Friday, 7 July 2023

Lessons From Prague 6

IN round six with white I reached the following position that Stockfish15.1 and Komodo14.1 reckon is completely equal.
Black has just played 40...Qe7-e2
*****
*****
*****
*****
*****
*****
*****
*****
I saw that 41.Qc5 leads to a simple draw, but instead I played 41.Qe3!?, reasoning that Black would probably play 41...Qc4, when 42.Qe5 is also a draw, although Black has to display a modicum of care.
Instead the game went 41...Qxe3!? 42.fxe3, which I had assumed, on general principles would be at least equal, and might well favour White as Black's pawn-structure is inferior.
However I had failed to properly take into account one of the most important factors in a pawn-ending, ie the activity of the kings.
After 42...Kf6 43.Kf2 Kg5 44.Kf3 Black keeps complete equality with 44...f5, but my opponent tried for more with 44...Kf5!?
Objectively this may be inferior, but I was worried about allowing ...Ke4.
I found the correct 45.g4+ hxg4 46.hxg4+ Kg5 47.e4 dxe4+ 48.Kxe4, while noting that after 48...Kxg4 White is a pawn down.
Following 49.d5 f5+ I faced a critical decision, although I did not realise it.
The game continued 50.Ke5? f4, after which both players queened and I failed to hold a pawn-down queen-and-pawn ending that endgame tablebases show is a draw.
Correct is 50.Kd4 f4 51.d6 f3 52.Ke3, which leads to 52...Kg3 53.d7 f2 54.d8=Q f1=Q.
As in the game, both sides have castled, and Black is a pawn up - but only temporarily 
*****
*****
*****
*****
*****
White can continue 55.Qd6+, after which White will win the h pawn.
True, the resulting position is drawn, as the Syzgy  tablebase shows, but the difficulty of Black's position can be gauged from the fact Komodo14.1 believes White is winning.
Throughout this ending I repeatedly misassessed the position, varying from undue optimism to undue pessimism, and rarely resorted to detailed concrete analysis.
LESSON: David Bronstein summed up the dangers of this type of faulty thinking when he said that "to lose one's objective attitude to a position nearly always means ruining your game."

Thursday, 6 July 2023

Lessons From Prague 5

IN round five with white I gained the bishop-pair and had the better pawn-structure, but Stockfish15.1 and Komodo14.1 reckon the following position is equal.
I have just played 17.c3
*****
*****
*****
*****
*****
Stockfish15.1 recommends 17...Nf6, claiming more-or-less complete equality, although Komodo14.1 is not so convinced.
The latter engine prefers 17...c5, although this time Stockfish15.1 doubts that the position is equal.
At any rate both moves are better than the 'clever' 17...Nc5?, which my opponent said afterwards he recognised as bad immediately on playing it.
The point is 18.Bxf5 wins a pawn, and although 18...gxf5 (the engines reckon this is better than 18...Nd7) 19.dxc5 cost me the bishop-pair and gave me a compromised queenside pawn-majority, White is very much on top.
LESSON: as with my round-four game, playing ...Nc5 is another example, albeit in a slightly different form, of 'automatic' thinking. White saw that immediately capturing the knight would not win a piece, and so assumed ...Nc5 did not lose material, but he missed Black's simple zwischenzug.

Wednesday, 5 July 2023

Lessons From Prague 4

IN round four with black I played an 'automatic' developing move that allowed my opponent to immediately gain a substantial advantage.
White has just played 8.Nd2
*****
*****
*****
*****
*****
*****
*****
*****
Reasonable is 8...Nxd2, while Stockfish15.1 and Komodo14.1 recommend 8...Bh3!?
Instead I played 8...e6? (technically this is not a developing move in that it does not develop a piece, but it opens a diagonal for Black's light-square bishop and it is normal to refer to such a move as developing).
After 9.Nxe4 dxe4 (9...Bxe4?! 10.Bxe4 dxe4 loses a pawn to 11.Qxd7+ Kxd7 12,0-0-0+ and 13.Rd4) 10.Qxd7+ Kxd7 11.0-0-0+ White was ahead in development and had a target at e4.
LESSON: there are genuinely forced moves in chess, and they usually should be played quickly, but playing an 'automatic' move quickly brings to mind the humorous title of  international master Willy Hendriks' popular improvement book, Move First, Think Later. Playing like that is a good way to come up with blunders.

Tuesday, 4 July 2023

Championship Chess

PLAYED this evening in round one of the Battersea club summer championship.

Spanton (1932) - Ariel Barrett (*)
Spanish Classical
1.e4 e5 2.Nf3 Nc6 3.Bb5 Bc5 4.c3 a6!?
This delayed putting-the-question-to-the-bishop is not particularly popular but is a speciality of Serbian grandmaster Ivan Ivanišević, at least at fast time-controls.
How should White respond?
*****
*****
*****
*****
*****
5.Bxc6!?
The mainline in ChessBase's 2023 Mega database runs 5.Ba4 b5 6.Bb3 Nf6 7.0-0 d6 8.d4 Bb6 with a slight advantage for White, according to Stockfish16 and Komodo14.1. However the engines prefer the text.
5...dxc6 6.d4
The engines give 6.Nxe5!?, and if 6...Qg5 then 7.0-0!? Qxe5 8.d4, claiming at least a slight edge for White.
6...exd4 7.Nxd4!?
We now have a pawn-structure closely associated with the Exchange Variation of the Spanish.
7...Nf6 8.e5
More in keeping with Exchange Variation ideas is 8.f3.
8...Ng4 9.0-0 0-0 10.f4?!
Almost certainly better is 10.Bf4.
10...f6 11.e6!?
Or 11.exf6 Nxf6, after which Black is positionally winning, according to the engines.
How should Black proceed?
*****
*****
*****
*****
*****
11...Bxe6?!
Probably best is 11...f5, eg 12.h3 Nf6 13.Be3 Nd5 with strong pressure against White's position.
12.f5 Bxd4+?!
Presumably hoping for 13.Qxd4? Qxd4+ 14.cxd4, after which 14...Bc4 is very good for Black, but almost certainly better is the immediate 12...Bc4, when 13.Qxg4 Bxf1 14.Kxf1 gives White two knights for rook and pawn, but 14...Re8 leaves Black with at least a slight edge, according to the engines.
13.cxd4 h5??
Black has maybe a slight edge after 13...Bc4 etc, according to the engines.
14.fxe6 Qd6!?
The engines reckon this is best.
15.Bxf4 Qxe6!?
They also approve of this, but Black is lost.
16.h3 Ne3 17.Qe2 1-0
*No ECF but 1726 Fide.

Summer Nights

AM playing tonight in the first round of a summer version of the Battersea club championship.
It is split into four sections, with a total of 46 players (so far - entries will remain open even after the tournaments are underway) and a time limit of 75 minutes plus a 15-second increment.
It is being held over five consecutive Tuesday evenings, although, due to prior commitments, I will only be able to play in the first and last rounds.

Lessons From Prague 3

IN round three I had a golden opportunity with white to beat a player rated almost 300 elo higher.
White to play and win
*****
*****
*****
*****
*****
*****
*****
*****
*****
White is a pawn up and has attacking chances against the black king.
Accordingly I played 26.Qd4? I knew Black could meet this with 26...Rf7, but I reasoned an exchange of rooks would not harm my position.
Stockfish15.1 and Komodo14.1 reckon this is good enough for a win, but later positional inaccuracies meant the game was drawn.
Instead I could have won on the spot with 26.g4, after which the black queen has only one move that does not leave it en prise, namely 26...Qb1+.
But then comes 27.Rc1, and if 27...Qxb3 then 28.Qd4 leads to mate, eg 28...Rf7 29.Rxf7 Kxf7 30.Qf6+ Kg8 31.Bh6 etc.
Admittedly 26.g4 is slightly counterintuitive at first sight as White moves a pawn in front of the king and weakens the white position on the f file, but these factors are trivial when compared with the benefits of the move.
LESSON: when an enemy piece, especially a major one, is short of squares, be extra sure to look at ways of hitting it.

Monday, 3 July 2023

When Only The Best Will Do

A new version of the free engine Stockfish is available at the creators' website.
An accompanying blurb about Stockfish16 states: "This updated version of Stockfish introduces several enhancements, including an upgraded neural net architecture (SFNNv6), improved implementation, and refined parameterization. The ongoing utilization of Leela’s data combined with a novel inference approach exploiting sparsity, and network compression ensure a speedy evaluation and modest binary sizes while allowing for more weights and higher accuracy. The search has undergone more optimization, leading to improved performance, particularly in longer analyses. Additionally, the Fishtest framework has been improved and is now able to run the tests needed to validate new ideas with 10000s of CPU cores."
One can only hope the engine is better than the creators' English, but the bottom line for most people is probably: "Stockfish continues to demonstrate its ability to discover superior moves with remarkable speed. In self-play against Stockfish 15, this new release gains up to 50 Elo and wins up to 12 times more game pairs than it loses."

Lessons From Prague 2

THE position in the first diagram below arises from a line in the Semi-Tarrasch that runs (minor transpositions are possible) 1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Nf3 c5 5.cxd5 cxd4!? 6.Qxd4 exd5 7.e4!? Nc6 8.Bb5 dxe4 9.Qxd8+ Kxd8 10.Ng5.
In this position I have always played 10...Be6, but I am aware 10...Ke8 and 10...Ke7!? are also possible
At Prague with black in round two I played the related line 1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Nf3 c5 5.cxd5 cxd4!? 6.Qxd4 exd5 7.e4!? dxe4!? 8.Qxd8+ Kxd8 9.Ng5 (see the game for the exact move-order), reaching the following position.
This is the same position as in the first diagram, but without the moves ...Nc6 and Bb5
Previously I have always continued 9...Be6, but, drawing on the analogy with the first diagram, I decided to keep the bishop-pair and played 9...Ke8?
It was only after playing it that I considered what would happen if White replied 10.Bc4, which is indeed what happened, after which White recovers his pawn and has the better position.
Note that, if in the first diagram, Black plays 10...Ke8, then 11.Bc4?! would be fairly pointless as Black has, among other satisfactory moves, 11...Ne5.
LESSON: similar positions are just that - similar, not the same. What applies in one need not necessarily be true in a similar but different position.

Sunday, 2 July 2023

Lessons From Prague

IN round one I had white against a junior rated more than 400 elo higher.
Nevertheless I soon achieved what Stockfish15.1 and Komodo14.1 reckon is a winning position.
Black to make his 16th move
*****
*****
*****
*****
*****
*****
*****
*****
White is a pawn up and has a promising attack against the exposed black king.
I think many club players would try to defend on the kingside, while hoping the presence of queens would later give chances of complications that a much lower-rated opponent could go wrong in.
However, Black presumably reasoned that being a pawn down is not his major problem - the possibility of getting mated is.
Accordingly he played 16...Qe8!?, which the engines agree is the best move.
There followed 17.Qxe8 Rfxe8 18.Bxg7 Kxg7,  after which the position is transformed.
Black remains a pawn down and is still losing, according to the engines, but there is no longer any danger of a mating attack
Not only is the black king safe, but it is slightly better placed than the white king for central action. Black also has the only active rook.
The game carried on for another 23 moves, but Black achieved a draw, something that I suspect not many people would have been willing to bet on from the position in the first diagram.
LESSON: seeking complications is not the only way to save a lost game. It is more important to identify the main problem with a position, and to rectify that.

Saturday, 1 July 2023

Summing Up Prague

MY score in the open of +2=4-3 gained exactly 33 Fide elo.
Sightseeing and fishing on the Vltava