I find GNS considerably less useful than the original GDS theory which Ron Edwards derived it from.
Basically, Edwards had a very specific preference for a very specific kind of Dramatist play. He labeled this Narrativist, but then he found himself with a whole bunch of other Dramatist playstyles that didn't fit into the N box of his new GNS. So he took all of those playstyles and kind of shoved them into Simulationist play, which he had never understood all that well in the first place.
Basically, GDS has a very specific and useful definition of its three axes. And it largely constrained itself to the description of the rationale used for any individual decision point. In terms of each decision point, one is forced to make a trade-off when it comes to the rationale used. Overall playstyles are described as an aggregate of decision points.
Edwards did put important emphasis on the impact of rule system on playstyle. Although his purity-driven style of system design results, by its very nature, in games of specifically narrow appeal. Edwards may not like the "incoherence" of classic D&D, for example, but it is specifically its broad mechanical flexibility that made the game appeal successfully to many different gamers.
Both theories tend to overvalue purity of theory at the expense of practical compromise at the gaming table. Although, IME, the GDS theory is an immensely useful terminology for discussing and reaching those compromises.
