One Bad Move Loses the Advantage, Two Bad Moves the Game

Once again on Smyslov's Sparklers, I'm starting to understand Why Did White Lose in Reshevsky - Smyslov 1945?. In the diagrammed position, Reshevsky played 28.Nd3. Kasparov gave the move a '?' and wrote:-

I don't understand why White didn't play 28.e4! (perhaps because of time trouble?). The opportunity is also not mentioned by the commentators, although 28...Nc3?! 29.bxc3 Bxc3 30.Rf1 Bxd4 (30...b2? 31.Rd7) 31.Bxd4 favors White. And if the Knight moves to another square, he also has no reason to complain -- on the contrary, he may be able to seize the initiative. Apparently Sammy thought that e3-e4 would not run away, and he decided to play more quietly, underestimating his opponent's clever reply.

I analyzed 28.e4 and agree completely with Kasparov. White keeps a small advantage with the move. Was 28.Nd3 the losing move?

Radio Match 1945
Smyslov, Vasily

Reshevsky, Samuel
(After 27...Nb6-d5)
[FEN "r1r3k1/1q3ppp/4p3/pB1nN3/Pb1R4/1p2PP2/1Pb1QBPP/R5K1 w - - 0 28"]

The game continued 28...e5 (a very classy move) 29.Nxe5 (29.Rc4 also looks interesting.) 29...Bc3 30.Nc4. Here Smyslov and Kasparov give the same line: 30.bxc3? Nxc3 31.Qd2 Nxb5 32.axb5 b2. The variation 30.Bc6 Rxc6 31.Rxd5 Bxb2 32.Rb5 Rb6 also looks bad for White.

There appears to be a little problem with the analysis of the two World Champions. After 30.bxc3 Nxc3, White has 31.Qf1, which is better than 31.Qd2. Now 31...b2 doesn't work as well because of 32.Re1 b1=Q 33.Rxb1 Bxb1 (33...Nxb1 34.Bc4) and then 34.Rd7 Rc7 35.Bc6. Here I looked at a lot of variations and White always comes out better, although Black can force a draw.

What am I missing? It appears Reshevsky made one bad move to throw away the advantage and a second bad move to lose. This is a pattern seen in many games at the highest level.