Over the board, if you choose a dubious opening, you may score a quick win but against a well-prepared opponent you will end up in a bad or losing position more often than not. That’s the value of mainline openings: they are sound at all levels, and if you know some theory and the general plans and where your pieces and pawns should go, you will reach a decent middlegame where you can start to play chess. And that’s all you can ask from an opening.

In high-level correspondence chess, where the moves are backed by engine analysis, it’s a different matter. The top engines operate at the 3500 rating level. In so far as you can ascribe emotional states to a machine, they are ruthless: they will exploit any mistake, no matter how small.

So I have discovered, to my initial surprise, that some mainline openings are no longer playable in correspondence chess. Let me give two examples with illustrations from games played by Rodney Barking.

The King’s Indian Defence

Since February 2022 Barking has started 218 correspondence games. 173 are finished and 45 are ongoing. Of the finished games, Barking has won 57, drawn 111, and lost only 5. Of the losses, 2 were down to carelessness: Barking played out a linear conditional sequence but got the moves in the wrong order at the point of implementation and lost material. This will not happen again.

The other 3 losses, where Barking was simply outplayed over the board, were all on the Black side of the King’s Indian Defence. Despite relying on the magisterial 2-volume set King’s Indian 1 and King’s Indian 2 by GM Gawain Jones (published 2022), Barking was simply pushed off the board. In this opening Black concedes a significant space advantage. As the engines show, White’s best strategy is to concentrate his forces on the kingside and neutralise Black’s initiative, before gradually expanding all over the board. Computers have become very good indeed at exploiting a space advantage. Barking also relied on engine analysis, but did not always follow its recommendations where the engine departed from Jones’s ideas, and this did not work out well. I am not saying anything against Jones, who is possibly the world’s leading expert on the King’s Indian Defence, and whose books were surely checked with the best-available engines at the time.

White: Michael Moyses (1800), Black: Rodney Barking (1800), ICCF 2022
E94: King’s Indian, Classical Variation
Online game – https://lichess.org/QyQdEy2c
1.c4 Nf6 2.Nc3 g6 3.e4 d6 4.d4 Bg7 5.Be2 0-0 6.Nf3 Nbd7
This is Gawain Jones’s trademark and he analyses it in detail in his book. Of course the main line is 6…e5 followed by 7…Nc6 and Jones analyses that in much greater detail. I’m not ascribing the loss to the 6…Nbd7 line despite losing with it in another game as well. The third game I lost was in the 6…e5 main line.
7.0-0 e5 8.Be3 Qe7 9.Qc2 c6 10.d5 c5
The plan endorsed by Jones is to close the centre, slowing down White’s queenside play, and to prepare kingside expansion with …f5. Often Black will play …Ng8 and …Bh6 to exchange the bad bishop.

11.Kh1
Jones considers a range of 11th moves for White, but this is not one of them. It’s the prelude to a kingside build-up by White. Alarmingly, the engines already evaluate White’s advantage as over +1.00 for several different 11th moves.
11…Kh8 12.Rae1 Ng8 13.g4 Ndf6 14.h3 h6 15.Qc1 Nh7 16.Rg1 Bd7 17.Rg2 a6 18.a4 f5 19.gxf5 gxf5 20.exf5 Bxf5 21.Reg1 Rf7 22.Nd2 Ngf6 23.Qd1 Bxh3 24.Rh2 Qd7 25.Bh5 Rff8 26.Qf3 Bf5 27.Rhg2 Rg8 28.Bf7 e4 29.Qf4 Bh3 30.Rg3 Qxf7 31.Rxh3 Qd7 32.Rgg3 Raf8 33.Qh4 Qf5 34.Ncxe4 Nxe4 35.Qxe4 Be5 36.Rxg8+ Kxg8 37.Rxh6
Over the course of the game, White has steadily increased the pressure on the kingside and has now won a pawn. His advantage is now around +2.00. Black’s king is also very exposed. I give the remaining moves although the game is effectively over as a contest.

37…Qxe4+ 38.Nxe4 Nf6 39.Nxf6+ Bxf6 40.Kg2 Kg7 41.Rh3 Kg6 42.b3 Kf5 43.Kf3 Bg5 44.Bxg5 Kxg5+ 45.Ke3 Rb8 46.Rh7 b5 47.Rd7 bxc4 48.bxc4 Kf5 49.Rxd6 1–0

The Sicilian Najdorf 6.Bg5

This is one of the most exciting and heavily-analysed of all opening variations. Fischer chose it as Black and defended the Poisoned Pawn line where Black snaffles the White b-pawn and aims to weather the storm in potentially quite irrational positions. In practical play, Black often declines the pawn in favour of completing development, and a complex position arises with chances for both sides. The variation as a whole is fully-playable both over the board and in correspondence chess. However, what used to be the main line – a Black option at move 13 – is now regarded as a serious error. Watch and learn.

White: Rodney Barking (2000), Black: Luis A. T. Frazão Ferreira (1668), ICCF 2022
B99: Sicilian Najdorf, 7…Be7 main line
Online game – https://lichess.org/FRZOtJZX
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Be7
7…Qb6 leads to the Poisoned Pawn variation. Because it is so theoretical, it tends not to be seen as a practical choice in over the board chess, and not many opening books analyse it for Black. One exception is The Sharpest Sicilian by Georgiev and Kolev, but that dates from 2012 which is a lifetime ago in the theory of this opening. Nor are there any Chessable courses on the Black side of this variation. In contrast, 7…Be7 is well-covered from the Black side, both in Playing the Najdorf: A Practical Repertoire by David Vigorito (2019), and on Chessable in Lifetime Repertoires: Giri’s Najdorf Sicilian by GM Anish Giri (2020).
8.Qf3 Qc7 9.0-0-0 Nbd7 10.g4 b5 11.Bxf6 Nxf6 12.g5 Nd7 13.f5

Black has a major choice here. Currently the best is thought to be 13…0-0! when Black seems to be castling into a kingside attack, but theory suggests that Black can successfully defend these positions.
13…Nc5?
The question mark may seem to be audacious for what was the main line of this variation until relatively recently. Vigorito considers it briefly and dismisses it in these words: “The big theoretical main line used to be 13…Nc5 but this never appealed to me in view of: 14.f6 gxf6 15.gxf6 Bf8 16.Rg1. The ensuing variations can be quite irrational and Black’s king is always in danger, with Rg7 and various knight sacrifices in the air. Maybe in a correspondence game Black can hold his own, but over the board it looks too scary. One recent example went 16…h5 17.a3 Rb8 18.Re1 Nd7 19.Nxe6! with a winning attack for White in Vallejo Pons–Wynn, Bangkok 2016. No thank you.”

Well that was 2019. Roll on 2021 and the 2-volume set The Najdorf Bg5 Revisited by Lukasz Jarmula. This is the most detailed and up-to-date coverage of the Bg5 Sicilian from White’s point of view. His assessment is even more stark: “This is the old main line, which has been pretty much refuted in correspondence practice.” He demonstrates this in 8 pages of detailed analysis backed up by recent games, concluding that the line “…has been simply refuted by modern engines. The whole line is unplayable unless as a provocation in a rapid or blitz game against an unprepared opponent.”
14.f6 gxf6 15.gxf6 Bf8 16.Rg1 Bd7
I have also faced 16…h5 in a more recent correspondence game which is still ongoing. We are only at move 20. The engine evaluation in that game is +1.60 and I am already winning.
17.Rg7 Bxg7 18.fxg7 Rg8 19.e5 0-0-0 20.exd6 Qb6 21.a4

White has sacrificed the exchange for an irresistible attack. 21.a4 opens up the queenside. The evaluation is now +2.50 and the game is effectively over.
21…Qb7 22.Qh5 f5 23.axb5 axb5 24.Bxb5 Rxg7 25.Qe2 Qb6 26.b4 Ne4 27.Qc4+ Kb8 28.Bxd7 Rdxd7 29.Nc6+ Kb7 30.Na5+ Kb8
Although Black has a slight material advantage, his king is completely exposed and is no match for the marauding White pieces.
31.Nxe4 fxe4 32.Qxe6 h5 33.Qd5 e3 34.Nc6+ Kb7 35.Ne7+ Kb8 36.Rf1 Rxd6 37.Rf8+ Kc7 38.Qc4+ Rc6 39.Rc8+ Kd7 40.Rxc6 Qa7 41.Qe6+ 1–0
It’s mate next move although Black could have resigned a long time ago.