Two years before this game, at the 1949 USSR championship, Bronstein (as White) defeated Ragozin in this opening. Ragozin played 7...a6 in that game, and Bronstein's win ( Bronstein vs Ragozin, 1949) made a real impression on Botvinnik, who later wrote about it in his red notebook:

<"Sicilian a la Rauzer (Bg5, Qd2, 0-0-0). With the energetic e4-e5 and Be4! forced a dark-square weakness at d4. After f7-f5 again captured en passant - meanwhile, the maneuver of the black knight to e4 was, evidently, nothing to be afraid of - gave the opponent probable chances. By skilfully maneuvering and exchanging queens (on the 26th move!) prevented e6-e5. Exploiting Ragozin's passive play, subtly won the endgame (without counter-chances!!) after a lengthy adjournment session. But he also schemed! In general played well! Very accurately!"> (thanks to Resignation Trap for posting this)

Brontein went to co-win the 1949 USSR Championship (with Smyslov) ahead of a strong field, and Botvinnik took note of him well before he was the official challenger for the title.

Fast forward to 1951, and Botvinnik played a training match with Ragozin just weeks before his World Championship match with Bronstein. He surely asked Ragozin to play this way with White against him, knowing that Bronstein would be confident with it after his impressive win vs Ragozin in 1949. In the training match Botvinnik tried out 7...h6 and was successful (Ragozin vs Botvinnik, 1951). He wished to avoid the main line (7...a6) against Bronstein, who had played so well against it. So here in this game he played 7...h6 and scored his first win of the match.

Fast forward again. In 1954, Petrosian decided to out 7...h6 for Black but was defeated by Ivkov (Ivkov vs Petrosian, 1954). Then in 1956, Keres beat Botvinnik in this line (Keres vs Botvinnik, 1956). Botvinnik decided to switch to the main line, 7...a6, for his World Championship match with Smyslov in 1957.

In preparation for the Smyslov match, Botvinnik played another training match -- this time against Averbakh (he was friends with Ragozin but I think he realized he needed a stronger opponent in his training matches) -- and probably asked Averbakh to play this way for White. Botvinnik played 7...a6 twice against Averbakh, winning both times (only one of those games is listed in the CG database (Averbakh vs Botvinnik, 1957), but there were two). So he used it in his match against Smyslov and this time scored his first win of the match with *that* line (Smyslov vs Botvinnik, 1957).

I wasnt sure about 50 f6 but I had the feeling Bronstein was playing to win. But when 57 Kc2 came up I almost fell off my chair.

Sounds like another Bronstein fairy tale.

<57.Ne6 followed by Nd4 was enough for a draw, but Bronstein forced the game with 57.Kc2 in order to reach black’s “e” pawn (he was one point ahead from Botvinnik up to this round, and a win would have enlarged the gap, with good prospects for the world title). Of course, he didn’t see the geometrical walk 57…Kg3! that reached the f2 spot in the same time that across f3 (if you pay attention, this is the “Reti maneuver” in a minimized scale).>

click for larger view

It's not just Botvinnik. Of 63 games in our database with this variation, 8...gxf6 was played 61 times.

The basic reason is tactical. After 9.Ndb5!, Black is almost surely going to lose material. White's threats of 10.Nxd6+ and 10.Nc7+ can only be delayed by the time-wasting 9...Qd8, but then 10.0-0-0 piles on.

Black might be able to hang on for now with 9...Kd7 10.0-0-0 e5, but who would want to play that?

It's a good question, though, since 8...gxf6 gives Black doubled pawns and makes kingside castling hazardous. But there are compensations. Black's advantages in the Sicilian include a pawn superiority in the center, and 8...gxf6 strengthens this control. Also, with so much pawn shelfter in the center, Black can choose to go without castling completely. And every doubled pawn means a half-open file.

However, the whole line with 7...h6 seems to be out of favor now. Some of the earlier kibitzes to this game reveal that Botvinnik had been preparing the variation before the match, but eventually abandoned it.

Bronsteins quote after 34..Rh7:
"We have here a typical position of dynamic equilibrium. Two other players would without doubt have gladly agreed to a draw. But in the 1951 match, on the initiative of Botvinnik and in my formulation, the rule of offering a draw only once was employed for the first time.And so neither of us wanted to suffer a moral defeat..."

The comment 58.Ne6 draw is completely wrong and also pointed out by Bronstein as previously noted.

This position as mentioned by Bronstein is winning for black:

click for larger view