Much well-deserved praise has been heaped upon David Bronstein's book of the 1953 Zurich/Neuhausen Candidates Tournament. It is not just a record of an important chess competition; it is a book of chess ideas.
"The author has tried not to overload this book with variations," wrote Bronstein in the preface to the first edition (1956.) "Variations are interesting if they demonstrate the beauty of chess; they are useless if they exceed the limit of what a human being can calculate during a game; and they are harmful if they are offered as a substitute for investigating and explaining positions where intuition, fantasy, and talent decide the outcome."
I have to say right now that I do not completely agree with this statement. The accurate calculation of concrete variations is fundamental to playing chess at any reasonable level; therefore concrete variations, whether beautiful or not, must be fundamental to understanding the game. And isn't one of the goals of analytical work that of extending the limits of what one can calculate? Nevertheless, in reading Bronstein's book I allow myself to suspend this minor disbelief and enjoy the stream of chess ideas flowing from his very creative pen.
Some time ago I read the claim that, despite years of scrutiny by Soviet players and analysts, only two or three analytical errors had ever been found in Bronstein's Zurich book. With most of my chess library in storage at the moment, I'm temporarily unable to track down the source of this claim, but I believe it was made by Boris Vainshtein in his biography of Bronstein entitled Improvisation in Chess.
Time to look at the first position. This is a snapshot from the game Boleslavsky-Averbakh, played in round 13, and shows the position after White's move 16.e4!? Bronstein writes that this advance would not be considered by most players because it weakens the d-pawn, and besides, where is the weakness in Black's position that would justify this concession? It is the unfortunate position of Black's queen, he says, and he gives the following variation to prove this: 16...dxe4 17.Nxe4 Qg6 18.Bd3 Bf5 19.Nh4 Qh5 (see the second diagram) 20.Nxf5 Nxf5 21.Nxd6 Nxd6 22.Qxc7, and White wins an important pawn.
If you haven't already done so, try to spot something stronger for Black than 20...Nxf5. Yes, it's a queen move.
So where does this leave Bronstein's variation? On the slag heap. After 19...Qh5, White has nothing better than 20.Nxd6, but this move order allows Black the interpolation 20...Bxd3 21.Qxd3 Rxd6 22.Nf3 c6, and I'm with the people who were worried about White's d-pawn.
Instead of 18.Bd3 White can play the immediate 18.Nxd6, and although he might be a bit better after 18...Qxc2 19.Rxc2 cxd6, Black doesn't look to be in any serious trouble.
This analytical error is not one that was found by any of those Soviet players or analysts. How do I know this? Because it is repeated in all three Russian editions of Bronstein's book. Hard to believe, but nevertheless true.