Anyway, I just did it, and the computer thinks both 18. Be4 and 18. Rfe1 give White only a slight advantage, but after 18. Be4 Black should reply <18...Ngf6> first, which leads to 18...Ngf6 19. Bb1 Bb5 20. Rfe1 Rac8.

No, I didn't. Having taken a look, there is an <analysis> button with greater analysis available to premium members.

<...computer thinks...but after 18. Be4 Black should reply 18...Ngf6 first, which leads to 18...Ngf6 19. Bb1 Bb5 20. Rfe1 Rac8...>

Geez, Capablanca's not very helpful here. I mean come on. He can't give even a bit of thought on blacks most logical 18...Ngf6 reply? Putting a question to the B and trying to maintain the N outpost.

Have the feeling Capa wasn't thinking 19. Bb1 in reply. Seems to make 18. Be4 kind of useless as it doesn't look like it causes black weakness. So is the idea 18...Ngf6 19. Bxd5 Nxd5 and ...?

After 18. Be4 Ngf6 19. Bxd5 Nxd5 20. Ng4 Rfc8 21. Qh6+ Kg8 22. Ng5 and only now 22...f6 23. Nf3 Qg7 24. Qd2 h5 25. Ne3 Bc6, the computer eval is exactly 0.0. Yes, Black has advanced his kingside pawns, but the area looks solidly defended and I see no serious winning attempt for White.

click for larger view

Yeah, unless there's mate or black has to give up something due to significant threats vs. the King, there's probably nothing for white in the Be4 toward Qh6 line. Hard to see a crushing attack without either white B though.

On the other side, Blacks outpost on d5 probably ensures equality and if there is no real white attack, the Qh6 looks a bit silly.

Don't have a problem with white playing the line but feel that Capablanca showed little analysis integrity with his dodge of "I leave it up to the reader to work it out" when looking at 18.Be4.

Maybe he could've provided a hint at what he was aiming for? Maybe a couple of possible continuations with some sort of conclusion? No. OK. But then the old "Oh, it's far too complicated and deep for me to explain." That's pretty sad.

Get the feeling that while most of the old masters where great players. With play deserving of respect equal to those greats today. Much of their published analysis appears weak and self serving. Not even close to the far more intellectually honest self-review in the modern era. Maybe starting with Botvinnik.

This reminds me of a story I read once in the press about an encounter between some journalists and Kasparov.

When they asked him to explain his moves his answer was something like this:

He said that he couldn't explain his moves to them because their knowledge and skill was not advanced enough. In order to understand his explanation their own knowledge and skill would have had to reach a stage sufficiently near his own to enable them to understand the terms or references made in his explanation.

Out of the few chess books I have, Fischer, Botvinnik, Petrosian, and even Kasparov all provide very useful and some fairly deep analysis. Haven't noticed any of them punting on especially key moments in the play. And much discussion on positions seemingly more complicated than 18. Be4 in this game.

Also didn't notice any warning label on the books that said for GM's only. Reminds me something GM Larry Evans said to me, referring to a copy of Fischer's MSMG's. Said, paraphrased, "If you want to get good, find every instance of a Fischer analytical conclusion, like "better", "with a bind", "with pull" etc., and if you can figure out how he came to that conclusion; you'll be good."

Good advice. But might be much more difficult to undertake when the analytical conclusion is as Capa prefers, "So I leave it up to the reader to work it out by himself."

Capablanca said that he had thought about the move for a very long time.

As 18...Ng8-f6 seems one of the obvious alternatives this suggests that Capablanca had considered it.

I suggest that after 18...Ng8-f6 the right procedure for White consists of moves that are so difficult as to appear senseless or incomprehensible to the ordinary player.

This suggests that from the point of view of the ordinary player after 18...Ng8-f6 the winning procedure consists of moves which are senseless or incomprehensible, moves which an ordinary player will therefore not consider.

After 18.Be4 Ngf6, that computer prefers 19.Bb1 with 0.0 evaluation.

I understand your sentiment.

14...Nfd5 much better than ...Nbd5

15...Ng8 slightly better than ...Bd7

16. SF doesn't understand the position I think. 16, a3, Qb3, h4 seem best rated around 30 plies.

17.h4 slightly better than Bxe7

18. Run longer as Capablanca gave it a long thought, not stable @34 ply. 18. Rfe1.37, h4 .22, a3 .07.

18...Ngf6, ...Rfc8, ...f6 all 0.0 30 ply.

18...Bb5 19,Rfe1 +1.00. Looks like C missed best moves as did A.

Capablanca was on record as believing in playing right moves and not trying to bluff one's opponent, and that's generally what we see in his games.

There are many examples of Capablanca being self-critical when annotating his own games, pointing out what he thinks are his mistakes and why he made them.

Capablanca was not a practitioner of deep analysis when annotating games. He's even on record saying that he was a better analyzer over the board than post game. I doubt he even did any new analysis when writing these annotations.