There’s multiple threads on this if you want more info, but basically in most cases the perception is
a) longer story length = good. So if you buy a book that’s 200k words long, and it’s relatively railroaded, you might see at least 100-150k words on a single play through so it feels like a longer game you got your money’s worth out of. If it highly branches, you might only see 30-40k for example.
b) Many people only read games once or twice. So they feel cheated if they pay for a 200k story and were expecting close to that in story word count, but only get 30k on a playthrough unless you’re the kind of reader that wants to go back and explore all the other choices to get the full possibility of stories that could happen depending on the choices made. (Which from polls seems to be uncommon for most games people are picking up.)
Both these things lead to bad reviews, which leads to poor sales, which leads to most authors not writing highly branched games that split into many different outcomes instead of following a branch and return type formula which is most common.
Choices are part of the attraction, but can be used cosmetically, for significant direct changes, additive changes or delayed ones. These can be used in any style of game. It’s more whether the game is more of an interactive story (1 main story with reader imput) or more of an interactive game (multiple stories that branch out.) Neither is better or worse than the other, it just comes down to reader preference and COG’s audience seems to prefer the former overall if you look at app store feedback. It’s not the only thing that counts certainly, but it is an important factor IMO.