Is it good practice to substantially edit older questions?

Question

This was brought up by a comment I received on an old answer of mine, regarding this question from 4 years ago.

Until yesterday, the question asked if/why a robot could even perform the function required to take a home pregnancy test. I remember this quite well, as it was a difficult question to answer within the guidelines of the site, and I was quite proud of my eventual ability to keep my answer clean and civil.

Yesterday, the question was edited twice. The edits eliminated a few spoiler boxes and substantially streamlined the text (which is good) but also changed the core question asked in the test from 'How could a robot take a home pregnancy test' to 'How could a robot get a positive result on a home pregnancy test'.

The editor of the question then provided their own answer, which completely ignored the original question of 'why would a robot be able to urinate' and instead answered the new question of 'how could a robot get a positive pregnancy test result'.

Obviously, making a question easier to read is a good thing, but what about the remaining actions - is it a good thing for people to edit a question in order to make an accepted answer a poor fit, so they can provide their own answer?

Answer 1

The short answer is no. In my haste to widen the question to encompass the issue of how a

robot

could provide a positive pregnancy result, I accidentally zapped a major part of what the OP had put into the body of the question; e.g.

How can a robot even pee on a stick?

In fairness, the original question header was "Why was Buffy's test result positive?", which is probably what confused me.

---

Moving forward, instead of asking a new question that would essentially be a duplicate of the existing question (or rather, a wider question that the original would now be a duplicate of), I've re-edited the question to widen and encompass both the mechanics of her taking the test and the result she provided. That makes your existing answer perfectly valid and allows the wider question (the one in the header) to be answered.

Answer 2

It seems to be more-or-less accepted that editing a question in such a way that invalidates existing answers is, if nothing else, massively impolite. See our meta, Meta.SU, and Meta.SE (and others, I'm sure). There hasn't been substantial (vocal) disagreement with this that I've seen, although of course I haven't seen all contexts, and am welcome to correction and argumentation.

Substantial edits, regardless of the age of the post, aren't necessarily bad; edits that fundamentally change the nature of the question are undesirable.

For what it's worth, though, I don't think the specific example you point to is necessarily a bad edit, although it's also not a great one (and I think your most recent edit is a great improvement). There was already a disconnect between the question and its title, but what most interests me is the last line of the second revision (emphasis not mine):

Given that this is that case, how was she able to provide a positive test sample?

This line, to me, prompts two (not unrelated) questions:

1. How was not-Buffy able to test positive for pregnancy. This is the question that you (and Tim, your commenter) have already noted
2. How was not-Buffy able to provide a pregnancy test sample in the first place; that is, how was she able to pee on the stick. This part is obviously less immediately clear, which is why I like your subsequent edit, although now we still have a disconnect between the question itself (which is asking about bodily functions) and the title (which is asking about the test result)

Personally, I probably would have posted a self-answered reference question, but that's a discretionary thing; given the question's title, it doesn't seem entirely unfair to assume that the OP was also curious about the outcome of the test.