Sequences of Unexplained Moves

Continuing with Smyslov's Sparklers, I'm going to take another stab at Why Did White Lose in Reshevsky - Smyslov 1945?.

The diagrammed position looks complicated. Reshevsky played 20.dxc5, and the game continued 20...Nxc5 21.Nd4 Rdc8 22.f3 Nb3 23.Nxb3. Both Smyslov and Kasparov thought 21.Nd4 was a good move and gave it a '!'. Smyslov noted, 'White concentrates his minor pieces on c6, leaving his g-Pawn en prise. The sacrifice is a sham.' Kasparov copied the remark and the subsequent analysis verbatim. The other moves in the sequence went unremarked.

Radio Match 1945
Smyslov, Vasily

Reshevsky, Samuel
(After 19...Nf6-e4)
[FEN "r2r2k1/1q2bppp/1n2p3/pBpbN3/Pp1Pn3/4PN2/1P2QPPP/R2RB1K1 w - - 0 20"]

Call me blind, but I don't see that 20.dxc5 is forced and I don't see why White can't play e.g. 20.Rac1. Now 20...Bb3 is a blunder because of 21.Bc6. Black can play 20...f6, forcing the Knight to 21.Nc4, from where it came a few moves ago. In the meantime, Black has weakened the Kingside slightly and there is nothing particularly wrong with returning the Knight to c4. Note that White could also play 20.b3 instead of the text. The Pawn is immune.

The habit of leaving long sequences of moves unexplained is frequently annoying and sometimes misleading. Many top GMs do this routinely when annotating. If the variation is forced, let's understand why. If it's not forced, why is it preferable to the alternatives?