This game is full of deep calculations but thanks to the very instructive annotations, even a novice chess player such as me can follow what is going on.

<An Englishman> I had a similar reaction to yours. But here is what Vuckovic has to say after his seven pages of annotations on the Pillsbury-Tarrasch game:

<This famous game has taken its place in chess literature as a classic example of an attack on the castled king which is stronger than the counterattack on the queenside. But in light of my annotations, which differ fundamentally from all annotations hitherto, from Tarrasch to Reti, this game takes on a new appearance. Black's numerous opportunities to strengthen his game at various stages point rather to the equal balance between attack and indirect defense, and it was only the decisive mistake on the thirty-eighth move which swung the balance in White's favour. This view of the game also qualifies it as a good example of the importance of indirect defence against an attack on the castled king, and of the part played by operations with wing pawns in particular.>

Vukovic then makes an interesting assertion:

<Indirect defense by means of advancing a pawn majority on the queenside was at one time held in high esteem and was more often practiced than nowadays; the real reason for this is that we have acquired a better understanding of the centre and the technique of centralization, thanks to the hypermodern school. It is clear that the player who has a majority on the wing, when the material is equally balanced, will usually not have the greater influence of pressure in the center, nor whill he have the necessary conditions for a central action. A majority on the wing is created at the expense of the centre and means giving up lateral pressure on the centre, and that is the sort of strategy which the masters of today are reluctant to adopt.>

Perhaps you could say that you see "indirect defense" in Vukovic's sense as practiced by Pillsbury and Spassky today more in positions in which the center pawns are locked, as in the classical King's Indian, where white advances his queen-side pawns and defends his king.

But here is an example from a QGD in which a queen-side pawn advance fared better:

Pillsbury vs Schlechter, 1895

To start off with, 13.g4 is fun for sure, but after 13...Nxe5 14.dxe5 Ne4 black has equalised. Hence, 13...b5?! might be considered Spassky's first real error.

The real issue however is this: 23...Rxd5 is surely an error. After 23...Bd6! instead, white's getting nowhere fast. Following the best line: 24.Bd4 Rc7 25.Qf3 Nef5! (there are no playable alternatives) 26.h6 Nxd4 27.hxg7+ Kxg7 28.Rxd4 (28.Qh1? is met by ...Ne2+; 28.exd4?! Be7 and black's holding on, while white's pawn structure is a mess) Be5 and the position is pretty much even (Fritz would even argue that black holds a small plus).

From white's 18th move to black's 22nd (inclusive) the game is forced, so if the evaluation here is wrong, then white must try 23.Nxd7 which leads to 23...Qxd7 (23...Bxd7 24.Qd4 Rg8 25.a3 Ba5 26.Bf6 Qe8 and white has reasonable compensation) 24.Qxd7 Bxd7 25.Kg2 (cutting out ...Bh3), and again white looks a little happier, but there's no clearcut win.