Editing policy is contradictory and unclear

Question

The current guidelines for editing are very scarce. I would especially like to know the actual consensus on the philosophy of editing the answers of others.

Current policy

1. We welcome all constructive edits, but please make them substantial. Avoid trivial, tiny one-letter edits unless absolutely necessary.

2. fix grammatical or spelling errors.

3. clarify meaning without changing it.

4. correct minor mistakes.

5. add related resources or links.

6. always respect the original author.

7. this site is collaboratively edited, like Wikipedia.

Issues

• Point 1 is incompatible with 2 and 4. The vast majority of questions and answers that require corrections for clarity or spelling are going to range from trivial to small and affect a sentence or two, without risking conflict with point 6.

• Point 1 is fundamentally incompatable with 3, 6, and very likely 5. Real "substantial" edits to improve the quality of a question or answer are almost guaranteed to modify meaning, especially if they involve correcting factual or coding errors. Additionally, important coding corrections are likely to appear "unsubstantial".

• Point 7 (from the FAQ), from what I gather is a very bad analogy. Per the previous bullet, the types of edits that "boldly" modify content are discouraged, sometimes even if unambiguous and indisputable (e.g. coding errors again). This makes stack overflow very unlike any other wiki in which the goal is to collaborate in order to make a page or article as good as it can be.

• Point 5 seems rarely exercised in practice. Doing so especially with accompanying text, again, often entails modifying meaning. Adding links or resources is usually done in comments.

Personally, I don't see the point in whitelist-only heavily filtered editing. Why can't it work like most wikis where if an original author disagrees with an edit, they can simply revert or improve it themselves? Things naturally work themselves out except in the case of comparatively rare edit wars requiring intervention.

Public forums like S.O. where editing is impeded run the risk of succumbing to what I call The W3Schools problem. That is, the most popular resources with high search rankings and containing information that appears plausibly correct to the non-expert can be, and often is dangerously flawed, and can spread misinformation to the masses for years if nobody is allowed to edit them. Respecting the original author means leaving critical mistakes intact in answers clearly written and upvoted without adequate research. The positive impact I can make by downvoting or leaving a comment is limited especially if the author disagrees, and the current reputation system gives no incentive to this important task. The most constructive types of edits, i.e. those that improve the accuracy or completeness of an answer, are sometimes unwelcome.

But that's just my initial observation. The main point of this question is that a philosophy should be decided upon, made clear in Policy, and enforced consistently. The FAQ in particular says almost nothing about editing philosophy other than that questions and answers may be edited, leaving new users to guess.

Answer 1

I believe you're reading point n°1 too literally, the substantial part of it doesn't mean edits should change 50% of the post: it means you shouldn't fix a single-letter typo when it doesn't introduce ambiguity and there are other things in the post to fix. Point 2, 3 and 4 don't conflict at all.

Point 6 is simply a matter of politeness, and point 7 is a fact - nothing to argue about that, anyone can click the "edit" button.

The idea and guideline is as I see it by combining points 1 to 6:

• If you change a post from meaning "A" to meaning "Z", the edit will (usually^*) be rejected, regardless of whether Z is indeed better than A or not. Don't edit for that, post an new answer if you can and want to, comment if that would help, and use your votes.
• But as long as the meaning of the post is still "A" after your edit, and the overall quality is improved (if possible, without minor fixups left to do), then your edit is welcome and will (usually^*) be approved.

---

To elaborate:

That first point is not incompatible with points 2, 3, 4 (except possibly for code) at all. Fixing spelling mistakes, grammar errors, or rephrasing to improve "sub-optimal" English are more than welcome as long as you don't change the meaning/intent of the original author.
Edits that only correct a pair of typos and rephrases a sentence are welcome, anything to improve readability (and searchability) is good. Fixing a single typo in a post that is not formatted properly and has numerous other things to fix is not good.

Also take into account that the substantial part is especially valid for people that don't have full edit privileges (2k rep). Edit reviews take time from the community, so edit that aren't substantial, in the sense that they really add value to the post, sap resources from the voluntary community reviewers.
Once you gain the full edit privilege, use your common sense and remember that edits bump questions on the front page (this is especially a problem for low-traffic sites).

Point n°4 is a bit more contentious if you're talking about code (on Stack Overflow). Fixing minor errors in code is welcome too, but be careful there: fixing code in questions is a bad idea - any code error there could be part of the problem.
Minor in this context means something like an "obvious" typo (in a variable name for instance) - something that a compiler or proper IDE would have spotted right away. Changing an algorithm, changing what API/data type/standard library function a piece of code uses in an answer is generally a bad idea.
If you see an answer that is no longer correct because it uses now-deprecated language features for example, you have I believe three options to help maintain site quality:

• Edit the answer to add an updated version - if the change is minor (1/2 lines, not the whole algorithm/method/approach) If you do this, keep the original around, and do say when the change became necessary, or recommended. Adding links to official docs about the change is a good idea here.
• Or post an answer of your own with the current best practice/favored method of answering the question - that's ok too, even if it does build from an existing answer. If the change required is substantial enough (use your common sense and domain knowledge to define that), having it separate from the original is a good thing - allows independent voting.
• Add a comment saying "XYZ should now be used instead of ZYW when dealing with version FooBar of ..." if you can't/don't want to post a complete answer and the change required to the existing answers is too extensive.

The idea is that if you have a substantial contribution to make, that isn't covered by existing posts, add an answer so that it can be reviewed/triaged on its own merits. If you have a minor tweek to existing content, edit or comment.

Now, regarding point n°5, it is indeed rarely used, except for Tag Wikis. It should be used, IMO, only to add authoritative references (links to official documentation/samples, standards, whatever). (Adding links to an add-laced blog that only partially answers the question is not welcome.)

---

Addressing the last two paragraphs of your question: I disagree. I rarely encounter wrong content when I see Stack Overflow (or other SE sites) in my search results. And guess what: when that happens, I have tools to deal with it: editing, voting, commenting, even flagging.
Compared to results gotten from your average forum or mailing list, SE results are, in my experience, much more relevant and helpful.

The voting system is there to do the filtering. Incorrect content will eventually be weeded out, and good content rises to the top. Sure it doesn't always work as well as everyone would hope for, but, by and large, it's pretty good.
(And the flag/close/delete workflow takes care of the more egregious content.)

(Comparing SO to w3schools is apples to oranges. Stack Overflow isn't a tutorial, or a documentation repository, or whatever w3schools aims to be. It's a place to get your code questions answered.)

---

^* As for useful edits being rejected (or valid edits getting approved), yes, that happens. I probably rejected some - sorry if I did - but that's to be expected given the sheer volume of activity, the points above, and the fact that people are not perfect and mistakes happen.

Answer 2

I do agree that the guidelines need a bit of a facelift, but I disagree that they're in total conflict with each other. There are some conflicting issues, but overall it's reasonably straightforward. My feeling is that, if you're editing out coding problems, then you're not really editing the question.

---

• Point 1 implies that you're not going to make a non-substantial change to the wording or style of the question. Things like moving around curly braces would be seen as non-substantial.

  "Unless absolutely necessary" is an implication that tiny changes should be made only when they truly improve the quality of the question or answer, and if you're < 2K rep, others may or may not agree with that change.

• I soundly reject the second issue, and I offer this question on Ask Ubuntu as proof. I'd argue that one can damage the original intent of the question by correcting coding errors (i.e. adding a type to a variable, properly indenting Python code, etc), since that may very well be the original asker's code.

  Assuming good faith when the person asks a question with code in it, even if it contains syntax errors, is something that shouldn't be edited out - if there are syntax errors, tease them out in comments, and ask them to revise those. Or, if that's the actual answer to the question, answer it instead. (If it's something obscure or something you feel can't help others, it'd be fine to [flag to] close as "Too Localized".

If you could provide an example of an edit that was rejected that "improved the quality of the answer", then that would be one thing. I haven't come across any recently, so I can't really say. However, I will say that edits that improve a question or answer are rarely, if ever, rejected.

Edits that change coding conventions are something I'd reject since that has nothing to do with the question, or it could be the actual problem.

Answer 3

There are two large issues you mention: too small edits and too large edits.

The guideline to avoid minor edits, which is also hard-coded into the suggested edits feature, is purely a response to the problems too many small edits can cause. It is not a fundamental principle of SE as I understand it, but the necessary consequence of the SE mechanics. Edits must be reviewed, or the site is vulnerable to vandalism that is hard to detect. The mechanism by which this is done currently is bumping any edited post, no matter how minor, to the frontpage. These bumps displace newer posts that need the attention, so any edit harms the frontpage, but substantial edits offset this harm by the improvement they bring to that post.

Now, most post that have one spelling or grammar mistake, have a few others that could be corrected at the same time. There are cases where there is just this one, annoying mistake in a post, but they are pretty rare in my observation. So I don't think discouraging one-letter edits has an overall negative effect, the positive effect of encouraging users to fix all the problems with a post outweigh the negatives in my opinion.

On the matter of too large or substantial edits, I'd emphasize that edits are not the correct way to deal with completely wrong answers! They are the ideal way to correct small mistakes in an otherwise good answer, but completely rewriting an answer is not the idea behind the edit system.

You're mentioning wrong information on SE sites, if you see a wrong answer you should comment on it and explain why and how exactly it is wrong, preferably cite an authorative source that shows that you're correct. A well-placed comment can often guide voters and prevent incorrect answers that appear plausible to rise to the top.

Another very important mechanism to correct wrong answers is to post your own, correct answer. Combined with a comment pointing out the error and downvoting the wrong post this is the preferred way to deal with just plain wrong answers.

The comparison to Wikipedia is a bit misleading, in most cases SE doesn't work like a wiki where everybody edits the same post. That sometimes works, especially when the initial poster of an answer recognizes that others need to extend the first revision and declares it community wiki, inviting others to chip in. But it is not the common case, but more an exception.

The one big problem in my opinion with substantially editing answers is that it is incompatible with the voting system. It doesn't make any sense to edit a wrong answer that has been downvoted, the voting will still reflect the old, incorrect state of the answer.

Answer 4

You misquoted the policy. Points 2 to 5 are each an example of a generally-good edit. An edit isn't supposed to match all of them, it's supposed to match one of them (or none: this is not meant to be an exhaustive list).

Point 1 isn't contradictory with the others. It says to “avoid … tiny … edits unless absolutely necessary”. Fixing a dozen spelling errors isn't that tiny.

I don't know where you get that point 5 isn't exercised.

The one difficult thing about edits is balancing points 6 and 7: respecting the original author while striving to make the content as good as possible. As a rule of thumb when editing, ask yourself: would the author have written this if he'd known better? If the answer is yes, your edit is probably good. If the answer is no, your edit is probably bad.

Answer 5

Complexities in editing policy reflect the fact that simple rules are not sufficient.

For example, below edit looks reasonably OK doesn't it? (save for the questionable can->may and minor glitch of not changing i to I):

But this is only until one takes a closer look into the edited post - which immediately reveals that suggested edit is not substantial. Not even close!

The problem with suggested edit is that its author did not make an effort to look into post text that badly needed cleanup. I would say that substantial edit is one that takes into account all issues in the post. Note this does not imply that it's required to edit everything - this is at editor discretion, I can imagine cases when I'd skip some editing (say, when title and tags are severely broken and need quick fix while large post text requires more thorough check).