I don't see why you assume a well thought out setting means the characters can have no impact on the world. Nor do you have to have collaborative world building to have autonomy even if that's your preference.
Speaking for myself, it's because it sounds--consistently--like your fun is in having the world be precisely the thing you want it to be and never anything else. That's what looks like a lack of interest in player autonomy. Since your fun comes first, then anyone else's after that. It's a matter of priorities, and you've made quite clear that where fun is concerned, yours comes first, because you are GM. If that's
not the case, then I don't understand why you have expressed your position as you have, because...I just can't see any other way of describing it than the straightforward, "If I don't get all of the things I find fun, I won't enjoy the game, so I won't run it, and thus my fun must always come first."
Hence why I said some of the things I said before. Or, if you prefer, a pithy phrasing of what I see my own position as: "
Sometimes, player fun comes first, even if it isn't what I find (most) fun.
Sometimes, my fun comes first, even if it isn't what the players find (most) fun." Give
and take. I don't see much (if any) give in your position--so it looks like the GM asking the players to never do anything but give.
A homebrew world can give characters more autonomy in my experience. Nobody's going to say you can't blow up my version of The Yawning Portal because it's such an iconic location that's required for future scenarios for example.
I'd say it's entirely side-grade, not upgrade, in terms of autonomy (agency, whatever one wishes to call it).
Because the huge benefit of a setting written by someone else? The GM's connection is necessarily secondhand. It's not the GM's darling, it's someone else's darling, the GM just happens to like it enough to be comfortable running it. If the GM doesn't care about potentially locking out future adventure paths (which, in my experience, that's rare--and the GM who
does care will usually tell their players and work something out), the detachment can mean
The best (most direct and efficient) way for homebrew settings to avoid the "GM's darling" problem (or, if you prefer, the "World-as-GMPC" problem) very much is collaborative development. Doesn't have to be top-to-bottom collaborative either. Just enough flexibility that the GM is willing to happily accept constructive outside ideas. (Non-constructive ideas would be, for example, rude, dull, adversarial, too difficult to manage/implement, or just generally tacky or tawdry.)
I will note, "best" doesn't mean "only". I completely believe that it is
possible for any GM to avoid this problem with effort. I just think it will be varying degrees of harder. Unfortunately, it's especially tough for worlds that exist over a very long period of time. The problem gets harder, not easier, as more and more facts become well-established. That's why, even when I do all the worldbuilding myself, I always consciously and intentionally leave "
ʜɪᴄ ꜱᴠɴᴛ ʟᴇᴏɴᴇꜱ" (or
"...ᴅʀᴀᴄᴏɴᴇꜱ", if you prefer, but only one globe has that specific phrasing) places in any world I work on/with. As long as I always leave room for doubt, for
my knowledge to be incomplete, I'm always at least somewhat like the players, genuinely
discovering the setting to some degree, not just revealing it.
