A better system for determining policy on a case-by-case basis

The close-vote system works well as a tool for implementing policy, but on many sites it has also been co-opted as a democratic system for determining policy. This happens either as a matter of local site policy or as the de facto result of having relatively hands-off moderators or a general lack of an active meta community. In either case, the result is that when there is uncertainty about whether a question is on-topic or not the community responds by waiting to see if it gets closed or not, possibly going through several cycles of closing and re-opening the question by different groups of people.

The problem is that this amounts to closing questions if

(votes in favour - votes against) % 10 >= 5.

There are a number of problems that stem from this.

• In principle, for a very large community where many people participate in the review queues, this amounts to a rather inefficient random number generator.
• In practice, for most communities, most questions don't get through to a single reopen cycle, so this amounts to a very strong "if in doubt, close it" policy.
• If the community has 5 or more active users who dislike a particular sub-topic it becomes impossible to ask questions about that topic, regardless of what the rest of the community thinks or of the question's quality. This is a real problem for the larger science communities, in my opinion.
• As a result of the above points, site policies can become much narrower than they need to be, eventually catering only to a subset of the community they were set up to serve.
• Since policies determined this way are not written down anywhere it's impossible for a new user to know whether their question is on topic or not.

I should stress that these points are not about wanting to leave open questions that are low quality or clearly off topic. They are about those few rare gems from new users that are well constructed and would be of value, but whose style or topic is unusual for that particular community, so that there isn't a clearly defined policy for that kind of question.

The root of these problems is that the close vote system has taken on two roles:

(i) quickly closing questions about which there is no real question; and

(ii) deciding whether to close questions about which there is a question.

It works well for the former but very sub-optimally for the latter. If these tasks could be more appropriately divided, with a new system taking over task (ii), it might be possible to solve these issues.

Highlight the answerer's post in the comments

This has been asked a few times in the past (most recent request here), but since that was asked in 2014, I feel it's been sufficiently long enough to bring it up again.

People use comments to question or correct an answer, and a person will often comment on their own answer to offer a counterargument. Knowing that someone wrote the answer they commented on is an important bit of context in long comment chains, and I am very surprised to find that it hasn't been implemented yet.

Raise an automatic flag when a question posted by a user banned at Stack Overflow enters the close queue at another site, so that moderators at that site could quicker close and delete off-topic coding / debugging questions.

Stop wasting community time and efforts on the problems that were already discovered and handled at Stack Overflow.

A bit of the history for those interested in details...

Abusing a system by trying to circumvent a block by posting the question to some other site is a long known popular trick. Five years ago, the Stack Exchange team even had to add a special system feature to address this, so that such questions can't migrate to the site where user is blocked.

The approach picked back then was not quite fair: although it protected Stack Overflow from getting back questions from already banned users, this was done at the expense of other sites where users had to waste efforts reviewing and closing inappropriate questions only to find out that migration is blocked.

Back then it hardly was a problem because not many users did that. I myself was a regular at a site that had been getting some of such questions and I remember how these were rather infrequent so it didn't really matter that we spent some effort on handling these.

Things changed after quality project features started rolling out at Stack Overflow. More users started bumping into the blocks, and it became harder for blocked users to retry attempts to ask over there with new accounts, etc. As a result, more of them started trying their luck at other sites and wasting effort of these sites regulars.

Related:

• Block migration if user is suspended/question-blocked at destination
• Stats on how attempts to circumvent Stack Overflow question blocks impact other sites
  ^{Stats over there are probably obsolete because these ware taken before quality project took off at Stack Overflow. More recent (and more scary) stats for one of the listed sites can be found here.}

• Perform automatic checks for... question block at SO when post enters close queue, not only at attempt to migrate

• Let's help askers who are trying to circumvent question block at Stack Overflow
  ^{Complementary / alternative approach that focuses only on improving new users experience by helping them avoid frustration if they get down and close votes, angry comments, blocked migrations and question block at new site when trying to work around SO ban.}

Make reviewing edits with Mathjax more managable

Currently, when viewing an edit where Mathjax is edited, the rendered output side-by-side diff breaks the Mathjax rendering. This is particularly annoying when reviewing edits, as the actual result after editing is unavailable.

Of course, it would be best to change the diff view to not break edited Mathjax, but a perhaps easier improvement would be to simply show the rendered output without the diff, as then the Mathjax will be rendered correctly (if the edit placed the Mathjax delimiters correctly. In fact, this is one of the hardest things to check right now!)

This is a pretty old issue and my suggestions are nothing new:

• MathJax rendering for suggested edit queue

• Mathjax half-broken in suggested edit diffs

But this might be a simple improvement that would make (community) moderating easier.

This is going to be a very elitist* suggestion, but how about an introductory non-zappable video before asking first question:

"How to get the best out of Stackwhatever"

With concise, positive advice, which I'm not going to rehash here, I think it's been done any number of times already here on meta.

_{*Because it's rude to point out that the problem is that people don't read any more.}

Ability to retract reopen/delete votes

Close votes can be retracted; let's make it happen for reopen/delete votes as well.

This may resurrect an old feature request.

Allow the user to mark an answer, but sort answers by the number of upvotes

While not as common as it used to be, I'll occasionally find an old question with a specified answer with a handful of upvotes. The answer after that will have Several dozen to several hundred upvotes, the next one a couple hundred, etc etc. Sometimes the accepted answer should be sorted four or five answers down. Stack Exchange is about community consensus on what is the best approach, but our most valuable resource (the first answer to a question) is picked by one individual.

Yes, the individual should have the ability to select an answer that works for them, and even the only person who should change that answer is the OP, but that doesn't mean it's actually the right answer to the question. Why does it make sense to let someone who has asked a question, showing their knowledge is lacking in the subject, control which answer is the first shown to incoming users?

More statistics on each question for geeks (like me).

This is the general idea, from that idea to make it real there are many questions.

• What level and up a user can see that statistics?
• What statistics to show?

For example, something like

• Views per month/day
• Votes per day
• Votes that gets compare with the views
• The average score of viewer
• The non-register viewer’s vs the register.
• The users that see, what tags they are best on…

All that on the bottom of the question hidden somewhere in a tab.

What is the goal behind that?

Apparently on most question that statistics will mean nothing but we can see if some technology is die or continue to interesting the people on some specific questions by looking if they still view and vote about.

For sure is only very interesting for some question and not for all !

Also can be a "gift" for high score users and motivate to get points to see that stats !

I've seen a lot of spam posts lately and earlier this week I just thought a Flag feature for users might come in handy on the user's profile.

I just remembered another feature-request of mine that comes with absolutely minimal (I think) coding requirements for maximum benefit — very small buck but significant bang.

I keep cancelling my edit rejects!

The gist is: if you want to do something that has a custom text box, e.g:

• send a custom mod flag
• close a question for a custom reason
• post an answer
• …

The required keyboard shortcut after having written in that custom text box is tab → enter. This even works for posting questions: if I forgot to add tags I’ll be right in the tag bar, if I did add them it also automatically posts.

And then there is reject suggested edits, where the workflow is tab →tab→ enter. And I swear it catches me every time.

A button to quickly/easily select all the code in a code block.

Especially useful when looking at questions to answer, whereby you want to copy the asker's code paste it into your own environment and try it out. It would save time with those code blocks that are so long that they need to show a vertical scrollbar.

I know this can be solved with user scripts etc. but still, it would be nice to have it built in to SE.

Example feature request on meta: "Select all" for code blocks

Identify and Provide Guidelines for Responders with Communication Issues

I really like the counselling that's automatically provided when I ask a question. Some responders, their information is sound and they mean well but they come off as needlessly rude because of their choice of language. Clearly they mean well but they could use some counselling on their communication skills.

Could warnings be incorporated into the answer posting process that identify highly emotionally charged words in a responders answer and provide suggested alternatives in the same way that the question posting provides advice when a question appears to be opinion-based?

New users usually start on one site, then find other sites they find interesting through the Hot Network Questions (HNQ). One problem that I have already indicated half a year ago and shouldn't exactly require be hard to fix, is the inconsistent link CSS these HNQ have across sites.

It feels like there are a dozen of different templates for this CSS, and what makes it even more confusing is that they can't agree on what form the contrast difference between :visited, :hover and plain should take. I've screenshotted 5 different sites, and each of those sites has a different take on HNQ links. One of them says :visited should be darker, one of them says it should be lighter, one of them says it should be a little bit darker (but not as much as the first site, because that would make too much sense), one of them says it should be cyan instead of the default red,... Both for new users and users that already have an expansive SE presence, this works confusing.

HNQ is one of the few page elements that can be found on every site, so users have a certain expectation of continuity when going from one site to the next through this method. Many existing users browse this network during their lunch break, at home, or while legitimately waiting for certain long-winded processes to finish. They may not be interested in a specific site and just want to read what interesting questions have been asked on sites they like to read. Many new users may find new communities they are interested in. For both use cases, consistent link CSS helps the user decide which question they want to visit next. It also can lead new users to sites they may enjoy.

I feel like we should at least consolidate the contrast across the different sites, even if we decide to not unify the CSS for the HNQ entirely. It makes no sense that going from Stack Overflow to Super User has a completely mirrored contrast for :visited compared to plain links.

Get rid of historical locking

This is probably the most counter-quality feature Stack Exchange has come with. Like Gilles said:

What we're doing here is to frame the garbage.

The historical lock has been used for stuff that isn't "history" at all. Stuff that was deleted by the community and was kept deleted for a long time, sometimes over a year, are undeleted and locked for the dismay of the curators and quality-aware users.

Instead, promote more aggressive course of actions like editing, migrating, re-asking, re-scoping, etc. that are more desirable since nobody will have claims that the content should be deleted, and frankly, content that is available on the internet isn't lost even if SE deletes it.

A queue just for new questions from new users, like Help & Improvement but before the question is shown to the public.

Most new users post bad questions, without intending to. This would truly help the community out.

Make the new user help available in languages other than English

A lot of users whose first language is not English use SO to ask questions. When one of those new users is confronted with a raft of new-user information in English, it's a larger barrier to entry than if it is available in their native language. Whatever we create to assist new users should be available in the languages that the surveys have identified as the most common in our community.

Hide deleted posts by default.

This has two implications, that I can see: First, if hichris123's suggestion is implemented, it will prevent spam from even being effective against most users. Second, it is annoying to read a bunch of really good content, then to be slapped in the face by trash, or perhaps have to scroll past it to get to the bottom of the page to the answer box after reviewing a section of the question.

Put the answer box immediately beneath the question.

It would make it a lot easier to recheck something in the question while writing an answer--especially if there are a few answers or a really long answer already.

We need a mechanism for asking the questioner a complicated question, one that supports full Markdown and can be unbounded in length.

I saw multiple proposals in here having to do with discouraging answers that are "more properly comments". This is about one scenario I've found myself in, repeatedly, where it was impossible to ask the question I needed to ask in the form of a comment. A nice, clear example is Bin data read differently in AIX vs Linux; look particularly at the revision history of my answer, https://stackoverflow.com/posts/38576285/revisions, which started out as a request for the OP to run a test program. Test programs can't go in the comment box, unless they're shell one-liners, and even then it can get awkward.

I would like to request that when marking a duplicate, if your answer is better than the duplicate, leave yours and punish the lesser quality one instead. I am asking that duplicates be marked based on quality not timestamp as that should improve the overall quality of SE as an information source.