Tuesday, 31 December 2024

Kraków Round Six

Spanton (2011) - Hanna Bartczak (1657)
Pirc
1.e4 d6 2.d4 Nf6 3.Nc3 g6 4.Nge2!?
A rare continuation - 12th-most popular in ChessBase's 2025 Mega database.
How should Black respond?
*****
*****
*****
*****
*****
4...c6!?
This is liked by Stockfish17 and Dragon1, along with 4...b5!?, but both lag well behind the obvious 4...Bg7 in terms of popularity.
5.h3!?
As will be seen, the idea of this is not just to prevent a black piece landing on g4.
5...b5 6.a3 Nbd7 7.g4!? Nb6 8.Bg2 Bb7!?
Possibly a novelty - the engines prefer it over the known 8...e5.
9.Ng3 Qc8?!
Taking the sting out of lines where White pushes the e pawn, followed by capturing on b5. But probably a better way is 9...Qc7.
10.b3 a5 11.Be3 Qc7!? 12.Qd2 a4 13.0-0 h5!?
This is Dragon1's top choice for a while, but it makes the black kingside brittle.
14.g5 Nfd7 15.f4 Bg7 16.f5 0-0!?
The engines reckon this is best. Perhaps the point is there is no safe haven for the black king.
Now both sides have castled, how would you assess the position?
*****
*****
*****
*****
*****
White's central and kingside space provide a springboard for attacking the black king. The engines reckon has a positionally won game.
17.Qd3 axb3 18.cxb3 Ba6 19.b4!?
This looks awkward, not least because it gives up the c4 square, but is effective.
19...Nc4 20.Nce2?
This throws away White's advantage. The engines reckon 20.Bf2 is correct.
20...Nxe3?
Black is at least equal after 20...h4, according to the engines.
21.Qxe3 Qb6?!
The engines do not like this, preferring 21...c5 (Stockfish17) or 21...Qa7 (Dragon1).
22.Rad1 Ne5?
This loses control of the f6 square.
23.f6 exf6
The engines prefer 23...Bh8!?
24.gxf6 Bh8 25.Qg5 Kh7 26.Kh1 Nd7
White's advantage is so large that the method of winning is largely a matter of taste
*****
*****
*****
*****
*****
27.Nxh5 Qd8 28.e5 Ra7 29.Be4 d5 30.Bb1 Re8 31.Rf4 Nf8 32.Rh4 Kg8 33.Qh6!? gxh5 34.Rxh5 Ng6 35.Bxg6 hxg6 36.Qxg6+ Bg7 37.Qh7+ Kf8 38.fxg7+ 1-0

Kraków Round Five

TOP board.

Andrzej Kawula (1922) - Spanton (2011)
QGD Tarrasch
1.d4 d5 2.c4 e6 3.Nf3 c5 4.cxd5 exd5 5.Nc3 Bf6 6.Bg5!?
Stockfish17 and Dragon1 fluctuate between the text and the Schlechter-Rubinstein 6.g3.
6...Be6 7.e3 Nc6
This is a popular continuation, but the engines much prefer 7...c4!? or 7...a6!?
8.Bb5 Be7 9.dxc5 0-0 10.0-0
Now both sides have castled, how would you assess the position?
*****
*****
*****
*****
*****
Clearly the most significant factor is the isolated queen's pawn, and the engines reckon that means White has the upper hand.
10...h6 11.Bh4 Bxc5 12.Rc1
Even better, according to the engines, is 12.Bxf6!? Bxf6 13.Nxd5 Qxb2 14.Nc7 Rad8 15.Qc1!? Ba3 16.Qxb2 Bxb2 17.Nxe6, when they claim at least the upper hand for White.
12...Be7 13.Nd4 Nxd4 14.Qxd4 a6 15.Ba4 Rc8 16.Bb3
Black threatened 16...Rc4.
16...Qa5 17.Rfd1 Rfd8 18.Qd3!?
How should Black proceed?
*****
*****
*****
*****
*****
18...d4!
The engines agree this pawn sacrifice equalises.
19.exd4 Bxb3 20.axb3 Qb4?
But this is the wrong follow-up. The engines like 20...Rc7, intending ...Rcd7.
21.Na2
Even stronger is 21.Bxf6! Bxf6 22.Nd5, when 22...Qd6 23.Rxc8 Rxc8 24.Nxf6 Qxf6 gives White excellent winning chances, thanks to the passed d pawn.
21...Qa5?!
Almost certainly better is 21...Qb6, keeping pressure on both b3 and d4.
22.Rxc8 Rxc8 23.Nc3 Rd8 24.Ne4!?
There seems nothing wrong with immediately pushing the d pawn.
24...g5 25.Bg3 Nxe4 26.Qxe4 Qb4 27.Qd3 Rd5 28.Qc4 Rb5?
The engines reckon 28...Qb5 more-or-less equalises, and if 29.Qxb5 Rxb5 30.d5 then 30...Kf8 holds - White does not seem able to make progress.
29.Qxb4?
White returns the favour. Instead 29.d5 is promising, but the engines agree that definitely winning is 29.Bc7!, the point being to support the d pawn on its journey to d8, and if 29...Qxb3 then 30.Qxb3 Rxb3 31.d5 wins a piece.
29...Rxb4
Possibly better is 29...Bxb4!?, meeting 30.d5 with 30...Ba5 or 30...Kf8.
30.Rd3
White can afford to sac the b3 pawn with 30.Bc7!, and the engines reckon 30.d5!? is also better than the text.
30...Kf8 31.Kf1 Ke8 32.Ke2 Kd7 33.f3 Bf6 34.Ke3 Bg7 35.Be1 Rb5 36.g4!?
The engines like this.
How should Black respond?
*****
*****
*****
*****
*****
36...f5
This is the engines' second choice, but they prefer keeping lines closed with 36...Rb6!?, and if 37.d4 then 37...Re6+.
37.gxf5 Rxf5 38.Bg3 Rd5?!
The engines strongly dislike this, preferring 38...Bf6, and if, as in the game, 39.Ke4 then 39...Rb5.
39.Ke4 Ke6 40.Bf2
Even stronger, according to the engines, is 40.h4!?, and if 40...Bf6 then 41.h5.
40...Rb5?!
Another move much disliked by the engines. They suggest 40...Bf6, claiming White has no more than a slight edge.
41.Rc3
The engines like this, but again prefer pushing the h pawn.
41...Bf6 42.Be3 Rb4 43.Bf2 h5?!
This is probably weakening.
44.Be3 h4?
It was better to keep the pawns flexible.
45.h3 Rb5 46.Rd3 Rd5 47.b4!? Be7 48.Bd2 Bf6 49.Bc3 Be7 50.Rd1 Bf6 51.Ra1 Bd8 52.Rg1 Bc7 53.Bd2
What should Black play?
*****
*****
*****
*****
*****
53...Rxd4+?
'Clever', but bad. It is not clear if White can make progress after, for example, 53...Bb6.
54.Kxd4 Bb6+ 55.Kc4?
White is winning easily after 55.Ke4 or 55.Kd3, according to the engines, eg 55.Ke4 Bxg1 56.Bxg5 Bf2 57.f4, and the passed pawn decides matters.
55...Bxg1 56.Bxg5 Bf2
How big is White's advantage in this this bishop-and-pawn ending?
*****
*****
*****
*****
*****
Dragon1 awards White the upper hand, but Stockfish17 reckons White only has a slight edge.
57.b5!? axb5+ 58.Kxb5 Kf5 59.Bd2 Ke5?!
59.Bg3 gives complete equality, according to the engines.
60.Kc4 Bd4?
This definitely loses, according to the engines, whereas 60...Kf5 and 60...Kf6 may hold.
61.b4
I cannot work out why, but the engines reckon much stronger is 61.b3.
61...Bf2
And here Stockfish17, but not Dragon1, reckons 61...Bb6!? is much better.
62.b5 Bb6?!
But now this is apparently much weaker than 62...Kf5.
63.Bg5 Bf2 64.Kd3 Kf5 65.Bd2 Ke5 66.Ke2 Bg3 67.Be3 Kf5 68.Kd3 Ke5 69.Ba7 Bf4 70.Bb6
White only draws after 70.Bb8+ Kf5 71.Bxf4?
70...Kd5 71.Bf2 Bg5 72.Be3 Be7 73.f4
The advance of the passer is decisive.
73...Bf6 74.Bf2 Be7 75.Bd4 Bd6 76.f5 Bg3 77.f6 Ke6 78.Ke4 Be1 79.Kf4 Bd2+ 80.Kg4 Be1 81.Kg5 Kf7 82.Kf5 Bg3 83.Be5 Be1 84.Bf4 Bf2 85.Bg4 Be1 86.Kg4 Kg6 87.Bxh4 (1-0, 92 moves)

That Was The Year That Was 2024


I started the year where I was to finish it, in the old Polish capital Kraków, which legend has it was once terrorised by a fire-breathing dragon 
After a 4NCL weekend in Telford, I played at a CzechTour seniors' tournament in the spa town of Mariánské Lázně 

The English Chess Federation's 2nd Cambridge International was held in February in the city's University Arms hotel

In early March my collection of chess pens received a boost at the Bavarian spa town of Bad Wörishofen 

Also in March I enjoyed a late winter in Fagernes, Norway

I was back in Bavaria, beside the picturesque Lake Tegernsee, in April

Austria's Bregenz, on the Bodensee (aka Lake Constance), holds open and senior tournaments in May
Later that month the English seniors' championships were held at the Holiday Inn in Kenilworth, Warwickshire
Bronze Age graves were among the highlights when I took a break from chess to enjoy a walking holiday in Sardinia in June
Davos in Switzerland was a long way to go for a June weekender, but I had a few days left of my Schengen allowance, and I was interested to see what the famous resort looked like (the journey by train had stunning views - the town was so-so)
Playing in the Bavarian resort of Bischofsgrün in June allowed me to walk to the source of the River Main
A tree growing from the arched doorway of monastery ruins was a highlight of Bad Herrenalb in the Black Forest in early July
Hull, where the British championships were held in July-August, has a pub with a model train set running round one of the bars

Olomouc in Czechia in August has long been one of my favourite events

Darlington, whose parish church of St Cuthbert partly dates back to the early 1100s, was inevitably more picturesque than its reputation would suggest

At the end of August and the first week of September I played in the Riviera congress at Torquay, which is still a commercial fishing port

Utah's Bryce Canyon was one of many stunning highlights when later in September I took another break from chess

The Amateur Chess Organization's Crete event in October had an idyllic setting, but my enjoyment was spoilt by playing-tables set unnecessarily close together - very annoying when paying top whack 

No such problems later the same month at Mallorca's Colonia de Sant Jordi, but there was no heating and my room got chilly in the evenings - again especially annoying when paying a hefty rate

Cap Negret on the Costa Blanca in November is one of the newest seniors' events

Benidorm was brought forward a little this year, and included two double-round days at the Gran Hotel Bali, but sunshine in late November and early December is hard to beat

I played a weekender in Coulsdon, Surrey, in mid-December, before finishing the year where I started, in Kraków

Monday, 30 December 2024

Kraków Round Four

I WAS downfloated.

Spanton (2011) - Stanisław Tomalak (1750)
Philidor/Pirc
1.Nc3 Nf6 2.d4 d6 3.e4 Nbd7 4.f4 e5!?
Technically the opening is considered to come under the Pirc banner, but the idea of playing ...e5, while allowing White to deprive Black of castling rights, is common in the Philidor, although the Philidor arises after the moves 1.e4 e5 2.Nf3 d6, usually ruling out an early f4
*****
*****
*****
*****
*****
5.dxe5!?
Theory considers 5.Nf3 a more promising way of trying for a significant advantage.
5...dxe5 6.fxe5 Nxe5 7.Qxd8+ Kxd8
There are 356 examples of this position in ChessBase's 2025 Mega database - who stands better, if anyone, and by how much?
*****
*****
*****
*****
*****
Both Stockfish17 and Dragon1 reckon the position is equal - indeed Stockfish17, at least for a time, has Black 0.04 ahead. I guess the point is that losing castling rights is not such a big deal with queens off the board, and Black has a slight lead in development, while White has an isolated pawn.
8.Bf4
This is the most popular continuation in Mega25, but the strongest players have tended to prefer 8.Bg5 or 8.Nf3.
8...Bd6 9.0-0-0 Ke7 10.Nf3!?
How should Black proceed?
*****
*****
*****
*****
*****
10...Nxf3!?
Dragon1 at first regards this as a mistake, while Stockfish17 is happy enough with it. However, given enough time, both engines come to award White a slight edge, whereas they reckon winning the bishop-pair with 10...Nd3+! keeps the game equal.
11.Bxd6+ cxd6 12.gxf3 Bd7
The engines suggest 12...Be6!?, and if 13.Nb5 then 13...Rhd8 or 13...d5.
13.Bg2!?
A 'mysterious' bishop move, but there really is not a great alternative destination for White's light-square bishop.
13...Rhd8 14.Rhe1 Bc6
What should White play?
*****
*****
*****
*****
*****
15.Nd5+?!
White is at least slightly better after 15.Rd4 or 15.f4, according to the engines.
15...Bxd5
More-or-less forced, but good.
16.exd5+ Kf8 17.Re2?!
Stockfish17 suggests getting the bishop on on decent diagonal with 17.Bh3, while Dragon1, with a similar idea in mind, prefers 17.Bf1!?
17...Re8 18.Rxe8+ Rxe8 ½–½
I could not justify playing on. White's only decent move, according to the engines, is 19.c4, although they reckon that leaves Black with the upper hand (Dragon1) or at least a slight edge (Stockfish17), while they reckon every other move gives Black a positionally won game.

Playing Conditions


Before yesterday's third round - all three tournament rooms are similar, with plenty of room back-to-back, and adequate space side-by-side

Sunday, 29 December 2024

Kraków Round Three

Bogusław Czepczyński (1793) - Spanton (2011)
English Symmetrical
1.c4 c5 2.Nc3 g6 3.Nf3 Bg7 4.g3 Nc6 5.Bg2 e6 6.0-0 d5!?
Of the 4,921 games to reach the position in ChessBase's 2025 Mega database, 4,734 saw 6...Nge7.
How should White respond?
*****
*****
*****
*****
*****
7.cxd5
This is most popular, and scores 61% in Mega25, albeit from a fairly small sample. Stockfish17 suggests 7.Na4!?, and Dragon1 likes 7.d4!?, neither move occurring in the database.
7...exd5 8.d4!?
The engines prefer this aggressive continuation over the slightly more popular 8.d3.
Black only has one decent reply, according to the engines
*****
*****
*****
*****
*****
8...cxd4!
Yury Atanazevich (2045) - Vladimir Gritsenko (2264), Veronezh Regional Championship (Russia) 2014, went 8...Nxd4? 9.Nxd4 cxd4 (9...Bxd4 is not an improvement) 10.Nxd5 Ne7? (better are 10...Nf6!? and 10...Be6, but both can be successfully met by 11.Qa4+) 11.Bg5 f6 12.Nxf6+ (1-0, 28 moves). Also bad is 8...Bxd4?, while 8...c4?! is positionally suspect in this type of position, one continuation being 9.b3!? cxb3 10.Qxb3 Nge7 11.Ba3 0-0 12.e3, when White's pressure against d5, and on the queenside, give the upper hand, according to the engines.
9.Nb5 Qb6?
Trying for too much. Black should develop the king's knight and get castled.
10.Nd6+
Somewhat surprisingly, at least to me, this seems to be a novelty. The known move, from a 1975 Romanian championship semi-final, saw 10.Qb3?!, when the engines reckon 10...Nge7 would have come close to equalising (the game was drawn anyway).
10...Ke7 11.Nxc8+
11.Bf4?! Nf6 leaves White with, at best, a slight edge, according to the engines.
11...Rxc8
How would you assess the coming middlegame?
*****
*****
*****
*****
*****
White has won the bishop-pair, and Black's extra pawn is likely to prove temporary, after which d5 will be even weaker than it is now, but the most significant factor is probably the awkward position of the black king. The engines give White the upper hand.
12.e3!?
The engines suggest queenside play, eg 12.a4 Nf6 13.a5!? Nxa5 14.Nxd4 Nc4 15.b3 Nd6 16.Be3 Qd8 17.Rxa7 with continuing pressure.
How should Black proceed?
*****
*****
*****
*****
12...d3
Trying to keep lines closed, but probably a better way to give back the pawn is 12..Nf6!? 13.exd4 Rhe8. Almost certainly too greedy is 12...dxe3?! 13.Bxe3 d4 14.Bf4.
13.Qxd3 Nf6
This is almost certainly better than 13...Nb4?! 14.Qa3 Nf6 15.Bd2 a5 16.Bc3!? Rhd8 17.Qb3 Qd6 18.Bh3, when Black remains uncoordinated, and is positionally lost, according to the engines.
14.Bd2 Rhd8 15.Qa3+ Ke8
The black king has had to stay in the centre
*****
*****
*****
*****
*****
16.Bc3 Bf8 17.Qa4 Ne4 18.Ne5?!
The natural 18.Rfd1 gives at least a slight edge, according to the engines, but they reckon 18.Bd4?! Qb4 19.Qxb4 Bxb4 is equal.
18...Nxc3 19.bxc3 Bd6?!
Both 19...Qa5 and 19...Bg7 equalise, according to the engines.
20.Ng4!
The engines come to marginally prefer this over the more obvious 20.Nxc6, when 20...bxc6? is met by 21.Bxd5, but they reckon 20...Qxc6 leaves White only slightly better.
20...Kf8?!
Breaking the pin on the c6 knight, but 20...Be7 is probably preferable.
21.Bxd5
Even stronger, according to the engines, is 21.Nf6!?, but the text works out well for White.
21...Bxg3??
Pressure against the black king means White is better after 21...Qc5, but there is a lot of play left.
22.Bxc6 Bxh2+ 23.Kxh2 Qxc6
How big is White's advantage?
*****
*****
*****
*****
*****
White has knight for pawn, but the advantage is worth a minor piece (Dragon1) or at least close to three pawns (Stockfish17).
24.Qb4+!?
More pragmatic is 24.Qxc6, eg 24...Rxc6 25.Rfd1!? Rxd1 26.Rxd1 Ke7 (26...Rxc3 27.Rd7 increases White's advantage, according to the engines) 27.Rb1 b6 28.Rb3, when White stays up knight for pawn.
24...Kg7 25.Qf4 Rd5!?
This is probably not best, but it threatens counterplay, and obliges White to be careful.
26.Kg3
The engines reckon best is the cool 26.Rad1!?, and if 26...Rh5+ then 27.Kg3.
26...Rf5
The engines prefer 26...h5 or 26...g5!?
27.Qd4+ f6 28.f4
Even stronger, according to the engines, is 28.e4!?, and if 28...Rc5 then 29.Rad1!? Rxc3+ 30.Ne3. However they reckon best for Black is 28...Qxc3+!? 29.Qxc3 Rxc3+ 30.Ne3, although they agree White is winning.
28...h5 29.Nf4 h4+!? 30.Kh2
30.Kxh4? seems to give Black a draw, and quite possibly more in practical play.
30...Rd5
There is no perpetual after 30...Qf3? 31.Qd7 Kh6 32.Ne4!
31.Qd4 Re8 32.Qf3
Not 32.Qg2? Rxe3.
32...Qc5!? 33.Rae1 Rd2!? 34.Qxb7+ Re7
White to play and lose
*****
*****
*****
*****
*****
35.Kg1??
BC had about two minutes, plus the 30-second increment, to make the 40-move time control.
35...Rxb7 0-1

A Little History

A TINY village on Wawel Hill, beside the Vistula river, grew into the Polish capital of Kraków by 1038, holding that position until the royal court was transferred to Warsaw in the late 1500s.
Kraków was destroyed by Mongols in 1241, but was rebuilt, only to be ravaged again in 1259, although a third attack in 1287 was beaten off thanks to improved fortifications.
The city's barbican, in front of St Florian's Gate, looms out of today's early-morning mist
The barbican was erected in the late 1400s in preparation for an expected Turkish attack following an abortive Polish invasion of the Ottoman empire's client state Moldavia.
That counterattack never reached Kraków, but the barbican saw action in subsequent invasions by Swedish, Transylvanian and Russian troops.
According to Wikipedia, the barbican is considered by experts to the best preserved of three such surviving European buildings.

Saturday, 28 December 2024

Kraków Round Two

Spanton (2011) - Jagwida Gasik (1752)
Caro-Kann Bronstein-Larsen
1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.Nxe4 Nf6 5.Nxf6+ gxf6!?
This formerly popular recapture has fallen from favour since the dramatic rise of ...exf6.
6.c3 Bf5 7.Nf3
How should Black proceed?
*****
*****
*****
*****
*****
7...Nd7!?
More popular in ChessBase's 2025 Mega database are 7...Qc7 and especially 7...e6. Stockfish17 and Dragon1 fluctuate between the latter and the text.
8.g3!?
This is easily the commonest continuation in Mega25, despite the white light-square bishop already having an open diagonal.
8...Qb6!?
Five moves are more popular in Mega25, but there is no consensus over which is best.
How should White proceed?
*****
*****
*****
*****
*****
9.Bg2
Consistent, but the engines prefer the more popular 9.Qe2!?, which prevents Black from interfering with White castling kingside, and also defends b2, making developing White's dark-square bishop less problematic.
9...0-0-0?!
The engines reckon this is a serious error. They want Black to play 9...Qa6 or 9...Qb5, after which they reckon White is better, but much less so than after the text, which seems to be a novelty.
10.0-0 e5 11.Nh4!?
The engines prefer getting on with it on the queenside with 11.b4 or 11.a4.
11...Be6 12.Be3!?
Is the b2 pawn poisoned?
*****
*****
*****
*****
*****
No, is the simple answer, but the position is sharp.
12...Qxb2!?
This is Stockfish17's top choice; Dragon1 prefers capturing on d4 first, eg 12...exd4 13.Bxd4 Qxb2, but later switches to 13...Bc5.
13.Qa4!?
Possibly the wrong plan. Instead the engines suggest pushing the d pawn, either immediately or after a preliminary Rb1, eg 13.d5!? Nb6 14.Bxb6 axb6!? (14...Qxb6 15.c4 gives White the upper hand, according to the engines) 15.Rb1 with an unclear position, one line running 15...Qxc3 16.Rxb6 Bxd5 17.Bxd5 Rxd5 18.Qg4+ Kc7 19.Rfb1 Rb5 20.R1xb5 cxb5 21.Rxf6 (not 21.Rxb5?? Qc1+ 22.Kg2 Qc6+ etc) Qa1+ 22.Kg2 Qxa2, when Black is up two (doubled) pawns, but has much the weaker king, with the engines agreeing the position is completely equal.
13...Qa3 14.Bxc6!
The only way to avoid being much worse.
14...Qxa4
Not 14...bxc6?? 15.Qxc6+ Kb8 16.Rfb1 Nb6 17.Rxb6+! axb6 18.Qxb6+ Kc8 19.Qc6+ Kb8 20.Rb1+. However the engines' 14...Nb6!? seems to give equality.
15.Bxa4 a6!?
The engines reckon White is only slightly better after 15...Nb6 or 15...Kb8.
16.Bb3 Be7
The engines prefer 16...exd4!?. meeting 17.Bxd4 with 17...Bc5, and 17.cxd4 with 17...Nb6.
What should White play?
*****
*****
*****
*****
*****
17.d5?
A fundamental misreading of the position. White should get the rooks into play, eg 17.Rab1 or 17.Rfe1. The text gains space, but it gives Black the c5 square and lets Black activate her kingside majority.
17...Bh3 18.Rfd1 f5
The position is equal, according to the engines, but Black has an obvious attacking plan, while White has to defend accurately to maintain the balance 
*****
*****
*****
*****
*****
19.Ng2?!
I rejected the engines' choice, 19.Nf3!?, because of 19...Bg4, with ...e4 to come, but the engines point out the strength of 20.d6 Bf6? 21.Bd5, when 21...e4? 22.Rdb1 gives White a very strong attack. Instead they give 20...Bxf3 21.dxe7 Rde8 with what they reckon is an equal position.
19...Bd6
White's passed pawn is firmly blockaded, and now ...f4 is a major threat.
20.f4 f6!?
This may be a little slow, although the engines reckon Black still has a slight edge. They prefer 20...Rde8 or 20...h5. After the former, White could give Black three isolanis, two of them doubled, by playing 21.fxe5, but 21...Nxe5 leaves the white king looking very vulnerable.
21.Bc2 Rdg8
The engines suggest 21...Rde8 or 21...Kb8!?
22.Rf1 Rf8!?
Black is losing time. Instead 22...e4!? takes the tension out of the position, but effectively ends Black's kingside threats. The engines suggest 22...h5!?, when they reckon 23.fxe5 fxe5 24.Bxf5 Bxf5 25.Rxf5 h4 gives Black enough play for a pawn.
23.Rf2 Rhg8 24.Re1 Rf7 25.Nh4 e4
Black's last move was forced, but it fundamentally changes the dynamics of the position
*****
*****
*****
*****
*****
26.Ng2!? h5 27.Bd1 Rh7 28.Kh1 Nc5
Using the square that was made available to Black's pieces by 17.d5?
29.Be2 Bg4 30.Bxg4?!
This lets the black knight into d3, and helps Black improve her pawn-structure.
30...fxg4!?
The engines prefer 20...hxg4.
31.Rb1 Nd3 32.Re2 f5 33.Bd4!?
Dragon1 much prefers 33.Nh4, but Stockfish17 is indifferent.
33...Bc7
The engines suggest 33....Be7, preparing ...h4, or the immediate 33...h4!? (the idea of the latter will become apparent next move).
34.c4 h4!?
The game again becomes sharp
*****
*****
*****
*****
*****
35.Nxh4 Nxf4! 36.Rf2
Stockfish17 slightly prefers 36.Reb2!? Dragon1 prefers the text for quite some time, but eventually agrees with Stockfish17.
36...Rxh4!?
Also equal, according to the engines, is 36...Nh5!?, eg 37.Rxf5 Nxg3+! 38.hxg3 Bxg3 39.Kg2 Bxh4, although there is plenty of play left in the position.
37.gxh4 g3 38.hxg3
Another line giving dynamic equality is 38.Rg1 Nh3 39.Rxf5 Nxg1 40.Kxg1 gxh2+ 41.Kh1 Rg4!, but again the position remains sharp.
38...Rxg3?!
Almost certainly better is the engines' 38...Nh5!?, eg 39.Rg1 Nxg3+ 40.Rxg3 (forced) Bxg3 41.Rxf5 (41.Rg2?? loses to 41...f4 and 41...Rg4) Bxh4 42.Rf4 Bd8 43.Rxe4, when White is a pawn up, but Black has complete equality, according to the engines.
39.Rg1?
39.Rb3 gives at least a slight edge.
39...Rh3+ 40.Rh2 Rxh2+ 41.Kxh2 Ne2+
How should White continue?
*****
*****
*****
*****
*****
42.Kg2?
Blocking the rook is fatal. Instead 42.Kh3 Nxd4 43.Rg8+ holds, one point being 43...Kd7 44.Rg7+ Kd6?! 45.c5+! probably gives White the better winning chances.
42...Nxg1??
After 42...Nxd4 Black's passed pawns, supported by bishop and knight, should win easily, but the text gives a bishop-and-pawn ending that is very difficult for Black.
43.Kxg1
Material is level, but White can much more comfortably cope with Black's connected passers than Black can with White's far-apart passers
*****
*****
*****
*****
*****
43...Bf4 44.Kf2 Bh6 45.Bf6 Kd7 46.Bg5!? Bg7
The pawn-ending after 46...Bxg5? 47.hxg5 is an easy win for White.
47.Ke3 Be5 48.c5 a5?!
Allowing a pawn to be fixed on a dark square does not help.
49.a4 Bb2 50.Ke2
Also winning is 50.Kf4!? Bc1+ 51.Kg3.
50...Bd4 51.Be3 Be5 52.h5 f4 53.Bd2 f3+ 54.Kf2
Not 54.Ke3?? Bf4+! 55.Kxf4 f2 etc.
54...Bd4+ 55.Be3 Bf6 56.h6 Bh4+ 57.Kf1 Bf6 58.h7 Kc7!?
This is Stockfish17's top choice, briefly, which only goes to show everything loses without much chance of putting up a fight.
The game finished:
59.Bd2 b6 60.Bf4+ Kb7 61.c6+ Kc8 62.d6 1-0