Summing Up Bischofsgrün

MY score of +3=4-2 in the seniors' championship of Upper Franconia lost exactly six Fide elo.

Because my game yesterday again finished quickly, I decided to walk up the Ocksenkof one last time, stopping by a lake for coffee and by the source of the White Main for bread and cheese, taking a final batch of photos along the way.

Bischofsgrün Round Nine

I WAS downfloated.

Achim Heller (2036) - Spanton (1954)

Queen's Gambit Declined

1.d4 d5 2.Nf3 e6 3.c4 Nf6 4.e3!?

More popular are 4.Bg5!?, 4.g3 and especially 4.Nc3, but there are 13,345 examples of the text in ChessBase's 2025 Mega database.

How should Black respond?

*****

*****

*****

*****

*****

4...Bb4+

Stockfish17 and Dragon1 suggest 4...Be7 or 4...b6.

5.Nbd2!?

This scores 65% in Mega25.

5...0-0 6.a3 Be7 7.b3 c5 8.Bd3 Nc6 9.Bb2 b6 10.0-0

Now both sides have castled, how would you assess the position?

*****

*****

*****

*****

*****

Space is equal, and the pawn-structure almost symmetrical. The white king's bishop is more active than its opposite number, but the white queen's knight less so than its opposite number. The engines reckon the game is equal.

10...Bb7 11.Qe2 Re8 12.Rfd1 Bf8!? 13.Rac1 Rc8

Michael de Verdier (2276) - Sergey Ivanov (2530), Swedish Team Championship 2010, went 13...cxd4 14.exd4 g6 15.c5!? bxc5 16.dxc5 e5 17.Bb5, with a slight edge for White, according to the engines (but 0-1, 57 moves).

14.cxd5 Nxd5?!
Michał Olszewski (2543) - Fabrizio Bellia (2376), Bratto (Italy) 2010, saw 14...exd5 15.Ba6!? Bxa6 16.Qxa6 Qe7 17.Ne5 Nb8!?, with an equal game, according to the engines (but 1-0, 48 moves).

After the text the engines reckon White is winning.

15.dxc5 Bxc5 16.b4 Be7 17.Nc4 Bf6?

The engines reckon Black had to play something like 17...Rf8, 17...Rc7 or 17...a6, but agree White is winning.

18.b5?!

This is enough for a comfortable advantage, but stronger is 18.Bxf6, when Black has three ways to recapture, but all are deeply unpleasant:

A) 18...Qxf6 19.Nd6 Nc3 20.Rxc3 Qxc3 21.Nxb7;

B) 18...Bxf6 19.Bxh7+ and 20.Rxd8;

C) 18...gxf6 19.e4! Nf4 20.Qe3.

18...Nb8?!

Almost certainly better is 18...Na5!?

19.Nfe5

Also still good is Bxf6.

19...Nd7?

19...Qe7, or even 19...Bxe5!?, keeps the game going.

20.Nxf7?!

Even stronger is 20.Bxh7+! (20.Nxd7 is also better than the text) Kxh7 21.Qh5+ Kg8 22.Nxf7 Qe7 23.Ncd6 g6 (otherwise 24.Qh8#) 24.Qxg6+ Kf8, when White has a winning attack, eg 25.Qh6+ Kg8 26.Bxf6 Qxf6 27.Qxf6 N7xf6 28.Nxc8 Rxc8 29.Rxc8+ Bxc8 30.Nd6, with an advantage worth about a rook, according to the engines.

20...Kxf7 21.Nd6+ Kg8??

Easily the worst of three possible king moves. After 21...Kf8 22.Nxc8 Bxb2 23.Qxb2 Bxc8 24.Bxh7 White has the upper hand (Dragon1) or is at least slightly better (Stockfish17).

22.Rxc8!?

Also very strong is 22.Qh5.

22...Bxc8 23.Qh5 Nf8?!

23...h6 is a modest improvement, but 24.Qf7+ Kh8 25.Qxe8+ Qxe8 26.Nxe8 Bxb2 27.Nd6 wins easily enough.

24.Qf7+ Kh8 25.Nxe8 Qe7

Forced - White threatened 26.Qxf8#.

26.Qxe7 Bxe7 27.Nxg7

The game finished:

27...Kg8 28.Nh5 Ng6 29.Bc4 Nc7 30.Nf6+ 1-0

Summit Special

GOOD weather returned yesterday, and I again had a fairly short game, so that gave me virtually the whole afternoon for walking.

Rather than use the cable car to ascend the Ochsenkopf, I decided to again walk it, stopping for coffee beside a lake, and for lunch at the source of the White Main.

Not much to say about the walk, except that it was a lot of fun and I took some more photos.

Bischofsgrün Round Eight

Spanton (1954) - Michael Schuh (1829)

English

1.c4 Nf6 2.Nc3 e6 3.e4!?

This is a well-known way to prevent Black playing a Nimzo-Indian or a regular Queen's Gambit Declined.

This position occurs 13,848 times in ChessBase's 2025 Mega database, but what should White's system be called?

*****

*****

*****

*****

*****

I understood it to be the Mikėnas-Flohr System, named after Vladas Mikėnas, who could be variously described as Estonian, Lithuanian and Soviet Russian, and Salo Flohr, who had claims to be Austrian, Polish, Ukrainian, Czechoslovakian and Soviet Russian.

However Wikipedia's article on the English Opening five times refers to the variation as the Mikėnas-Carls, honouring Vladas Mikėnas and the German player Carl Carls.

To further confuse the issue, the same article also calls the variation the Flohr-Mikėnas Attack, albeit spelling Mikėnas without an accent (as also happens one time in Mikėnas-Carls).

There are 16 games in Mega25 in which Carls played 3.e4!?, compared with just three for Flohr (plus a further one in which Flohr had the black pieces) and 11 for Mikėnas (plus five for Mikėnas's son Alius). 

From these numbers it could be argued that the fairest name is Carls-Mikėnas, especially as Carls seems to have been the first of the three to have played it.

On the other hand there is a long history of openings being named after the famous, rather than the originators, in which case perhaps Aron Nimzowitsch's name should be added to the list - he has six games with 3.e4!? in Mega25, including a 1931 win over Flohr.

If anyone has a definitive answer to the conundrum, I would be interested to learn it.

3...d5 4.e5

Carls usually played 4.cxd5, Flohr and Nimzowitsch preferred the text, Mikėnas (senior) began with 4.e5, but later in his career switched to 4.cxd5, and Mikenas (junior) plays 4.cxd5.

4...d4 5.exf6 dxc3 6.fxg7 cxd2+ 7.Bxd2 Bxg7 8.Qf3!?

This positional pawn sacrifice seems to a novelty. The main continuation is 8.Qc2.

How should Black respond?

*****

*****

*****

*****

*****

8...Nc6!?

Stockfish17 and Dragon1 fluctuate between accepting the gambit and declining it with 8...Bd7!?

After 8...Bxb2, play might continue 9.Rb1 Qf6!? 10.Qb3!? Be5 11.Nf3 Nc6!? 12.Nxe5!? Qxe5+ 13.Qe3 Qxe3+ 14.Bxe3, when the engines reckon White's bishop-pair does not fully compensate for being  a pawn down. However, both sides have plenty of alternative move along the way.

9.0-0-0 Qf6 10.Qxf6 Bxf6

With Black having (more-or-less) forced queens off the board, how would you assess the position? 

*****

*****

*****

*****

*****

Black has an extra pawn-island, but has the only pawn on the central two files and has an actively placed dark-square bishop. The engines reckon the game is equal.

11.Nf3 Bd7 12.Bd3 0-0-0 13.Bg5!?

The engines reckon 13.Ng5 is well-met by 13...Ne5.

13...Bxg5+ 14.Nxg5 Ne5 15.Be2 h6 16.Ne4?!

This is probably a mistake, so 16.Nf3 or possible 16.Nh3!? should have been played.

16...Bc6 17.f3 Rhg8

After 17...Bxe4!? 18.fxe4 Black has a strong knight against a bad bishop. The engines also like 17...Rdg8!?, with ...Bxe4 to come, one point being 18.Ng3?! can be met by 18...h5.

18.Rhg1?

Better is 18.Ng3, or 18.Rxd8+ Kxd8 19.Ng3.

18...Rxd1+ 19.Kxd1 f5?

Missing the win of material by 19...Bxe4 20.fxe4 Rd8+ and 21...Rd4.

20.Nd2 Rd8 21.Kc2 Ng6 22.g3 Ne5

MS offered a draw.

Who, if anyone, has the advantage?

*****

*****

*****

*****

*****

White's better pawn-structure gives a slight edge, according to Dragon1, although Stockfish17 is not so sure.

23.Re1 Bd7

Not 23...Rxd2+? 24.Kxd2 Bxf3 as White, rather than losing to 25.Bxf3?? Nxf3+ etc, plays 25.Bd3, and after the forced 25...Nxd3 26.Kxd3 Black has virtually no chance of surviving the rook-v-bishop ending, despite starting it a pawn up.

24.f4 Nc6 25.Kc3 e5 26.fxe5 Nxe5 27.Bh5!?

How should Black proceed?

*****

*****

*****

*****

*****

27...Nc6?!

Probably better is 27...Ng4, not fearing 28.Bxg4.

28.Nf3 Rf8 29.Nd4?!

Almost certainly better is going straight after the f pawn with 29...Nh4.

29...Nxd4 30.Kxd4 Kd8

The engines point out 30...f4!? is possible, as 31.Rf1 can be met by 31...Rd8.

31.Ke5 Bc6?

Still possible is ...f4 as 31.gxf4?? blunders the white bishop to 31...Rf5+.

32.Rf1 Be4?

It seems there is no defence, eg the engines' 32...Be8 is simply met by 33.Bg4, after which the f pawn falls, but there would still be quite a bit of play before Black would be obliged to resign.

33.g4 Ke7 34.gxf5 Bd3?!

Or 34...Bc2 35.f6+ Kd7 36.Rf2 Bb1 37.Rd2+ Kc6 38.b4 with an advantage the engines reckon is worth much more than a rook.

35.f6+ Kd7 36.Rd1 1-0

Reining It In

THUNDERSTORMS and showers threatened yesterday, so instead of going high into the Fichtel mountains, I walked Bischofsgrün's 10.4-kilometre panoramaweg.

Despite the overcast sky, visibility was good

I had a momentary shock when this sight greeted me down a sidetrack

But I quickly realised it was a carving in the mould of ones I saw earlier.

And I soon spotted another (two actually, as a second deer can just about be seen, although mostly hidden in the bushes)

I also took the chance to make a small detour to walk round a military cemetery I visited last year.

Bischofsgrün Round Seven

Spanton (1954) - Roland Welsch (1890)

Caro-Kann Panov-Botvinnik

1.c4 c6 2.e4 d5 3.exd5 cxd5 4.d4 Nf6 5.Nc3 e6

This is the main move in ChessBase's 2025 Mega database, but it scores three percentage points lower than the 48% of second-most popular 5...Nc6.

6.Nf3 Be7

How should White proceed?

*****

*****

*****

*****

*****

7.Bg5!?

This has been played by grandmasters, but the main line in Mega25 runs 7.cxd5 Nxd5 8.Bd3 Nc6 9.0-0 0-0 10.Re1, with a slight edge for White, according to Dragon1, although Stockfish17 calls the game equal.

7...0-0 8.Rc1!?

The main moves are 8.c5!? and especially 8.Bd3!?

How should Black respond?

*****

*****

*****

*****

*****

8...Qb6!?

Apparently a novelty, in a position in which there are eight known moves, headed by 8...Nc6.

9.Qc2

The engines suggest 9.c5!? Qxb2 10.Bd3, claiming a slight edge.

9...Nc6

The engines prefer 9...dxc4 10.Bxc4, and now 10...Nc6.

10.c5!? Qc7 11.Bb5!? Bd7 12.Bxc6!?

The point of this manoeuvre is to slow Black getting in ...e5, the classic answer to White's queenside push.

12...Bxc6 13.b4 a6 14.0-0

Now both sides have castled, how would you assess the position?

*****

*****

*****

*****

*****

Barring blunders, the game is likely to revolve around whether Black can generate central counterplay to distract White from supporting and advancing the white queenside pawn-majority. The engines give White a slight plus.

14...Bb5?

Black spends a tempo obliging the white king's rook to move to a better location, and meanwhile provides White with a tempo for advancing the a2 pawn.

15.Rfe1 Rfe8 16.a4 Bd7

Probably no better is 16...Bc4.

17.Ne5 Nh5

Offering to give up the bishop-pair in the hope of getting Black's kingside pawns rolling.

18.Bxe7 Rxe7 19.Qe2 Nf6 20.b5 axb5 21.axb5 Be8!? 22.Ra1 Qb8 23.Qb2 Rxa1 24.Rxa1 Nd7 25.Nd3 f6

At last threatening to play ...e5, although d5 has to be protected first.

26.Qa3 Bf7??

Black is positionally lost anyway, according to the engines, but this allows an immediate finish.

27.c6 1-0

Surprising Find

No big walk yesterday, but these carvings, left in the open, but off the beaten track, made interesting viewing

Bischofsgrün Round Six

FACED a Frenchman.

Patrick Chochoy (1757) - Spanton (1954)

Maróczy Bind

1.e4 c5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 g6 5.c4

Entering the Maróczy Bind scores a very healthy 57% in ChessBase's 2025 Mega database, while the more popular 5.Nc3 returns a disappointing 49%.

5...Bg7 6.Be3 Nf6 7.Nc3 d6 8.Be2 0-0 9.0-0

This position occurs 19,854 times in Mega25

*****

*****

*****

*****

*****

9...Bd7 10.Qd2

The one time Géza Maróczy had the position after 9...Bd7, in a 1906 game, he played 10.h3, which had been played against him two years previously in the stem game. That also seems to be Dragon1's top choice, very marginally ahead of the unusual 10.Re1!?, although the engine does for a while prefer the text. Stockfish17 likes the trendy 10.Nc2!?, again ahead of 10.Re1!? 

10...Nxd4

This is why 10.Nc2!? has become fashionable - to prevent Black reducing the cramp in the black position by initiating exchanges.

11.Bxd4 Bc6 12.f3 a5!?

The main point of this move is to secure c5 for the remaining black knight. It also gains space, but more-or-less gives up hopes Black might have of breaking the Bind, or at least challenging it, with ...b5.

13.b3 Nd7

We are still very much in mainstream theory - this position occurs 2,404 times in Mega25

*****

*****

*****

*****

*****

14.Be3!?

This is easily the commonest continuation, following the advice that when you have a bad bishop (White's light-square bishop is hemmed in by its own pawns) avoid exchanging your good one.

14...Nc5 15.Rab1

Again easily the main continuation.

15...Qb6!?

Self-pinning looks strange, at least at first sight, but White cannot immediately take advantage of Black's setup on the g1-a7 diagonal.

16.Rfc1

This position occurs 1,114 times in Mega25

*****

*****

*****

*****

*****

16...Qb4!?

The engines prefer this over the much more common 16...Rfc8.

17.Rc2

This is the usual response, but the engines suggest 17.Qb2!?

17...Bxc3!?

This unusual continuation - normal is 17...Rfc8 - is the top choice of the engines.

How should White recapture?

*****

*****

*****

*****

*****

18.Qxc3

This lets Black follow through with the main idea of 17...Bxc3!? - getting queens off the board.

Instead, after 18.Rxc3!? it appears at first that Black can take on e4 with the bishop as 19.fxe4?? loses to 19...Nxe4, followed by snaffling the c3 rook. But White does not have to fall in with Black's plans, and can play 19.Rb2!? Bc6 20.a3!, meeting both 20...Qxa3 and 20...Qb6 with 21.b4, winning material. The engines also point out 19.a3!?, which also seems very good.

However Black does not have to play 18...Bxe4? - rather 18...Na4 forces an exchange of queens, as in the game.

18...Qxc3 19.Rxc3

How would you assess this queenless middlegame?

*****

*****

*****

*****

*****

White has more space and the bishop-pair, although the latter factor is tempered by there being 14 pawns on the board, with all but one white pawn on the same-colour squares as White's light-square bishop. The Black knight is well-posted, for now - it faces being evicted by White playing a3 and b4. Dragon1 gives White a slight edge, but Stockfish17 calls the game equal.

19...f5?

This novelty is misdirected. Better are the known moves 19...Ne6 and 19...b6, but two other known moves - 19...Rfc8? and 19...Rfb8? - are mistakes because of the reply 20.e5.

20.exf5?!

This is good enough for an advantage, but 20.e5 is stronger, eg 20...f4 (20...dxe5?? 21.Bxc5) 21.Bf2 Rfe8 22.exd6 exd6 23.Bf1, when the isolated d pawn is a serious weakness.

20...Rxf5 21.Bxc5?!

White should almost certainly not give up the better white bishop.

21...Rxc5

The engines slightly prefer 21...dxc5!?

22.Re3 Kf7 23.f4 b6 24.Re1 e5

The engines prefer passive-looking defence with 24...Re8 or 24...Ra7.

25.fxe5

The engines give White a slight edge (Dragon1) or the better part of equality (Stockfish17) after 25.Rf1.

25...dxe5

The pawn is isolated, but passed. However the engines prefer 25...Rxe5!?

26.a3 e4 27.Bd1 Re5 28.Bc2 Rae8 29.Rd1 Kf6 30.g3!? g5 31.g4 Kg7!? 32.Rd6 R5e6 33.Rd1 Rf8

How should White proceed?

*****

*****

*****

*****

*****

34.Rde1?!

The engines strongly dislike this, reckoning 34.b4 maintains equality (Stockfish17 is also fine with 34.h3, but Dragon1 reckons it leaves Black with a slight edge).

34...Rf4 35.h3

This looks natural, but the engines prefer 35.Bd1.

35...Kg6

The engines give 35...Kf6!? as much stronger, eg 36.Bd1 Ke5 37.Be2 Kd4 38.Bf1 a4, when suddenly White's queenside is vulnerable.

Going back to the last diagram, after the suggested 34.b4 there might come 34...axb4 35.axb4 Rf4, but 36.h3 is now fine, presumably because there is little chance of the black king attacking White's queenside.

This is really quite subtle stuff, and I do not suppose even a small fraction of it crossed either player's mind.

36.R1e2?!

The engines much prefer 36.b4 or 36.Rd1.

36...h5

They again like ...Kf6, and if 37.Rd2, then 37...Ke5, with ...Rd6 to come so the black king can gain access to the queenside.

37.gxh5+ Kxh5 38.Rg2?!

The engines suggest 38.Rd2, or 38.Bd1 followed by Rd2.

38...Rf3?

The engines reckon taking the open d file with 38...Rd6 is strong.

39.Bd1 Rd6 40.Bxf3+ Kxf3 41.Rf2

Possibly better is 41.Rg3, hoping to give back the exchange in return for the f pawn. However 41...Rd1+ 42.Kf2 Rd2+ 43.Kf1 a4!? (43...Rd1+? 44.Re1 Rd3 45.Re5) 44.bxa4 Ra2 seems to give enough counterplay.

41...g4 42.hxg4+ Kxg4

Now the f pawn is fully protected, and the position is completely equal, according to the engines, but tricky

*****

*****

*****

*****

*****

43.Re7 Rg6?

This lets White threaten to activate his king. Instead both 43...Kf4 and 43...Kg3 keep a lid on the position, while also holding are moves such as 43...Bd7 and 43...Rd1+ (meeting 44.Re1 with 44...Rd2).

White to play and (probably) win

*****

*****

*****

*****

*****

44.Kh2

After 44.Rh2! the only way to stop White's king getting active with 45.Kf2 is 44...Kg3, but the engines reckon 45.Reh7! wins, eg 45...Rd6 46.R2h3+ Kf4 47.R3h6!  Rd1+ 48.Kf2 Rd2+ 49.Ke1 Re2+ 50.Kf1 Be4 51.Re7 Re3 52.Rh4+ Kf5 53.Rf7+ Ke6 54.Rf8 Ke7 55.Rb8 etc.

This is a difficult line to foresee, and White has to find a lot of good moves, eg 47.R7h6, instead of 47.R3h6!, does not win, eg 47... Rd1+ 48.Kf2 Rd2+ 49.Ke1 Re2+ 50.Kf1 Be4 reaches a similar position, but 51.Re6, instead of 51.Re7, can be met by 51...Bf5 52.Rf6 Rb2, with sufficient counterplay, while the alternative 51.Rxb6, which at first might be thought an improvement, can be met by, among other moves, 51...Ke3, eg 52.Rd6 (52.c5? Rc2 53.Kg1 Bf5 54.Rg3 Rc1+ 55.Kh2 Kf4 wins for Black) Rf2+ 53.Kg1 Rg2+ 54.Kf1 (not 54.Kh1?? Rg4 with ...Ke2 and f2 etc to come) Rf2+ etc.

44...Rh6+ 45.Kg1 Rg6?

Giving Black another chance to play Rh2.

46.Kh2 Rh6+ ½–½

Muddling Along

FOLLOWING yet another game finishing in under two hours, I set off from the venue, Hotel Kaiseralm, again intending to take a scenic route to Bischofsgrün's cable car and use it to ascend the Ochsenkopf.

But I had not gone far when I saw a sign stating Fichtelberg was less than 10 kilometres away, and thought I would walk there instead.

I had seen the town of Fichtelberg on the map, and assumed it got its name from Mount Fichtelberg, which at 1,215 metres is higher than Ochsenkopf's 1,024 metres.

In fact I had got my geography in a muddle - Fichtelberg, the mountain, is not in the Fichtel range, but some 60 miles away in Saxony.

The walk started pleasantly enough with sun-dappled forest, albeit I was walking on a straight, gravelly trail 

As on the previous day, virtually the only colours to be seen were greens and grey-browns

The main exceptions to the colour duopoly were the occasional white butterfly and, here, foxgloves

It soon became apparent that, rather than taking a completely new route, I had merely cut off a corner of the day-before's journey along the White Main.

And since it seemed the whole way to Fichtelberg was going to be on a gravel trail, I decided to cut my losses and more-or-less repeat the walk of the previous day (this was probably a lucky choice, bearing in mind that, as I explained above, I had got my Fichtelbergs confused).

So once again I stopped for coffee and water at a lakeside hostelry, and again had lunch, this time pretzel, cheese and water, at the source of the White Main.

Speaking of choice, it's not that I had a limited number of routes to pick from 

A single file of foxgloves, seemingly marching down the mountainside

One change on this second ascent of Ochsenkopf was that the cafe at the summit was open, having been closed on Monday.

Rock Art

Maria Alm, a biergarten-style open space near the foot of Ochsenkopf, was also over Montag Ruhetag syndrome

A final crossing of the White Main, before returning to my hotel

Bischofsgrün Round Five

I WAS upfloated.

Spanton (1954) - Rüdiger Schönrock (1884)

Petrov

1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nf3 Nxe4 5.d3!? Nf6 6.d4 Be7!?

More popular in ChessBase's 2025 Mega database is 6...d5, transposing into the Exchange Variation of the French.

7.Bd3 Bg4 8.h3 Bh5 9.0-0 0-0

Now both sides have castled, how would you assess the position?

*****

*****

*****

*****

*****

White has more space in the centre, giving a slight edge, according to Stockfish17 and Dragon1.

10.c4 Nc6 11.Be3 Nb4!? 12.Be2 d5!?

After 12...Bg6 13.Na3, Black may be for choice, but better is the engines' 13.Ne1!?, and also better, as RS pointed out in the postmortem, is 13.Nh4. My planned 13.Nc3!? Nc2 14.Rc1 Nxe3 15.fxe3 loses the bishop-pair but strengthens White's centre, and also seems playable.

13.a3 Nc6 14.Nc3 dxc4!?

How should White respond? 

*****

*****

*****

*****

*****

15.d5!?

The engines dislike this, preferring 15.g4!? Bg6 16.Bxc4, claiming a slight edge for White. Also reasonable is 15.Bxc4 as 15...Bxf3 16.Qxf3 Nxd4 17.Qxb7 gives White at least a slight edge, according to the engines. However they reckon Black should meet 15.Bxc4 by getting on with development with 15...Qd7, giving what they reckon are equal chances.

15...Na5

The engines narrowly prefer this over 15...Bxf3 16.Qxf3 Ne5.

16.g4!?

The engines fluctuate between the text and 16.Qa4.

16...Bg6 17.Ne5?!

This is strongly disliked by the engines. They suggest 17.Qa4 b6 18.Bxc4, claiming equality.

17...Bd6 18.Nxc4?

Winning back the pawn, but allowing Black to exchange his offside knight. The engines give 18.Nxg6, but much prefer Black.

18...Nxc4 19.Bxc4 Ne4

Even stronger seems to be 19...Nxg4!?, when 20.hxg4 Nxc4 21.Bxc4 Qh4 leaves White in all sorts of trouble, eg 22.f4 Qg3+ 23.Kh1 Qh3+ 24.Kg1 Qxe3+ etc. Better may be 20.Nxa5, but 20...Nxe3 21.fxe3 Qxg5+ 22.Bg4 (forced, as 22.Kh1? loses to 22...Qg3) f5 gives Black a huge attack.

20.Nxe4 Bxe4 21.Qd4?!

Probably better is 21.f3, intending 22.Qd2, or immediately 21.Qd2, with the possibility of f3 to come.

21...f5!?

Protecting the light-square bishop and preparing to bring the king's rook into the attack against an increasingly vulnerable-looking white king.

How should White proceed?

*****

*****

*****

*****

*****

22.g5

This seems best. Definitely bad is 22.f3? as Black has 22...Qh4!, while 22.gxf5? can also be met by 22...Qh4.

22...f4

The engines' 22...Qe8 may be even stronger.

23.Qxe4

Black to play and win

*****

*****

*****

*****

*****

23...Qxg5+?!

This looks promising, but Black has a defence, whereas the simple 23...fxe3 leaves White in insurmountable difficulties, according to the engines, eg 24.Qxe3? Rf4! puts White's bishop and g pawn en prise, and after, say, 25.Rac1 Qxg5+ 26.Qg3 Qf6 27.Qe3 Kh8! (keeping queens on the board) White's position collapses. A better try is 24.f4, but the engines' 24...Qe8! is strong, eg 25.Qe6+ Kh8 26.Qxe8 Raxe8 27.Be2 Bxf4 gives Black a pawn advantage, and with more White weaknesses to target.

24.Qg2 Qxg2+

I was worried about 24...Qe5 25.Bd2 f3? 26.Qg3 Qd4?, missing 27.Bc3!, when White remains a piece up.

25.Kxg2 fxe3 26.fxe3 Rae8 ½–½