keypusher: Had Shredder look at this game at a setting of three minutes a move. The engine doesn't care for the challenger's plan of welding his bishop in place and letting the knight take it on f5. The few times this variation has come up Black has generally preferred 6....Bg6.
The position after 4.h4 came up five times in the match. Botvinnik played 4....h6 three times and 4....c5 once (and got into some trouble after 5.dxc5).
In the game 8
.c5? was premature. As <talisman> suggested long ago, White could get a clear advantage with 9.Bg5 Qa5+ 10.Qd2 Qxd2+ 11.Nxd2. Simply 9.dxc5 is also strong: 9
.Bxc5 10.cxd5 Qxd5 11.Qxd5 is obviously bad for Black) 10.Qd2 Qxc5 11.cxd5).
Instead of trying to grab a pawn, Tal prefers to give up one of his own: 9.cxd5?! Qxd5 10.Nc3 Qxd4 11.Qf3? (11.Qxd4 cxd4 12.Nb5 still keeps some advantage for White after either
Na6). By move 14 Shredder thinks Black is clearly better.
Botvinnik then gives the advantage back with the normal-looking 17
.Rad8 (the engine likes 17
.Bd2 18.Re2 Qd6 19.Qe4 Bxc3 20.bxc3 a6 21.Bd3 Rhg8). White is only slightly better after that, though, and the position gradually peters out into equality. Botvinnik's cold-blooded 26
.Rxb2! 27.Nf5+ Kf6 28.Nxh6 Rdd2 secures the draw, since 29.Rxa7 Rxf2+ is perpetual check.