Is it acceptable to edit other people's answers to include new information?

Question

Recently I have noticed a few suggested edits that included adding new information to an existing answer. Is that acceptable?

My intuition is that edits should not add new information that the author did not include -- because it may compromise the responsibility and trustworthiness. E.g. if a nonsensical edit is suggested and somehow passes the review, it will look as if the original answerer had this bad idea, while it actually belongs to the editor of the answer. I think that in such situations, a new answer should be posted instead, or perhaps a comment under the existing answer prompting the original answerer to update his answer accordingly (so that the original answerer can decide on it himself/herself).

Answer 1

I'd say No in broad terms, but as usual much depends on the details.

Positively, substantial new information could help a thread considerably, but that should typically be added through comments or very possibly as part of an answer.

But what's substantial? For example, if someone edited a literature reference to add detail or to add a URL, that's an excellent contribution in my view. If the editor wanted to do that without claiming credit, that's fine too.

Negatively, the purpose of editing (see e.g. Editing questions: dos and don'ts) is essentially to clarify a question, including correcting trivial errors, but not to amplify it.

The possibility of a misguided or incorrect edit being accepted by mistake exists always, independently of this issue, and is to be matched by a rollback and/or further editing.

Answer 2

I agree with @NickCox that as a general policy answer edits should only be "minor" (copy editing), i.e. improve grammar, fix spelling and LaTeX mistakes, format the quotes and code blocks, insert full references and hyperlinks to papers, etc.

Substantial (major) changes should not be made via edits; if you think that an answer is incorrect or not entirely correct, consider downvoting, and/or commenting, and/or posting an alternative answer.

However, I have encountered some situations where I think it is justified (with extreme care!) to violate this rule. For example, if a highly upvoted / accepted answer in a popular thread contains some sloppiness or misinformation and if the answerer is not with us anymore or simply does not react to the comments, it might be worth to edit. It is practically impossible for a new answer in such a thread to get more upvotes than an existing popular answer; and there might be so many comments already present that a new comment would not be very noticeable.

I did it a couple of times, e.g. with the accepted answer here: What are principal component scores?

Very rarely I have even decided to edit to improve an answer that was correct but I felt could be extended or clarified. E.g. the accepted answer here: PCA on correlation or covariance?

That said, I re-iterate: I do agree that our policy should remain that such edits are generally not allowed.