Pawn and Two: <whiteshark & beatgiant> Interestingly, Kotov considered the position to be won for Black after 42...b4. In "Grandmaster at Work", Kotov stated, <...the presence of a strong passed pawn on b4 gives Black a decisive advantage.>
After 49...Ke4, Kotov stated, <Here is where the problem is solved. The pawn ending, despite the extra pawn, is lost for White.>
However, after 50.e3!, as noted by <whiteshark>, it appears White can hold the position.
Fritz provides the following analysis: (-.66) (24 ply) 50.e3! Kf3 51.f5 Kxf2 52.e4 Rb5 53.Kd3 Kf3 54.e5 Rxe5 55.Rxb2, (-1.02) (26 ply) 55...Rxf5 56.Kd2 h5 57.Ke1 Kg4 58.Ke2 h4, (-1.02) (30 ply) 59.Rb3 Re5+ 60.Kf2 f5 61.Ra3 Re4 62.h3+, or (-1.02) (30 ply) 59.Rb4+ Rf4 60.Rb3 f5 61.h3+, and Black appears unable to make any further progress.
After 50.e3! Kf3 51.f5, <beatgiant> suggested the move 51...h5. This move, however, leads only to equality: (-.08) (27 ply) 51...h5 52.Kc2, (-.02) (27 ply) 52...Kxf2 53.e4 Kf3 54.e5 Ke4 55.e6 f6 56.Rf1.
In this line, instead of 55...f6, Black could also try 55...fxe6, 55...Kxf5, 55...Rf8, 55...Rh8, 55...Rd8, 55...Re8, 55...Rc8+, 55...Ke5, 55...Ra8, 55...Rg8, 55...h4, or 55...Kf4, but all of these moves give only an equal position: (.00) (22 ply).