Future generations will visit this site, read the above comment and think.. 'I'm just so glad that valuable piece of information was available to us!?'

I ran SF10 for two hours (60 ply) after 42.a7. The main line was 42....Bc5 43.Rdd7 f5 44.Kf1 Ra1+ 45.Ke2 Ra2+ 46.Kd3 Bxf2 47.Rf7+ Kg8 48.Rfe7 Kf8 49.Rxe8+ Kxe8 50.Rb8+ Kf7 51.a8/Q Rxa8 52.Rxa8 (+1.02), with R+g and h pawns v. B and f, g, and h pawns in the ending. Continuing SF a few moves: 52....h5 53.Rh8 g6

click for larger view

I glanced at FCE, BCE, and Dvoretsky's endgame manual. If I'm understanding Dvoretsky p. 227 correctly (no sure thing), Black has an unbreakable fortress even without the f-pawn. So, although 42.Ra7 is definitely worse than 42.a7, I don't think it cost Tal a half-point.

I checked Tal's autobiography to see if he had anything to say about the game. Having lost to Torre, Estevez, and Huebner in the early rounds, he recounts a tragicomedy at p. 412:

<I was nevertheless playing 'creakily', with difficulty, and this led to my amassing 6(!!) adjourned games: a difficult position against Korchnoi, and better positions, as I thought, against Kuzmin, Karpov, and Tukmakov, as well as against Quinteros and Cuellar. If they were all to end up as I was planning, then all would not yet be lost in the tournament. But the resumption brought a serious correction, since it was impossible to analyse all the positions anything like thoroughly. We managed this only in the game against Tukmakov, finding a win at around three in the morning, and merely making a superficial acquaintance with all the others.

The resumption began with a loss against Korchnoi, which was more or less planned. Then came Kuzmin, and I suddenly saw that in my nocturnal analysis (at about 5 a.m.) I had made two moves in a row for myself(!!) which, of course, one could not count on doing in an official situation. I should have agreed a draw, but I thought for about forty minutes, and then, just at the point when I was ready to conclude peace, Kotov, the chief controller came up and informed me that Tukmakov had resigned without playing on. The thought flashed through my mind that if I were to draw with Kuzmin now, I would then have to play against Karpov, and my adjournment analysis had not gone as far as that position. So I decided to continue, and played on...to a hopeless position.

I agree with both <Olavi> and you in your assessment of the computer assessment. To me, the best move is the most difficult to respond to by a human, not the one that gets the highest score with both sides playing perfectly. Tal of all people knew that very well (although computers didn't exist in his day)... so many of his sacs were technically flawed, yet so hard to respond to.