)) c6.
B 39...c5 40.dxc5 (40.d5+ Kf6 41.a4 a5 transposes to 41.d5+ in line A) c6.
C 39...Kf6 40.c5 (40.a4 a5 transposes to previous lines) Ke6 41.a4 a5 42.Ke3 Kf5 (but not 42...Kd5? 43.Kd3 Ke6 44.Ke4 Kf6 45.d5 etc) 43.Kf3 Kf6 44.Ke4 Ke6.
40.a4 a5 ½–½
Both 41.c5 and 41.d5+ win for White.

Fundamentals (part 10)

Spanton (151) - J Simons (130)
Kings Head (London) Major 1990

Black has just captured on e5 - how should White proceed?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
It might be thought Black is obviously better. After all, he has more space in the centre and he has an undamaged pawn majority. But Komodo10 reckons Black has just a slight edge, while Stockfish10 rates the position as level. However, the engines agree White's next move is critical.
27.b4?
The engines give 27.Kc4. But as soon as the move is played on the board, Stockfish10's assessment switches from dead-equal to Black being the equivalent of almost a pawn ahead. However, given enough time, Stockfish10's assessment changes to about a half-pawn advantage for Black, and eventually to equality. Komodo10 goes through a similar process.
Best play after 27.Kc4, the engines agree, is 27...Ke4 28.g4 c6 29.Kc5 d3 30.cxd3+ Kxd3 31.a4 Kc2 32.b4 Kb3 33.a5 Kb2 34.f3 Kb3 35.f4 Kb2 36.g3 Kb3 37.f5 Kc3 38.Kd6 Kxb4 39.Ke6 c5 40.Kf7 c4 40.Kxg7 c3 41.f6 c2 42.f7 c1=Q 43.f8=Q+ - a long line that is largely forced and seems to lead to equality.
27...c6?
The engines reckon Black should play 27...b5 or 27...Kd5, in both cases preventing 28.Kc4.
28.a4? Kd5
Black seems to be winning in all lines.
The game finished:
29.g4 b6 30.f3 c5 31.b5 c4+ 32.Kd2 Kc5 33.c3 Kd5 34.g3 dxc3+ 35.Kxc3 Kc5 36.f4 Kd5 37.g5 hxg5 38.fxg5 Kc5 39.g6 Kd5 40.Kc2 Kd4 41.Kd2 c3+ 42.Kc2 Kc4 43.Kc1 Kb3 0-1

Fundamentals (part nine)

Spanton (151) - Martin Cutmore (152)
Barbican (London) Rapidplay 1989

Black has just captured on e6

*****
*****
*****
*****
*****
Black's king is more centralised, and White has a backward e pawn, but neither factor seems very important - the position is dead-equal, according to analysis engines Stockfish10 and Komodo10.
26.Kf3 h5
Komodo10 reckons 26...f5 27.Kf4 Kf6 is a tiny bit better for Black, which is why the engine preferred 26.g4. Stockfish10 rates the position after 27...Kf6 as still dead-equal.
27.e4 dxe4+ 28.Kxe4
We now have rival pawn majorities, which makes it much easier to go wrong, as Black does immediately.
*****
*****
*****
*****
*****
28...f5+?
A useful general rule-of-thumb is that you should first advance the unopposed pawn of a pawn majority. But here that rule-of-thumb is wrong as, after the text, White's h pawn is holding up two black pawns. And since the white g pawn is holding up the f pawn, Black's pawn majority can no longer make a passed pawn with normal play.
Indeed, after ...
29.Kf4
... Black is obliged to reply ...
29...Kf6
(29...g6 30.Kf5 Kf7 is no improvement) ... and White is free to create a winning passer on the queenside.
30.c4 g6 31.d5 cxd5 32.cxd5 a5 33.b3 b5 34.a4 b4 35.g3
Having this reserve tempo is nice, but it is not necessary as 35.d6 also wins.
35...Ke7 36.Ke5 1-0

Fundamentals (part six)

P Sargeaunt (155) - Spanton (151)
Middlesex League Division 3 1989

White has just captured on e3 - who stands better?

*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
Black has doubled isolanis, which are unlikely to ever be dissolved, but it is far from obvious how White is to get at them.
White has a backward e pawn. He should be able to get rid of that, but after e4 dxe4 White will have an isolated d pawn.
Neither side has a space advantage, or prospects of creating a pawn majority. Perhaps it is not surprising that the analysis engines Stockfish10 and Komodo10 reckon the position is dead-equal.
24...Kf7 25.Kf2 Kf6 26.Kf3 Kf5
Both players have centralised their kings as far as they are able, and it is time to think of pawn moves.
27.b4 b6 28.g4+

Black to play and draw

*****
*****
*****
*****
*****
Often in pawn endings it is fatal for a king to retreat, but here 28...Kg5? loses as Black eventually ends up in zugzwang after 29.Kg3, eg 29...a6 30.a3 a5 31.bxa5 bxa5 32.a4 Kf6 33.Kh4 Ke6 34.Kg5 Kf7 35.h4.
28...Kf6
Also fine is 28...Ke6.
29.h3
If 29.Kg3, then 29...g5 holds.
White offered a draw.
29...g5 30.e4
White gets to dissolve his backward pawn (or turn it into a protected passer) and gets his king to the fourth rank, but the position remains dead-equal, according to the engines.
30...dxe4+
Actually, the engines reckon 30...Ke6 31.e5 is also level, but not 31.exd5+?? Kxd5 32.Ke3 Kc4 etc.
31.Kxe4 Ke6 32.a3 a6

White to play and draw

*****
*****
*****
*****
*****
33.a4?
For the second time in this ending, a king is better off retreating than following normal good-practice by standing firm.
Both 33.Ke3 and 33.Kd3 draw, according to the engines, eg 33.Ke3 Kd5 34.Kd3 b5 35.Ke3 (not 35.Kc3? as the d pawn falls after 35...Ke4) Kc4 36.Ke4 Kb3 37.Ke5 Kxa3 38.Kd6 Kxb4 39.Kxc6 Ka4 40.d5 b4 - both players queen and the game is drawn.
That is not an easy line to calculate, but White had to try it as the text loses immediately.
33...a5 34.bxa5 bxa5 35.Kd3
No improvement is 35.Ke3 Kd5 36.Kd3 c5 37.dxc5 Kxc5 38.Kc3 g6 39.Kb3 Kd4 etc.
35...Kd5 36.Kc3 c5 37.dxc5 Kxc5
The game finished:
38.Kd3 Kb4 39.Kd4 Kxa4 40.Kc4 Ka3 41.Kc3 a4 42.Kc2 Kb4 43.Kb2 a3+ 44.Ka2 Ka4 45.Kb1 Kb3 46.Ka1 a2 47.h4 gxh4 48.g5 h3 49.g6 Kc3 0-1