It's part of sandbox worldbuilding, unless your play is entirely player-driven like yours, which neither I nor my players want. And it can be presented in a way that is setting-logical.
Okay. But that means you explicitly
do want some amount of narrative conceits engaged as part of play. This has genuinely gone beyond doing something with deep regret, metaphorically holding your nose and wishing there were a better way. This is, "I want that result, so I use the tool that achieves it."
Sure, but I feel the world and the game are richer experiences when activity is driven by both external events and internal PC drives. Not everyone (and certainly not me or my players) wants an entirely player-driven game.
I mean, in fairness, even my game isn't purely player-driven. The GM is the one who creates Fronts in Dungeon World, after all. The difference, at least as far as I can tell, is that you frequently position things as (effectively) being judged on a scale of verisimilitude, with higher verisimilitude presumed to be superior....unless one of your other goals happens to be on the chopping block as a result. Like having a game that isn't entirely player-driven, or a game that preserves pacing by having an unrealistically high proportion of engaging and worthwhile events, or a few other things.
Then a little bit of lost verisimilitude is an acceptable sacrifice, because the
real end goal is an engaging and worthwhile experience, not verisimilitude alone.
Which, as I said above, leads to a much more interesting discussion...but also one that necessarily is way more nuanced and full of grey areas. And, because it bears repeating and because I want to be totally clear, I am
not harping on this point as a means to say "well see you don't ACTUALLY care about versimilitude" nor "so every possible sacrifice of verisimilitude is actually totally okay and you're just being persnickety about it." I am harping on it because that's a conversation I would find far more stimulating and rewarding. Because I don't think anyone here thinks verisimilitude is worthless or harmful; I think most of us would, all else being equal, prefer a higher verisimilitude solution if that's the only meaningful difference between two approaches. Conversely, I also doubt that very many people--especially since it seems this applies to you, too!--would argue that
nothing is ever more important than increasing verisimilitude.
Hence, it becomes a question of cost vs benefit; we could view verisimilitude as a currency, where we are obliged to spend some (very,
very) minimal amount, but can elect to spend more if it would buy us something of greater value. That's an interesting thing to investigate! Much more interesting than debating what things count or don't count as "story" or as "game", and certainly more interesting than some of the other...well, to put it charitably,
non sequitur "discussions" that have cropped up in this thread.