I am pretty confident I have analysed more engine games than anyone else in the world, and I truly feel that my understanding of chess has benefited greatly from it. However, some aspects of engine play and evaluation are still difficult to grasp and internalise. I came across an instructive example of this while analysing a line of the Classical Pirc with 6…Nc6.

This blog entry can be read together with a video on my Silicon Road YouTube channel which looks at the lines covered here in a bit more detail (https://youtu.be/uGM6rV_s9AU) as well as a pgn (available here: http://cloudserver.chessbase.com/MTIyMTYx/replay.html) which contains all the analysis I have done of this line.

While analysing lines from Viktor Moskalenko’s latest New in Chess book “The Perfect Pirc-Modern” I noted that after 1.e4 d6 2.d4 Nf6 3.Nc3 g6 4.Nf3 Bg7 5.Be2 0-0 6.0-0 (Karpov’s classical handling of the white side) Nc6!? 7.d5

Position after 7.d5

Komodo Dragon 3.3 was very keen on 7…Ne5!?

Position after 7…Ne5!?

(Moskalenko focuses on 7…Nb8!? though he also gives a few lines to show the potential of 7…Ne5!?) Black’s intention is to reach this odd doubled e-pawns position after 8.Nxe5 dxe5

Position after 8…dxe5

Let me hasten to say that this is not the remarkable thing I’m talking about. Many 1.d4 players will recall a tricky line of the Dutch Leningrad (favoured by Nakamura) with a similar structure. 1. d4 f5 2. g3 Nf6 3. Bg2 g6 4. Nf3 Bg7 5. O-O O-O 6. c4 d6 7. Nc3 Nc6 8. d5 Ne5 9. Nxe5 dxe5

Comparison position from the Dutch Leningrad

What was remarkable to me was how the engines handled the white position in the Classical Pirc position.

Back to the Classical Pirc after 8…dxe5

Firstly, the engines identified that Black’s next move will be to play …e6xd5, undoubling Black’s e-pawn, and leaving Black with a pleasant and harmonious position. For that reason, they unanimously chose the barely-played 9.Bc4 as their main line.

Position after 9.Bc4

The idea is that Black has little compensation for the structural weaknesses after 9…e6 10.dxe6 Bxe6 11.Bxe6 fxe6.

Position after 11…fxe6

That’s again not too difficult to understand, although I always find it impressive when engines unanimously plump for a move that has not really touched the consciousness of human players! After 9.Bc4, then the engines all want the lovely Dutch-style plan 9…Ne8 intending …Nd6 and then a kingside pawn storm with …f5-f4 and …g5-g4

Position after 9…Ne8

Cool stuff! And this is where my intuition and understanding parts company with the engines! (which is where it gets interesting!)

My intuition / reasoning is the following:

Having played 9.Bc4 to dissuade 9…e6, I should keep my bishop on the a2-g8 when Black plays …Nd6, retreating either to b3 or a2 (if I play a4)

Position after 10.Be3 Nd6 11.Bb3

I don’t want to allow Black to gain too much space on the kingside, so I’ll meet …f5 with the typical riposte exf5 …gxf5, f4.

Position after 11…f5 12.exf5 gxcf5 13.f4

After …e4, I’ll go Bd4 and exchange off the dark-squared bishops.

Position after 13…e4 14.Bd4

Position after 14.Bd4

That’s not bad reasoning – indeed, toss the position above at Leela and Stockfish and they’ll give me +0.4 which isn’t too bad 😉 However, there’s a fly in the ointment when Black delays …f5 and strengthens his position on the queenside first. Take a look at this sample line from Stockfish (not perfect, just illustrative)

1.e4 d6 2.d4 Nf6 3.Nc3 g6 4.Nf3 Bg7 5.Be2 0–0 6.0–0 Nc6 7.d5 Ne5 8.Nxe5 dxe5 9.Bc4 Ne8 10.Be3 Nd6 11.Bb3

Position after 11.Bb3

11…a5 12.a4 b6 13.Qd2 Ba6

Position after 13…Ba6

14.Rfe1 Kh8 15.f3 Qd7 16.Kh1 f6

Position after 16…f6

Thinking of preparing further improvement with …Rf7 and ..Rg8 before finally playing …f5

As you can see, my chosen setup lacks any sort of momentum if Black doesn’t present an early target with …f5. I was thinking of somehow getting my c-pawn involved by moving the knight from c3 and playing c4-c5 but it doesn’t look too realistic against Black’s setup.

How do the engines do it?

Actually, the engines assess one key factor much more subtly than I did: namely that Black’s kingside counterplay can also provide the easiest entry point for White’s counterplay. So White doesn’t need to prevent …f5-f4 and …g5-g4. White needs to be able to absorb …f5-f4, place the white pieces to restrict the activity of Black’s pieces (for example …Ba6) and be ready to strike back on the kingside at the right moment.

This is the engine main line. I won’t cover all the subtleties of this line – I’m preparing a video on my Silicon Road YouTube channel for that. Here, I just want to get the main idea across.

9. Bc4 Ne8 10. Qe2 Nd6 11. a4

Position after 11.a4

Keeping the queen on e2 and bishop on c4 prevents a future …Ba6. Black can of course net the 2 bishops with …Nxc4, but firstly this knight is an important piece in assisting Black’s kingside play, and secondly after …Nxc4, Qxc4 Black needs to look out for attacks against the c7-pawn with e.g., Nb5.

11…f5 12. f3 f4 13. Bd2

Position after 13.Bd2

White simply absorbs …f5 and lets Black come on with …f4. It feels worrying to me, but not if you have a plan!

13…a5 14. Kh1 g5 15. Rad1 Qe8 16. Bd3

Position after 16.Bd3

Intuitively, I’m confused here! If I didn’t know better I’d be thinking I was looking at a silly computer from the 1980s, putting a rook on a central file in a blocked position and moving the light-squared bishop around aimlessly! Although I might confess that White’s rook on d1 is well-placed to meet …e6 (supporting a knight on d5 after dxe6) and that if …e6 is no longer dangerous, then the bishop no longer needs to be on c4! And after all, the bishop is better on d3 (supporting Nb5 and preventing …b6 and …Ba6) than b3.

But what plan does White have? What activity does White have?

16…c5 17. b3 Rf6 18. Rg1! b6 19. g3!

Position after 19.g3

That’s it! That’s what the engines have had in reserve all this time (and in many sub-variations as well). White’s counterplay will come against the advanced Black kingside. So much more efficient to let the black forces advance towards White’s position and then cut them down rather than make a huge effort to somehow open lines on the queenside by yourself!

This Rg1 and g3 idea is counter-intuitive because to avoid making pawn moves on the side where you are weakest, is a human defensive tenet! And yet I might have been able to guess the idea if I had thought back to this game analysed in depth in “The Silicon Road to Chess Improvement” 😉(https://tcec-chess.com/#div=sf&game=40&season=20)

Leela-Stockfish TCEC SuperFinal 20, game 40

In this Mar del Plata King’s Indian, instead of trying to blast through on the queenside as normal, Leela devised the plan to strike back at Black’s advancing kingside pawns… and it put Stockfish in grave danger!

Leela-Stockfish TCEC SuperFinal 20, game 40

Last thoughts

When I reflect on episodes like this, it strikes me that I hear the game with the wrong rhythm. I anticipate the struggle taking place within a relatively small timeframe / small number of moves and I place my pieces accordingly. I don’t mind dropping a bishop out of play on b3 because I expect a quick conflict with …f5 and I think I will quickly achieve a compensating positional gain with exf5, f4 and Be3-d4. Unfortunately, the struggle can be stretched out for much longer (Black plays …a5, b6, …Ba6 first)…

Position after 13…Ba6

…and then my plan looks short-sighted. I’m preparing for a battle that will never be fought!

The engines hear a longer rhythm, anticipating a struggle that extends over 10-15 moves rather than just 3-4 moves and their piece placement and plans reflect this. The light-squared bishop is kept on c4 (able to stay in play by retreating to d3 later), the initial break is absorbed (with f3) rather than immediately resolved.

Position after 13.f3

And the engines exploit the flexible idea that you can “Attack through the channel the opponent opened” (chapter 17 in “The Silicon Road to Chess Improvement” 😉)

Position after 19.g3

It’s certainly not easy to improve to the engine level, but I guess being aware of a shortcoming is always the first step on the road to fixing it!