I have just rolled back an edit for (as far as I can recall) the first time on this site. This edit was approved (and improved) by Jan Doggen, a user I respect, so I thought it might be a good idea to canvass some opinions on what to do with edits that rephrase questions where it's not strictly necessary. I'm not asking here about edits that make major, substantial changes to a question’s content, but about minor edits that just re-word things that didn't seem to need re-wording.
Here's the question that precipitated this meta-question. The revisions that caused me to revert were the following.
Which geological processes [will -> could possiby] destroy plastic?
sedimentary rocks will contain [plastic pellets at the least -> small plastic pellets]
geological [processes that create -> process that creates] metamorphic rock
I hope my syntax for edits above is comprehensible: the edited bit is in square brackets, with the original to the left of the arrow and the edit to the right.
In (1), the processes that "could possibly" destroy plastic in the original question are clearly a huge superset of the processes that "will" (with certainty) do so in the edit. In (2), the original didn't specify the size of the pellets, and added "at the least" (i.e. other things may be present); the edit added a spurious size specifier for the pellets, and removed the mention of possible other components. In (3), an unspecified (but plural) number of processes is changed to a singular process.
In my opinion all these edits change the semantics of the question, without correcting any perceptible problem in the original phrasing. Personally I tend to be as conservative as possible in editing other people's questions: unless there's a blatant error where the intended meaning was obvious, I leave the original wording, even if I personally consider it inelegant. Partly this is because of the danger of changing the author's meaning; partly it's a wish to respect their original choice of words. We're a broad international community and I'm aware that my idea of elegant prose isn't universally shared.
In short: whenever I'm in doubt regarding an edit, I will tend to come down strongly in favour of the original author. I'm concerned that, even in cases where the phrasing is undeniably improved, the questioner may end up with answers to a question they never actually asked, because of some slight accidental shift in semantics. If I'm reviewing an otherwise useful edit that contains random minor tweaks like this, I'll probably revert them before approving. If I'm reviewing an edit that consists only of random minor tweaks, I'll probably reject it.
Do others feel this way? Are there strong arguments for this kind of "free-and-easy" editing that I'm ignoring?