On the other hand, there are pre-tex (yet also other things) were some editing could increase considerably the readability. And, now if somebody has the idea to edit some old things as default I would assume they will focus on things were they see a lot of room for improvement, as opopsed to making a massive number of minimal or borderline pointless edits. Noone wil go through each and every (old) post. And, regarding "in some people's opinion, plain text and Unicode not only carry the same information, but also display it in just as clear and legible a way, as TeX" Fine. However, it could be interesting to test this opinion on reality against reality. I do not buy that (on avarage) it is just as easy (and pleasant) to read these things. And, the opinion that reading things like old volumes of Springer's Lecture Notes in Math. is not nice due to the typesetting (or rather the lack thereof) seems pretty wide spread. And, this is not only a question of aesthetics. (For centuries a lot of effort went into good typesetting even when it was quite difficult to do. Certainly not all these people were fools for thinking this was relevant.)

And, after Tom Church voiced his opinion I looked at some of his old answers. And, yes, on the one hand they look really good given the circumstances yet on the other hand will one actually claim that this answer would not be better if it used mathjax [also look at the other posts since, as said, he really made a lot of effort; not that the other are particularly bad].

Finally, of course one can also come to different conclusions how relevant such improvements are. And, the sorry state of many of my own contributions regarding such side-aspects documents that I do not consider it as a big priority to optimmize for this on MO. What however I do want to maintain is that it would/could be an improvement if such things were done. And also that in my opinion and observation too few people have a global approach to MO in that they would also consider improving (at least in their mind) others' contributions or do other things "for the MO community," so that it seems particularly unfortunate to me if an idea to do so is received with raising so many red-flags.

More seriously, the point seems to be that, at least in some people's opinion, plain text and Unicode not only carry the same information, but also display it in just as clear and legible a way, as TeX. I accept that others might disagree, but at least I want to argue against the presumption that TeX is always better.

On the other hand, as long as there is not a constant stream of questions being bumped to the front page by minor typographical changes, I don't feel that strongly about it.

Now, I'm not saying anyone actually edits old answers with this motivation, but if a perfectly understandable answer gets TeXified with no explanation, this would be the impression I was left with.

@bsteinberg and @Artie: Good point. Thank you for bringing that up.

As a side remark, I would certainly appreciate the possibility of non-bumping edits.

I think one needs to exercise good judgement in making edits of this kind on a significant scale, and restrict to editing answers for which there is a very clear benefit from adding TeX.

More generally, are there similar ways to improve the value of an old answer written by someone else, and which are considered acceptable edits?