In case you missed the first one, check out the tag. The quality project isn't one that we plan to ever finish, it's perennial and kicks in every 18 months or so after we've had ample opportunity to observe the efficacy of our previous efforts, changes in how people use our sites along with feedback regarding what they expected, and when new ideas seem promising enough to try.

It's time to talk about the areas that we're going to be focusing on. I'm going to give a high level overview of the types of things that we want to do, and touch briefly on things that we have in progress. I'll then turn it over to you.

It's then your turn to mention a that you think might be a great idea for us to evaluate for inclusion in this iteration, and we'll then take a look at all of it to see what we're going to be able to build.

We can't do all of it, but we're going to put the highest priority on small-ish things that could potentially make a big impact first, and then drill down into things that are going to take a bit of work, again with priority given to things that could help the largest number of users have a better experience on the site.

So, let's go over what we've got in mind:

Improving the new user experience

We want people to feel like they belong here. And in order to belong here you really need to be able to ask and answer questions competently, or contribute helpful things to documentation. It's that first part we're going to focus on first for the purposes of this project:

  • We plan to test a new, 'guided' version of the ask question page soon. This page would essentially break down all of the elements that make a great question, and give the user plenty of guidance as to why it's to their advantage to understand what's needed in each of them, and provide it.

  • We're launching an initiative with the University Of Melbourne in Australia. Some graduate studies have uncovered what (could) lead to much more precision in the detection of duplicate questions, as well as a huge improvement on what duplicates are suggested to the person asking. There will be a separate post about that this week. We need to help them validate a ton of data, they've come up with a system to make it easy for us to help. More to come on that soon.

Improving the experience of our existing users

We think the two things we've got planned in the new user experience category is going to have a pretty nice impact (most unhelpful questions don't need to be asked, if only we could let folks know we've already got their answers more effectively). But we want to do more.

  • We're going to explore some ideas for better filtering of the questions that you see indicated by information that you give us. If you really only enjoy solving medium to difficult problems, that's what we need to prioritize showing you. We're going to have some discussions on ways this can be accomplished when we come to it.

  • We're looking at some upgrades to the anti-spam and abuse mitigation systems. They've been good to us, but we need to keep them relevant. There are some awesome community-run projects that have been helping tremendously, and we're currently looking at ways that we might be able to integrate them more properly.

Improving the experience for our moderators

Because at the scale that we see on Stack Overflow, existing tools are really starting to show their age.

  • We want to end the need to manually suss out voting rings. We're working on making the scheduled tasks that identify these things much smarter, by teaching them to think like Shog9 and myself do as we analyze all of the data that we have access to. We'll find a lot more rings where action is needed, and stop showing moderators patterns that are actually pretty benign (but the tools they have really don't indicate that it's benign). Mods will get a sort of 'minority report' in cases where the system couldn't make an automatic decision explaining why it couldn't decide, show all of the information that the mod needs to make the call, and let the moderator decide. This is actively being worked on for feasibility, once we're sure the model scales we'll kick off a discussion with more detail.

  • We're going to do more to make certain that people understand what they're doing by flagging and help them choose the correct way to proceed (and in many cases that probably means not proceeding). We'll break out into a separate discussion when we get to this. Moderators are far too often called to look at things where they aren't actually needed, and this distracts them from looking at the stuff where they are actually needed. "Should I downvote, vote to close or flag? Should I do all of that?" - we need to make this much, much clearer. And we've got the inverse of that, folks downvoting spam when they really need to be flagging too :) It's not an easy problem.

None of these lists are exhaustive, there's more, but this is getting pretty long now. What's important is what you think we need to be prioritizing.

So what do you think we should include in this round? Try to link to the , but don't fret too much if you can't find it (just explain what it wanted, and we can rely on Shog9's RAM-like memory for the location).

  • 6
    What makes a FR a good fit for this project? I've looked at the other six questions with this tag and I'm not quite sure what sets them apart. – Catija Oct 17 '16 at 17:01
  • 46
    How much of this will be Stack Overflow-centric? For example, a guided Ask page will probably be great on ---Programmers--- Software Engineering, especially given the confusion in scope and recent name change (meant to fix the confusion in scope). – Thomas Owens Oct 17 '16 at 17:03
  • 12
    +1 for better duplicate handling and for improvements with flagging. – Stephen Ostermiller Oct 17 '16 at 17:53
  • 5
    From comment #3: or where existing users have a better experience of new users getting to know the site ---> The ability to close questions quickly that are wildly not ready for prime time perhaps by some user vote weighting method – Drew Oct 17 '16 at 18:17
  • 23
    "Some graduate studies have uncovered what (could) lead to much more precision in the detection of duplicate questions, as well as a huge improvement on what duplicates are suggested to the person asking." - Are these papers listed here? I like reading new papers/studies that utilize SE data. If they are not, but are publicly available, can they be added? – Andy Oct 17 '16 at 18:48
  • 6
    Did anything come of the post characteristics as images tests? – Andy Oct 17 '16 at 18:59
  • 4
    Can I just take a second to say that the changes you guys have made to the mod screens over the past few months have been very minor but very welcome? Maybe I just noticed it over the past few months and it actually was there all along but it seems a lot easier to mod lately. So umm, yay. – corsiKa Oct 17 '16 at 19:11
  • 12
    "If you really only enjoy solving medium to difficult problems, that's what we need to prioritize showing you." I am curious as to how you are gonna accomplish that. Grading the difficulty of a problem often requires to understand it first. If an AI can do that, I am scared. – Knu Oct 17 '16 at 19:53
  • 4
    @TimPost so you are planning to put the burden of grading on users? Now, in addition to downvoting and upvoting we will have (be able) to give a difficulty grade. Interesting. Gotta polish your UX for that one. – Knu Oct 17 '16 at 20:55
  • 8
    I notice that there is no statement of intent to implement things that are a lot of work. Do you think this is sustainable? Imho, there are a few things about the SE platform that need big changes. Do you plan to ever address such? – Raphael Oct 17 '16 at 22:48
  • 7
    Tim, do I understand it correctly that this discussion you opened is not in the scope this year: Empowering tag-badge holders part II - let's look at silver? (guess it's going to be kept ignored for quite a while since over two years have passed already and nothing happened) – gnat Oct 19 '16 at 7:31
  • 8
    When will we hear about the outcome of this initiative/? – Raphael Feb 5 '17 at 19:29
  • 21
    @Tim with all due respect (and there is respect, lots of it actually :)) the people who posted here wasted their time, and that's a fact. Proof? Let's take the top answer here, with 399 upvotes, posted almost year ago. You did post a promising comment, but those 3 weeks you mentioned turned into over 8 months now. So Robert has the full right to feel he wasted his time. I do not blame you. I do not blame anyone. Just stating the facts here. – Shadow The Vaccinated Wizard Jul 5 '17 at 6:16
  • 12
    And another full year went by. Do you by your comment "nobody wasted time", Tim? – Raphael Jul 15 '18 at 14:32
  • 5
    Any update on this @TimPost? And I actually mean any. – Script47 Jul 24 '19 at 12:37

81 Answers 81


Related : Could: "Possible duplicate of.." be given prominence for answering users?

Since you seem to be so much concerned about improvement on what duplicates are suggested to the person asking, it looks natural to additionally invest some effort into helping those who answer possible duplicates instead of aiding to properly handle these (by closing or by opposing the closure if duplicate suggestion is wrong).

This is especially so because it is so much simpler and it takes so much less engineering effort to help resolve matters after possible duplicate is discovered compared to searching for it, as proposed eg here:

showing a modal popup to the answerer if the question is voted / flagged as duplicate.

With the link (or better yet, full text) of possible duplicate and a message like: There is possible duplicate here, please make sure that you don't repeat already existing answers.

This would serve several goals:

  • For responsible answerers, such an explicit warning would help to make better informed decision, whether to abstain of answering or even support a duplicate with their own flag / vote, or do the opposite, that is answer and challenge duplicate suggestion if they believe it's wrong.
  • It will prevent complaints like "why am I downvoted, I simply didn't notice possible duplicate comment when answering".
  • The last but not the least, it will throttle FGITW shooters who will have to go through an additional screen before dumping their garbage into the question...

Implementing such a feature needn't be effort consuming, at least in its simplest form: warning message text, link(s) to duplicate(s) and two buttons for user to pick either to proceed or abstain of answering...

And it can be tested say, at MSO / MSE prior to making decision on whether to give it a go at main site.

Really, folks, you are going to go to such great lengths to educate askers about possible duplicates. But keeping the answerers oblivious about already discovered (!) duplicates kind of spoils all these efforts:

dupe close without answers teaches user: "next time, search and research". Close with answer teaches, "No need to research, just dump your question and it will be chewed up to you like it was to 10,000 guys like you before"...

  • 5
    Fix duplicate search first. Fix all the existing "canonical duplicates" with shitty canonical answers second. Only then worry about this. – zwol Oct 18 '16 at 14:09
  • 2
    @zwol you got to be kidding. Fix everything including most difficult and trickiest issues that may take years (review and rewrite all canonical answers, yeah sure) and only after that do the thing that takes just a day or two of developer time to implement. Just so you know, even with current insufficient system median time to discover and vote / flag duplicate at SO is mere 7 minutes, good enough to help against getting useless FGITW answers in almost half cases. Sounds like you are simply scared by proposed feature – gnat Oct 18 '16 at 14:22
  • 1
    For users with enough rep, the dialog should offer to vote-to-close. – Raphael Oct 18 '16 at 15:08
  • @gnat I don't see any of the things you seem to be worried about as significant problems, not compared to the problems I mentioned. Put another way, I don't see why the SE corp should bother spending even a day or two of developer time on the feature you want. – zwol Oct 18 '16 at 15:54
  • 1
    @zwol agree compared to mine your problems are more significant. But it doesn't make sense to put addressing these as a precondition to address mine, particularly because addressing your might take years while mine is a matter of days – gnat Oct 18 '16 at 16:01
  • 5
    @zwol the way you suggest will have the effect opposite to what you claim. There will be no improvement - instead there will be (heck there are) hundreds mediocre twin answers posted to hundreds useless duplicate questions. There is no room to improve when one can not see and compare how others already answered the similar duplicate. Now compare that to the way I mean, when duplicates are closed timely and answerers attention is refocused to dupe target... – gnat Oct 18 '16 at 21:38
  • 7
    ...answerers will try to post to that dupe target and they will have to try do better than prior ones. That will be real, direct competition, a clear road to improvement – gnat Oct 18 '16 at 21:38

The guidance for the "requires editing" button in the Triage queue currently states:

for questions where edits by the author or others would result in a question that is clear and answerable

However, if a question requires information from the asker, the question should be closed. "Requires editing" puts the question in the H&I queue, where editors are powerless to read the asker's mind.

My feature request is that you should strike "by the author" from this text. (There have been other proposals on Meta Stack Overflow about rewording these messages that may have better all-around text.)

Previous proposals/complaints:

In general, it would be nice to see triage and H&I evaluated before moving on to new things. I'm surprised it wasn't mentioned at all in this meta question.

  • 4
    I'd not only want "by the author" removed... I'd want emphasis on "by others". – S.L. Barth Oct 18 '16 at 9:05
  • 2
    This is just a symptom of a much, much larger problem: it's too hard for folks who actually know enough to know when a question is lacking information to close them as such. Expecting arbitrary reviewers in Triage to know what information is required for a question in a tag they've never seen is... Extremely optimistic; they're likely to miss cases where such information is needed as often as they incorrectly hammer questions that are already sufficient. In fact, we've seen exactly that every single time we've tried. This sort of content moderation requires folks who know the topic. – Shog9 Oct 19 '16 at 3:23
  • 3
    @Shog9 then adding the filter for triage is a must, maybe with a combination of this meta.stackexchange.com/q/254145/213575 – Braiam Oct 19 '16 at 15:43
  • 1
    I think more guidance for what to do as a triage reviewer and what happends with the questions after triage (lots of low rep users who vote in triage have never seen the subsequent queues) is a great idea. I suspect that I have cast hundreds of "wrong" triage votes just out of confusion. – Anders Oct 25 '16 at 12:09

Add cross-site links to the right-hand sidebar

As proposed in Display other SE site links in the linked questions sidebar.

The Linked section of the sidebar is an excellent tool for discovery. Over time we have seen the development of many smaller sites that would vaguely fall under the umbrella of an older, bigger site (like e.g. astro, space, scicomp, hsm and earth science taking up a fair bit of question space that might otherwise have ended up in physics), and one of the worries in this process is the fragmentation of question collections and of communities.

An excellent way to help avoid that is to make explicit those inter-site links where they exist: both highlighting off-site resources that are mentioned somewhere in a given thread and, more valuably, making it easier to discover off-site threads that comment or build upon the thread you're reading (i.e. linkbacks).

This is admittedly a relatively small population of questions, particularly if you insist on comparing apples and oranges by bringing up the total amount of questions on SO (as Jon Ericson does here). That's a dud argument - the measure is the high value that the feature brings to the threads where it does appear.

Ideally, this should also be coupled with a better cross-site question discoverability mechanism (how about a similar Related sidebar pulling from closely-related SE sites?), but first things first.

I'm also aware that there are nontrivial scaling issues in implementing this network-wide, but I find it hard to believe that they cannot be optimized with suitable indexing. If you're looking for places to put new code that will bring the network closer together, this is a good place to do it.

  • related discussion (in Russian) – jfs Oct 18 '16 at 9:34
  • (Scicomp a part of physics? You can't be serious! Computing is part of every science by now!) – Raphael Oct 18 '16 at 15:09
  • @Raphael The claim is (exclusively) that there exist questions that would have been asked on physics that are now asked on scicomp. It says nothing about the relation of scicomp with other sciences and sites (and indeed identical arguments apply to all the other listed sites). – E.P. Oct 18 '16 at 15:35
  • 3
    I would dearly love this feature. Last time we tried to get it implemented, devs decided it was too much work... I still burn a candle. – Shog9 Oct 19 '16 at 0:50

Easier tag handling

Currently, the most effective way to edit tag names or to create tag synonyms is to flag for moderator attention. They have the diamond power™. However, there is one issue that could, in my humble opinion, be made a lot easier a lot quicker; and a second issue that could be simplified but may require thought.

Allow renaming of a tag if it is only applied to one question

This is basically a rehash of this : A pox upon small sites: the thing that prevents you from creating plural or hyphenated versions of tags

Consider a site like Anime where tags are essentially series’ names. A user wants to ask a (the first!) question about , but for some reason misses the hyphen key and creates the tag . If they had created anything else, similar but distinct from (except , which would have the same issue) one could simply retag to the correct one; the wrong one would then be orphaned and deleted at 03:00 UTC. Unfortunately though, the system forbids creating a tag that differs only in hyphenation or a trailing s from an existing one. So we flag for mod attention, waste a moderators time for something that could easily have been handled by any user with edit and retag priviledges.

I propose the solution (not in the present feature request) of allowing renaming of tags, if they are only applied to one question or have been created within the last x hours — whichever is easier to implement software-side. It would allow the first editor to fix the problem immediately and not take up any diamond mod’s time.

Do something about tag synonyms

Improving Tag Synonyms

We all know the system isn’t working unless again you waste a diamond moderator’s time (see linked post). There are a number of issues mentioned in the post and some sound difficult to implement. Here’s a biased selection, not meant to be exclusive:

  • make synonym suggestions more visible

  • if a synonym is approved, make that automatically retag all questions associated with the now-synonym (i.e. like moderator merges)

  • increase the number of people who can suggest synonyms — I mean, what are generalists for, if not for this?

  • Allow users with a score of x to suggest a synonym for an existing tag without having to apply that synonym to a question first.

  • … open to further suggestions.


As far as I can tell about 15 millions views at Stack Overflow are on inappropriate (historically locked) questions.

(285 locked questions have views between 10K and 100K plus 158 questions with views between 100K and 1M plus 3 questions with views over 1M)

Visual difference between historically locked and legitimate questions looks quite subtle to me, even though I have few years experience on the site. It can be even harder for inexperienced readers to tell the difference. This seems to create wide open broken windows.

Due to high views historical questions tend to rank highly in web searches. Inexperienced visitors coming from searches may think that these questions are legitimate and try to ask similar ones.

Suggest to invest some effort into making it easier for new users to see that historical questions differ from regular ones.

  • Related: Use pink background when rendering historically locked posts. (similar analysis at Software Engineering shows about 5M views at historically locked questions over there - meaning this issue isn't specific only to Stack Overflow)

  • Also related: Automatic visual indication of old questions - this request shows an interesting mockup for how historic questions could be rendered:
      Aged image

  • Yet another approach worth considering was proposed in comments below - grayout with un-fade on hover like it is currently done on low-score answers:

    It already exists, it makes it very visually obvious that something is funky with the post, and it still allows for good reading conditions with minor effort. Moreover, it would bring the locked banner to greater prominence on the page.

  • 4
    This locked questions are a very important information source on the internet so should not be made harder to read. – Ian Ringrose Oct 20 '16 at 11:47
  • 9
    That mockup goes too far in making them difficult to read; if we want them to be unreadable, we should just delete them. But I agree some sort of visual marker would help. – Jeffrey Bosboom Oct 21 '16 at 2:36
  • 4
    @IanRingrose there would be no need in anything like that at all if visitors of historical questions were redirected to some kind of an "outcast" / "museum" site that would be physically different and in particular wouldn't show them that big prominent "Ask Question" button (which kinda tells, "hey here's an example question you can ask over here (and expect thousands upvotes and millions views)"). Quite a pity that all requests for such a redirect were declined by SE team so far (example)... – gnat Oct 21 '16 at 7:53
  • ...we are essentially forced to decide whether "harm to the Internet" - obscured text of historic question - outweighs the harm of it hanging in here and making a broken window for inexperienced users (who sometimes simply can't see / understand what historical lock means) – gnat Oct 21 '16 at 7:53
  • 3
    One such alternative is the grayout that's currently in use for score<-3 posts (which re-blackens the text on hover). It already exists, it makes it very visually obvious that something is funky with the post, and it still allows for good reading conditions with minor effort. Moreover, it would bring the locked banner to greater prominence on the page. – E.P. Oct 21 '16 at 10:11
  • I prefer my solution, the lock wasn't meant for being a permanent state either. – Braiam Oct 30 '16 at 1:43
  • 1
    @Braiam per my reading of the respective FAQ permanent state is intended: "Historically locking a post ends the debate over whether a question should be kept on the site or deleted, and is often the final state of a question..." Wrt your solution, I've seen it and while I didn't vote it down, nor did I vote it up because it assumes too much effort on questions that aren't worth it (I understand that these are kept to avoid link rot but besides that I have no interest in doing anything on them) – gnat Oct 30 '16 at 16:10
  • Read who wrote it... Then read this meta.stackexchange.com/a/36966/213575. I'm on mobile but there are more. – Braiam Oct 30 '16 at 16:23
  • @Braiam "post temporarily locked away" in a/36966/213575 talks about different kinds of locks, not historical ones (off-topic comments, edit wars, content dispute etc) – gnat Oct 30 '16 at 17:41
  • It would likely help if the locked banner were at the top of the post rather than below it. – TylerH Oct 31 '16 at 15:13
  • @TylerH I recall a feature request for that. If memory serves some skeptics in discussion over there expressed doubt that this would be of much help. If, on the other hand, the banner would be the only non-greyed piece in the question... – gnat Oct 31 '16 at 15:34
  • Jeez, I didn't even know this was here. Don't post anything late in October if it relates to me. I'm busy having muh spoopy time. youtube.com/watch?v=v4IC7qaNr7I – user1228 Feb 9 '17 at 16:20
  • but you know it @Won't be implemented anyway, so why worry – gnat Feb 9 '17 at 17:11

Give more power to the votes to close, etc. of silver tag badge holders.

The duplicate "hammer" for gold badge holders has worked very well, however we have a lot more silver tag badge holders than we do gold.

This could be one or more of:

  • A close vote by a silver tag badge holders takes longer to time out.
  • Close review queue put questions with a vote from silver tag badge holder higher up the queue.
  • A close vote by a silver tag badge holders result is an automatic vote from community, so doubling the power.
  • Two votes from silver tag badge holders is enough to close a question as a duplicate.

Ability to leave a time-limited comment that also mark another comment as obsolete after a time delay.

  • Often someone comments on one of my post, I edit the post in response and leave them a comment to say I have done so.
  • Both of these comments are of no interest to anyone else in the long term.
  • Having to remember to go back and delete may comment and flag their as obsolete is a pain.
  • If the other person agrees that their comment is now obsolete, it should be deleted without the moderators having to look at it.

Result, fewer obsolete comments, and fewer flags for the moderators.

  • 3
    Why a time limit? "Delete after read by @notified user" should be perfectly possible. – Raphael Oct 18 '16 at 15:11
  • @Raphael, maybe hard to define "read" – Ian Ringrose Oct 18 '16 at 15:18
  • The canonical interpretation would be "viewed on their screen" or maybe "clicked the notification". Of course we can't know if they read it unless we implement some kind of acknowledgement mechanism. – Raphael Oct 18 '16 at 15:19
  • 1
    @Raphael, "clicked the notification" would work, but I think there should also be a timeout, so even if they never come back, the comment does not live forever. – Ian Ringrose Oct 18 '16 at 15:23
  • 1
    "Clicked the notification" as long as that doesn't mean just "opened the inbox menu", for those times when you have more than one waiting. – jscs Oct 18 '16 at 23:58
  • 1
    You may like this proposal of mine going in a similar direction. – Wrzlprmft Oct 19 '16 at 5:01

Flagging Improvements will Help Moderators' Experiences

I find as a moderator that many users get confused about when they should flag and when they should down vote. I see it go both ways. I see spam that gets down voted (but not flagged). I see questions flagged when a down vote would have been sufficient.

Any improvements to flagging should also include changes to down voting. When somebody down votes, a box could pop up asking for a reason. Clicking on some of those reasons could escalate the down vote to a flag. There could also be an "other" field that would be automatically left as a comment.

Some reasons for flagging could just turn into a down vote. Something like "there are inaccuracies" or "I disagree with this".

I'd also like users to able to flag posts to enter queues. It would be nice to be able to flag questions that need to be edited for grammar, spelling, or formatting. There are plenty of times where you may not be able to do that edit right then, but putting it in a list of things to get taken care of would make the site better.

  • 4
    I agree with a more general shift from "what do you want to do" over to "why do you want to do it" and I'm not sure why you're seeing downvotes because the problem pretty much is people don't know when to vote or when to flag. I think the sequence would be more natural if they left the flagging box, realizing that they should just vote instead (if they have that ability) while we gather more anon feedback from folks that don't have the ability to vote, but indicated that they clicked on flag because "this just annoys me" :) – Tim Post Oct 17 '16 at 18:12
  • I'll ask the same thing I ask every time the downvotes-on-spam thing comes up: What real, practical harm does it do? – Undo Oct 17 '16 at 18:12
  • 2
    @Undo None if you flag it. But if you don't flag it and just vote, then it delays the luncheon meat slaughter :) This is going to be tricky to fix from a UX perspective. – Tim Post Oct 17 '16 at 18:13
  • @TimPost Of course... but is the few seconds we delay that really worth throwing time and work at? I'd like to see stats on how often it's actually a measurable issue. – Undo Oct 17 '16 at 18:14
  • @TimPost The downvotes are probably because people are reading too quickly, seeing "When somebody down votes, a box could pop up asking for a reason", and thinking "Oh no, not the discussion about justifying downvotes again!". I, for one, must confess that I had done exactly that before noticing your comment :) – duplode Oct 17 '16 at 18:27
  • I didn't "read too quickly". I don't understand what relevance changes to flagging guidance has to improving the quality of the material on Stack Exchange. – jscs Oct 17 '16 at 18:49
  • @TimPost: I downvoted because it seems like it focuses mostly on "flag mechanics," rather than education. Proper flagging is not a software problem, really. – user102937 Oct 17 '16 at 18:52
  • 4
    @JoshCaswell It surfaces stuff where mods are really needed much better, which helps them drill down into things that make a bigger difference. I can't tell you the nerd rage I had as a mod on SO declining "pls I need answer" as an "other" flag on SO, and it happened nearly 10 times a day. It's worse now. – Tim Post Oct 17 '16 at 18:52
  • 1
    Okay, that's a very good reason for looking at this, @TimPost, which I had completely missed. Thanks for explaining! – jscs Oct 17 '16 at 18:53
  • 2
    @RobertHarvey It's a UX problem, and a hard one at that. You have all of these options and it gets quite complicated remembering which one is appropriate in any given context (not for me or for you, but for many). So we're shooting for more education, without getting in the way of useful flags too much. I could see a downvote on a post that has been flagged as spam bringing up "Mmm, maybe flag too?" and giving the user the option to do that right there (for instance). – Tim Post Oct 17 '16 at 18:58
  • 1
    I've edited my question to explain this a bit better. I'm sorry @Stephen, this was a good direction in which to think and my vagueness there didn't fill in some gaps that most people don't ever see :) – Tim Post Oct 17 '16 at 19:06
  • 1
    No problem, Tim. Thanks for weighing in. – Stephen Ostermiller Oct 17 '16 at 19:08
  • 4
    To be clear, we're not going to ask for a reason for downvoting, we're more looking at is that really what you meant to do? based on data (flags, spam system score, etc) and present the other option for them to pick conveniently. And this is all in theory at this point. – Tim Post Oct 17 '16 at 19:17
  • 3
    @TimPost Could we remove "close" and "flag" in favor of a popup dialog that appears after a downvote? Along the lines of "Why did you downvote? ..."; it would then cast close votes or raise flags (is the distinction really all that meaningful?) according to the answers. – Raphael Oct 17 '16 at 23:03
  • 4
    Even if we didn't use it to replace flags and close votes, asking downvoters to select (the category of) their reason to downvote would be very useful. 'You received 5 downvotes: 3x "factual errors" and 2x "bad form"' is a lot more useful than 'You received 5 downvotes'. Yes, downvoters should comment, but we know they often don't. – Raphael Oct 18 '16 at 0:32

I can describe my big desire, as a power user and an expert in some cross-site areas: discoverability and manageability of communities and questions, preferably without needing to be actually on the network at all times.

It's not a feature request, at least in its present form, but:

The cross-site Filtered Questions on Stack Exchange are a bit awkward to use, making site and question discovery extremely difficult

It would be nice to have better discovery tools. There are a lot of sites in the network. Just looking at my professional interests, I can answer questions across over half a dozen sites.

This fits in with your filtering stuff that was mentioned in the question, but being able to put the right communities and the right questions in front of the experts is only beneficial. Ideally, you'd see higher quality answers to questions.

Something tangentially related to filtering and discovery:

Additional RSS feeds

As long as you aren't as aggressive in blocking feed readers as Reddit is (my Reddit feeds in Feedly rarely get updated because Feedly hits the Reddit rate limits and Reddit refuses to increase their allocation), RSS feeds can lead to some interesting discovery, data analysis in a closer-to-real-time situation, and integrations with third party tools (using something like IFTTT or custom software).

  • 3
    Going to dig up numbers on how often feeds are used. I know they plummeted when Google killed reader, but if it's steady and not tiny, it's definitely worth looking at this, even perhaps expanding it. – Tim Post Oct 17 '16 at 17:48
  • 7
    @TimPost I don't use feeds now, and there are two reasons. First, it's hard to find the feeds that do exist. Second, as far as I'm aware, the feeds I want don't exist. I would be hesitant to base a decision only on the usage numbers. If people don't know that the current feeds exist or the right feeds don't exist, of course the usage will be low. – Thomas Owens Oct 17 '16 at 17:49
  • Good point, re: usage. I'm going to look into it. – Tim Post Oct 17 '16 at 18:16
  • @TimPost Thanks. Admittedly, I don't know how easy it is to implement various feeds. I think that ease of implementation vs what people can do if they choose to use them would probably be better. I've got some ideas of what I would do with IFTTT (automatic posting of questions and answers on various site to Twitter/LinkedIn, for example) as well as building applications around discovery and RSS feeds across the network. – Thomas Owens Oct 17 '16 at 18:22
  • 2
    I just discovered that some searches have their own RSS feeds (queries with only tags work) but others don't. Giving every search query it's own feed (essentially, that would mean providing a search API) might be interesting. For instance, I may be interested in [favorite-tag] answers:0 score:1 closed:no for discovering questions I may want to answer. – Raphael Oct 18 '16 at 14:56
  • @Raphael: That kind of query actually worries me. That's what I would call a drive-by. It's meant to sus out the questions that are most likely to give you rep, not necessarily the ones that need an answer. The ability to filter that way would make we worry that many users' questions may never get good exposure if enough potential answerers utilized it. – Chris Pratt Oct 27 '16 at 18:47
  • @ChrisPratt Actually, this is a search I would make when I have some time to kill to discover unanswered questions that may be worth answering. In particular in beta, there is some incentive to keeping the percentage of answered questions high. I agree that rep farming can be a motive, but it's certainly not the only one. (And rep farming is only bad if you get upvotes on crap.) – Raphael Oct 27 '16 at 20:16

Multiple Entry Boxes

Right now, everyone enters their questions and answers into a single box, and we rely on all users to use understand what a good question should look like. For the legions of users who come here from Yahoo answers and other question sites, there is no clear indication that SE requires a little more from its users.

I think for users with less than 200 points, instead of a single text entry are for the question, there should be two:

  1. The actual question
  2. How the asker has already attempted to solve their own problem.

So for Mathematica SE, a users would see two boxes - one for the question, and one to enter code and describe how they have already attempted to solve their problem.

For History SE, the second box would required what a preliminary search has revealed.

Having a simple form like this will start their walk through asking a good question.

  • 6
    And do you propose requiring the user to enter at least 30 characters in each box? 15 in each?  If the use of the second box is mandatory, it will contribute to the "SE is too complicated/unfriendly" mentality (and possibly discourage and annoy users).  If it's not mandatory, then it will be ignored by the people who currently ignore all the other guidance they are offered. – G-Man Says 'Reinstate Monica' Oct 18 '16 at 20:38
  • People to lazy to read the guidlines won't know about initial research. A separate box, clearly labeled, won't be missed. – axsvl77 Oct 18 '16 at 22:14
  • 2
    It wouldn't be hard to label the two boxes: Tell us what you'd like to know: and Tell us what you found or worked out so far:. People then know exactly what they're expected to provide - a question along with some context and prior effort. It also places those two things in obviously traceable areas, so especially on sites that deal with a lot of HWQ, potential answerers can scan quickly for where the asker so far reached. – Nij Oct 20 '16 at 11:09

Another pet popular-but-not-implemented proposal: opt-in notifications for specific posts.

EDIT: The original question asks about notifications for particular questions. I'd actually like separate opt-in notifications for questions and answers; a new answer would generate a notification for a question, but edits to that answer would not generate further notification unless the user requests notifications for that answer. This would be an important distinction on questions with lots of answers (such as this "Plan the Second Iteration" question!).

Proposal stats

Teach the controversy!

This request seems potentially more controversial than my other request, so I'll try to address potential criticism in Q&A format:

Q: How is this different from marking questions as "favorites"?

Note that all of my "duplicate" questions asked above are actually marked as duplicates of the official "favorite questions" FAQ question. This is not completely ridiculous, especially since one of the OPs mentions that they're specifically looking for something to replace in-browser bookmarking.

A: The proposal is for new notifications.

Marking a question as a "favorite" simply puts it in a list somewhere. In order to see if a "favorite" question has been updated, the user must remember to periodically check each and every favorited question for updates. I just checked, and I have over three pages' worth of favorited questions across the network. There is essentially no chance that I will go back and check each of those periodically for updates.

Additionally, there are reasons to "favorite" questions even when the user doesn't care about updates to the question. For instance, I "favorite" questions that I find merely interesting or amusing, or which explain how to solve particular problems that I encounter periodically.

Q: Don't you already get all the notifications you need?

For comment discussions, we already have @-notifications and the auto-notification feature that kicks in when the OP write a comment and only one other user has commented.

A: No, the existing features really are insufficient.

  • There was a recent meta-question (I can't find it at the moment, unfortunately) asking about whether we could add a feature to promote use of @-notifications among new users.
  • @-notifications are limited to notifying a single user. If someone has already written a comment asking a question to which you'd like to know the answer, you'd have to ask the OP to @-notify you in a separate comment if and when they get around to answering the user who originally asked.
  • Similar to the above point, sometimes users may be interested in conversations to which they haven't yet actually contributed.

Q: Won't this generate too much noise?

A: That's why it's opt-in.

...and, once you've subscribed to updates, you should be able to unsubscribe, too.

This is a model that's already been deployed (and is in my opinion a success) on Bugzilla, GitHub (you can subscribe to and unsubscribe from updates for individual issues or for an entire repository), and Facebook. I believe it's even implemented on Discourse, another Jeff Atwood Project™.

Q: But wait, those example sites are social networks! We don't want more social networking features here!

A: This functionality is not specific to social networking.

It's unclear that the particular subscribe/unsubscribe feature is specifically related to social networking. This is a simple but extremely useful tool for customizing the notification features that already exist on SE sites.

Additionally, as pointed out by NathanTuggy below, Bugzilla (which previously wasn't mentioned in the list above) implements fine-grained control of per-issue notifications, and it is in no way a social networking site.

Random trivia

According to Jeff Atwood's answer on the linked proposal, this was actually implemented, then pulled, then replaced with the "favorites" feature.

  • 9
    This would, to a degree, mitigate the problem of downvotes that are never removed because the downvoter didn't know the post was improved. – user102937 Oct 24 '16 at 20:19
  • 3
    A better example than GitHub might be Bugzilla, which allows all kinds of coarse- and fine-grained notifications for bugs or categories of bugs, and is in no possible sense any sort of social networking software. – Nathan Tuggy Oct 24 '16 at 22:56
  • @NathanTuggy Thanks for pointing that out! I've added it as an example and modified the last "A" section. – Kyle Strand Oct 24 '16 at 23:20
  • Related feature request. There’s also a related other suggestion – Jan Oct 25 '16 at 13:24
  • @Jan Thanks for pointing that out! I've added it to the list. – Kyle Strand Oct 25 '16 at 16:09
  • 1
    This is even more important, as "favourites" are also marked as modified if the "community" bumped them to the current list of active questions. – samcarter_is_at_topanswers.xyz Oct 25 '16 at 16:12
  • @KyleStrand Just to reduce any confusion that may or may not occur: My proposal is about new questions in favourite tags, not any edits to favourite questions. But of course, a favourite tag can be loaded with new questions, too ;) – Jan Oct 25 '16 at 16:16
  • @Jan Definitely definitely definitely missed that word. It does bother me a bit that we've overloaded the meaning of "favorite" -- oh well. Thanks for clarifying! I'll make another edit. (I also didn't notice that that was your proposal.) – Kyle Strand Oct 25 '16 at 16:25
  • 1
    @KyleStrand It’s just that we love this network so much, everything is favourite, right? ;) – Jan Oct 25 '16 at 16:26

Improve searching operators by adding options to sort

Searching plays an important role in find posts, matching duplicates for closure votes, and possibly many more.

I enjoy the search operators that can be used to limit the scope of one's search. Perhaps one can add more of these:

  • Be able to sort your search as part of the search query using (say) sort:votes (for a descending list of posts by their score).

    At the moment, this is a click away with any query. When trying to find duplicates that I know are highly voted, I have to search and then click the sort. A sort: operator would help here.

  • Be able to reverse the sort order using (say) sort:oldest or sort:-newest to have a look at the oldest posts matching your query.

    At the moment, there is only one suggested order each of the tabs (newest, votes and active).

  • 2
    Can't you just navigate to the last page of a sorted search? – Nathan Tuggy Oct 24 '16 at 22:49
  • @NathanTuggy: There's an additional click (or two + scroll) to reach this sorted search (and then the last page). – Werner Oct 24 '16 at 22:53
  • Well, I mean, that's sad, but is saving two clicks and a keystroke on a somewhat unusual search option really worth implementing a new feature (set)? – Nathan Tuggy Oct 24 '16 at 23:00
  • @NathanTuggy: This is just a suggestion. Perhaps it's simple and saves a bunch of clicks. Perhaps it's not and won't be implemented because it's super difficult. I don't know, but my gut suggests the former. – Werner Oct 24 '16 at 23:01
  • It's a feature. So figure a couple hours to spec, a couple hours to implement, a few hours to test, a few hours to document, another few hours to handle regressions… and when all's said and done, it's just another search keyword that helps the few thousand users that even know about it to save a few seconds a few dozen times a year. – Nathan Tuggy Oct 24 '16 at 23:05
  • @NathanTuggy: Like most (if not all) feature requests here. – Werner Oct 24 '16 at 23:11

Enable the Triage Review Queue Everywhere

and possibly improve it.

As a moderator, these criteria make me edit questions the most often:

  • Has only one tag.
  • Does not have a top 5% (?) tag.
  • Has tags that don't usually occur together.
  • Has new or rare tag.
  • Has a tag from the (implicit) black list.
  • Has title that is very similar to many other titles.
  • Has no or very little Markdown.
  • Is only one paragraph.
  • Has MathJax in the title.

If there was a review queue with questions that fulfill any of these or similar criteria, that would not only help me find them but also enable others to help with the task(s).

When a reviewer views the question, the reasons for it being in the queue should be highlighted.

  • 8
    I could honestly see replacing "First Posts" with Triage on most sites. Having both of them is a waste of folks' time though. – Shog9 Oct 20 '16 at 1:18
  • @Shog9 Any movement on this? Asking for a user. – Raphael Jan 23 '18 at 15:12

Leave closed but change the close reason

Sometimes questions have been closed with a close reason (unclear, too broad, off-topic) which was correct at the time. Then, the OP edits the question. This now makes the former close reason obsolete but lets a different one kick in: for example, a question which used to be unclear is now off-topic because site-specific off-topic reason 2. Currently, the only thing possible is to either leave it as is, ask (/flag for) a mod nicely or attempt to reopen followed by reclose.

Instead, a functionality could be added to the reopen review queue along the lines of leave closed; apply different close reason. With a sensible majority of e.g. three people (which would kick the post from the queue anyway as is) the new reason would be applied.

If a closed question acquires both leave closed and change reason votes, let it go up to three identical votes. If the users cannot agree on a new close reason (e.g. two go for too broad, one for off-topic reason 2), wait for five close-type votes in total. If a majority of three hasn’t been reached, leave the reason as is.

I’ll admit that this is a not-so-often encountered feature, but in my opinion SE would benefit. It would be much clearer than commenting on a user’s comment ‘Why is my question still unclear?’


For sites with large review queues, automatically set the review filters based on the user's most "active" tag(s)

  • I can't be humanly expected to review 1000+ posts on Stack Overflow

    Yet that is the expectation conveyed to me whenever I glance at the top bar.

  • I don't grok half the questions because I am not familiar with those technology stacks

    So how can I be expected to make a call on what is enough information and what is too broad?

  • Yes, there is a filter that can be manually set

    But it is not obvious how to set this. And why should I have to set it when my site activity is captured as tag votes, question views, etc.?

I don't know what the right definition of "active" should be here. Here are some possible metrics:

  • upvotes
  • views on questions with those tags
  • proportion of questions/answers in that tag compared to total questions/answers
  • a combination of the above
  • 4
    Personally I think the filters should not be automatically set, but that the questions matching the filter should be sorted based on the skills and engagement the review has with each tag. – Ian Ringrose Oct 26 '16 at 13:37
  • Perhaps there should be a filter to give meaningfull numbers in the top bar, but I would indeed agree that for the actual queue sorting would be more appropriate. – Dennis Jaheruddin Oct 26 '16 at 15:10
  • Maybe the review queue should not show any numbers…. – Ian Ringrose Oct 26 '16 at 17:11
  • 3
    Can we use a bit of machine learning, to build a model of the review tasks that a person is most likely not to skip, and show them these? (Along with the tasks that are most likely to result in a outcome.) – Ian Ringrose Oct 26 '16 at 17:13
  • 2
    Related meta.stackexchange.com/q/254145/213575 – Braiam Oct 27 '16 at 0:05
  • "So how can I be expected to make a call on what is enough information and what is too broad?" - you aren't. That's what the Skip button is for. – Nij Oct 29 '16 at 1:03

Give an incentive for finding duplicate questions and at the same time can we have a platinum badge for questions that get more than 500 upvotes:-)

At present if you see a question that may be a duplicate and has an easy answer you can post the answers or post a link to the duplicate question. (Or the few hi-rep users can vote to close it as duplicate, think of normal users here, not the people that read Meta)

It is better for Stack Overflow if a link to the duplicate is posted as a comment, or the user votes to close the question as a duplicate. However, the user gets more rep if he/she posts the easy answer quickly.

  • 6
    If we're going to reward rep for linking to SO Documentation, we should reward rep for closing as duplicate. – Jeffrey Bosboom Oct 28 '16 at 22:33

Improve Searching

I know this is a can of worms, but let's face it: the search results usually ... disappoint. When I really need to find something, I use Google with site:x.stackexchange.com.

Part of this may be the "Relevance" sorting criterion. How does that work? Searching Meta I just got five-years-old posts that have right at the top.

  • What about inline search by Google? – RudolfJelin Oct 19 '16 at 17:42
  • @RudolfL.Jelínek I don't know what you mean by that. Should SE use Google to drive their search? I don't know. The SE-specific parts of the query language are useful. – Raphael Oct 19 '16 at 17:56
  • Yes, that's what I mean. Maybe it would be good to let users choose the search method before searching, SE search would be the default. – RudolfJelin Oct 19 '16 at 18:03
  • @RudolfL.Jelínek Could work. I wonder if the two could be combined, or Google result presented through the SE stack. – Raphael Oct 19 '16 at 18:06
  • Imagine the situation: Daily news: "Stackexchange combined forces with Google to make a new inline search engine!" – RudolfJelin Oct 19 '16 at 18:09
  • I just search site:stackexchangedomain.com stuff I want to find. Works way better. – enderland Oct 21 '16 at 13:00
  • 1
    The various SE sites are well positioned on google. Of course it doesn't feel right that the best way to find SE content is via an external resource, but personally I don't really know whether I consider this to be a priority issue. -- Semi serious note: if people can depend on finding stuff via google, they just have 1 to-go spot, also for things that are not properly covered on SE. Not sure if this is the best for SE, but it could be the best for the programmers of this world! – Dennis Jaheruddin Oct 26 '16 at 14:53
  • @DennisJaheruddin Honestly, I'm not sure we can generally say that people can rely on Google. I know that my filter bubble is probably tuned heavily towards computer science and Stack Exchange, so I can use Google for finding stuff on Computer Science. I don't know if that's the case for everybody else, especially newcomers. – Raphael Nov 7 '16 at 10:10

This suggestion is three years old. It currently has a net score of +17, is not tagged with /, and only a single answer at net score +4. The latter even missed the entire point of the feature request, which is

Allow a single close-voter to give multiple reasons.

I have hit a dozen questions across several sites in the past week where multiple reasons applied. Some of them are straightforward: it was both too broad and primarily opinion-based. Others were a bit more complicated: either reason X applied, but if it didn't, then reason Z would logically have to apply, instead.

This is problematic in more than one way:

  • if everybody votes closure for a single reason when two or more apply, and the first reason is inapplicable after editing, we now have a question closed on poor grounds, an asker confused or annoyed, and a dilemma for voters or moderators.

  • close-voters have to pick the one reason. As an extreme example, someone may be unable to actually make that choice, and therefore doesn't pick either. I have not experienced the difficulty yet, but it has caused real (physical and mental) anguish.

  • borderline cases get left unresolved, because each reason makes it hard to deal with others (it's hard to explain the way it is structured in my mind, but, imagine trying to see one tree in a forest without ever looking at others). Again, no close-vote.

There are several ways to implement this:

  • each close-vote is a close-vote. Only the number matters; the reasons are a check box that select the information a close-vote wants to show to the asker.

  • the close-vote gets divvied up. Pick two reasons and they each get a half-tick, three get a third, etc. If a reason no longer applies (other responses deal with changing the reason for a close-vote while it is still closed, due to edits) then reasons can be unticked, but the close-vote remains on the counter.

  • ...?

  • 6
    I'd above all love to see my voting reason shown, even if others voted for a different reason. (Especially when 2 or more others voted for "belongs on another site" when I voted "off-topic", "unclear" or "too broad" or "plain crap that should not be moved" if that would exist.) – Arjan Oct 23 '16 at 12:17
  • 5
    Right now, if multiple close reasons are voted on, the final close displays only one reason, but shows all voter names on it, even if that is not the reason they voted on. The close should display all reasons voted on, but not necessary who voted for which particular reason. Something more along the lines like "Closed X hours ago by Joe, Jane, Mike, etc for the following reason(s): - reason 1 - reason 2 - reason 3, etc" – Remy Lebeau Oct 29 '16 at 0:53
  • @remylebeau all reasons for closure that receive two or more votes are shown. This doesn't often happen because one reason is very obvious and others are not, or because there are two equally applicable reasons and one is easier to click to, but where a question is e.g. vaguely mentioning something completely off-topic, it might get three votes for unclear and two votes for blatantly off-topic. I believe that suggestion has been made on another answer, it's related but separate from this one. – Nij Oct 29 '16 at 0:58

Give us notifications for absolutely everything, and then make the non-essential notes opt-out.

Other responses have already pointed out a number of events that we receive zero notice of, despite the importance they have, to our work here.

But it's also been implied that some (especially higher-rep) users would be bombarded, and in any case, the individual might simply not want to hear about that event.

So, there are two requests here:

  1. Give us notifications for everything. See previous responses for some examples of what this includes.

  2. Let us make them go away, if we want to. I'll expand on this next.

That latter request can be fleshed out:

  • when a person first receives a new type of notification, add a quick phrase to the notification box saying that it is the first such time, and pointing out a check box they can use to "mute" this type of notification. Either grandfather existing notifications or apply this rule "from now on" at implementation.

  • add a link in the notification box which directs to a page describing the different notification types, and how to globally mute/unmute them when that is possible. This could even be on the same page. Make it a very obvious symbol and mode indicator - say, a red barred circle versus a green exclamation mark - right next to the name of the notification.

  • make the muting question-specifiable. I don't want to hear about new answers on just that question, but I do need to know about other activity - bam, "mute answer notifications on this question" and I'm happy. Not interested in anything except closevotes or deletion - bam, "mute [everything else] on this question" and I can be on my merry, knowing I'll be told if that question might or has been removed.

  • Subsequent to both of the above, separate the muting/unmuting for specific questions from the global toggle. I might normally block all the new comment notes, but want to keep track of a particular discussion for a couple of days, or change my standard preferences but keep the special mutes/unmutes applied. Possibly incorporate a "nuke" button as well, that removes all my current special mutes/unmutes.

  • This kind of functionality already exists in the ability to block tags on a particular site, or even entire sites from the main page and sidebars. It enables us to get all the information we need at the time we need it, and also to ignore the information we don't need or want, and the change is as simple as a couple of clicks while we happen to be thinking about that thing already.


Ban "Try this [code block]" answers

I am getting tired of those numerous posts (in my case on Stack Overflow, but I guess there are similar cases on other sites), that say:

Try this


I know we can't make rules for each and every low quality post indicator, but in my opinion this one does need attention.

I propose to show a notification to the user that "Try this" answers are discouraged and should be accompanied by some explanation. Maybe we should subtract the code block character count, so "Try this" would fall out the 30 character limit enforcing a little explanation already.

  • Given that code does not work in comments, sometime there are few other options then a "Try this [code block]" when you think you may understand the users problem. Hopefully the user responds and then the answer can be deleted or expanded. – Ian Ringrose Oct 27 '16 at 9:12
  • 8
    "Try this" is never helpful. At least explain what has changed and how that fixes the issue at hand. After such answers, the user is still puzzled what he did wrong. He will ask the same question the next time since he didn't learn how to fix it. @IanRingrose – Patrick Hofman Oct 27 '16 at 9:19
  • @IanRingrose: In situations where you need to post code, but aren't giving an answer, per se, you can use the community wiki feature. Just act as if you are posting an answer, and check the the community wiki check box before submitting. Community wikis do not grant rep, so you're not taking anything away from real answers. – Chris Pratt Oct 27 '16 at 18:35
  • 5
    Also include "Like this"... – Werner Oct 28 '16 at 21:38
  • 1
    @ChrisPratt community wiki isn't a excuse to give a poor answer. Either you properly answer the question or you don't. – Braiam Nov 7 '16 at 18:19
  • 1
    @Braiam I didn't say it was. However, there's many situations where code or long prose may be beneficial to the user, even if it doesn't actually answer the question. That's a perfectly valid use of the community wiki feature. – Chris Pratt Nov 7 '16 at 18:23
  • @ChrisPratt again, it's not excuse for a poor answer, including "answers" that don't answer the question. – Braiam Nov 7 '16 at 21:06

Support for tables/tabulation in posts

I'm not particularly concerned about how this feature is implemented, but it is cumbersome as a user to have to reach out to a 3rd-party ASCII-table utility (like this one) in order to convey one's message more clearly.


Open Stack Exchange to the other 80%

Something like 20% of the world population can read simple English. Stack Exchange should expand to the other 80%.

Yes, internationalization is hard. But Stack Exchange has been around for a long time. It's diversified away from being for English-speaking programmers writing about programming, to being for English-speaking programmers writing about diverse topics, and has opened up to English-speaking non-programmers. Now, very slowly, Stack Exchange is opening up to non-English-speaking programmers. Open up to the rest of the world!

6–8 years ago…

Our mission is to make the Internet a better place to get expert answers to your questions. Nothing about that mission says the questions have to be in English. It is our long term goal to make the Stack Exchange Network a great, planetary resource for all the world's citizens no matter what language they speak.

3–4 years ago…

ETA: still 6–8 arbitrary units of time.

A little over a year ago…

We're saying not yet.

And still nothing coming outside of a few programming sites. And no respect for the existing communities of the sites about languages), which have all (at least the ones about non-extinct, non-artificial) been built by a small set of very persistent native speakers driven by the hope that the site would eventually open up to everyone, and not just the few who also happen to speak English well enough to participate on Stack Exchange.

In concrete terms, I'm only making some very cheap requests:

  • Provide a user interface and the official documentation in all the languages of the existing sites (at the visitor's choice, of course). Cost: small, supporting a dozen languages isn't much more expensive than supporting 4.
  • On each site, officially allow meta posts and chat in the site's language(s). Cost: 0.
  • Allow proposals for non-English sites about topics other than programming on Area 51 (and allow them to go through if they reach the launch stage). “We can't allow T in L yet because we want to have an employee who speaks L” makes sense, “We won't allow T in L because T ≠ programming” doesn't. Cost: 0.
  • 10
    @Magisch If you'd like to spew this xenophobic crap, please use the “downvote” button and refrain from commenting. Also remember that Stack Exchange is not only about programming. – Gilles 'SO- stop being evil' Oct 19 '16 at 7:33
  • 8
    I'm not a native english speaker myself, and that isn't xenophobic. Programming happens in english, and a strong command of the english language is the single most important soft skill for any programmer. What happens when you allow non english posts on for instance TWP or parenting or whatnot is that you mitigate the existing communities' ability to self govern - lots of people on there speak only english. And if you feel that stuff like "parenting in japanese" could gain sufficient traction, I mean, no, probably not. – user308386 Oct 19 '16 at 7:35
  • 5
    I understand this answer to propose that the SE UI should be available in other languages, not that we should post in Babelian ways. (cc @Magisch) – Raphael Oct 19 '16 at 14:25
  • 6
    @Magisch Maybe you are not aware, but most sites on the SE network are not about programming. – Raphael Oct 19 '16 at 17:57
  • 10
    @Magisch Much of the content on German Language is in German. Much of the content on French Language is in French. Much of the content of Japanese Language and all of the content of ja.stackoverflow.com is in Japanese. etc. There is no necessity that content on Stack Exchange should be in English only, it's just a limitation that was supposed to be lifted at some point. – Gilles 'SO- stop being evil' Oct 19 '16 at 19:02
  • 5
    Maybe some SE admin can shed some light on @Magisch theses. I for one think it's a narrowing point of view; there's no reason to believe the diverse spread of sites attracted only programmer types. As a matter of fact, there are probably very few on sites like Mathematics and Computer Science (even though those all speak English for other reasons). Anyway, it's a self-fulfilling prophecy: as long as the platform is only available in English it will only attract people with command of the English language. Is there a particular reason for this platform-self-limitation? – Raphael Oct 19 '16 at 21:10
  • 10
    @Raphael To give an example, among the 36 users on the top page of French Language by reputation, 18 have >2k rep on SO (serious hobbyists or professional programmers), 8 have 200–2000 rep on SO (occasional programmers) and 10 have <200 rep or no account (non-programmers who at most asked about tweaking a script once or something). All can at least understand written English (otherwise, they just wouldn't be there) and most can write English (again, the ones who can't tend not to be interested) but I know at least a couple have trouble with English even if they can cope with the basics. – Gilles 'SO- stop being evil' Oct 19 '16 at 21:22
  • 5
    Calling people Xenophobes... The new Godwin's Law. – user102937 Oct 20 '16 at 1:08
  • 6
    I'm confused. Do you want the current SE sites to start accepting non-English questions and answers? Each SE site to automatically get other-language versions in each "supported" language? Just translate the interface, but expect English Q&A? Something else? – Jeffrey Bosboom Oct 21 '16 at 2:43
  • 9
    @JeffreyBosboom I want what Joel said in 2010: SE to open up to non-English sites (and not just about programming), and to provide a non-English interface to the sites where posts in other languages are already allowed (all the sites about languages other than English and other than Русский язык are bilingual: English and the topical language). I don't want to have omnilingual sites, I don't think that can work. – Gilles 'SO- stop being evil' Oct 21 '16 at 20:04
  • 9
    You might want to add that last comment to your answer. As it stands, I can't understand what you're actually suggesting. – terdon Oct 24 '16 at 13:59
  • 5
    @terdon Correct me if I’m wrong, but I believe this comes down to: “If there's a monolingual non-English XX.so site with an existing user interface in language XX, make its UI optionally available on the corresponding bilingual XX.se language site so that the language site can attract more members whose facility with their own language exceeds their facility with English.” So for example make the Spanish UI from es.so an optional UI on es.se, the Portuguese UI from pt.so on pt.se, and so on and so forth for all paired XX.so and XX.se sites. This seems easy enough, and desirable. – tchrist Oct 24 '16 at 14:25
  • 8
    @tchrist that does seem like a good idea. I just can't tell what Gilles is actually proposing here. – terdon Oct 24 '16 at 14:29
  • 4
    @MonicaCellio The “X Language” sites (where X ≠ English) are bilingual for content by community choice: you can ask and answer either in English or in X. This wasn't a Stack Exchange policy but AFAIK all sites came to the same conclusion. I expect most sites to be monolingual: only a few topics would warrant multiple languages (beyond sites about languages, perhaps topics like “Indian culture” and “Canadian law”, none of which currently exist). – Gilles 'SO- stop being evil' Oct 25 '16 at 11:50
  • 5
    @DennisJaheruddin “Provide a user interface and official documentation” means just that, it doesn't mean to allow site content in all those language. At least French Language should have a French UI in addition English UI. I don't see why you'd forbid a German UI on French Language: assuming the UI exists anyway (for German Language), why wouldn't the UI language be a global preference of the user? And you can ask questions in French on French Language, this has been possible since day one. French Language is bilingual in terms of content (as are all the non-English language sites). – Gilles 'SO- stop being evil' Oct 27 '16 at 10:12

Help users understand that comments are not for answers

Across all sites on the network (though with varying frequency), inevitably, questions end up with "mini-answers" in the comments (even from high-rep users who should know better). Clearly, something isn't working here. Comments are for comments on the question (like asking for clarification), not for answers.

From my experience, here are the main types of question comments (ignoring spam and rude/offensive comments, as they aren't really relevant here):

  1. Asking for clarification/trying to help improve the question: This is the main intended use for comments.
  2. References to other questions/answers: These are very helpful - they populate the "Linked" section on the question with posts that are related (as determined by users, rather than the posts in the "Related" section that are chosen by the software). Often, the question asker or a potential answerer will see information on a linked post that helps them arrive at a solution.
  3. Giving a partial answer: This is not a valid use of the comment system, nor is it a good answer. Incomplete, unhelpful answers are not welcome as answers or comments.
  4. Giving an answer that the commenter is not certain works: In my opinion, this is not really different than giving a partial answer: it may or may not be useful to the question asker, and it certainly isn't of the quality that we expect from answers.
  5. Jokes and chatter: While these sorts of comments can be well-received, they're not really what comments (in their current state) are meant for. In my opinion, chat is a better vehicle for discussion of questions (we have the "migrate to chat" button for a reason).

We need to emphasize that types 1 and 2 are the appropiate uses for comments, and other types should be made in chat (or not at all).

  • 3
    You may want to elaborate on the three main reasons for this: 1. helping the asker of an unsalvageable question; 2. giving an extremely partial answer; 3. giving a provisional answer the user is unsure of. (The last two partly overlap, and also partly overlap with legitimate troubleshooting advice intended to make the question more detailed.) – Nathan Tuggy Oct 17 '16 at 23:59
  • @NathanTuggy I agree with your suggestion. I wanted to add more elaboration, but I hadn't yet figured out how I wanted to elaborate. – user307833 Oct 18 '16 at 0:53
  • 1
    We do enforce the no-answering-as-comments rule on Code Review. A lot of it has to do with site culture and moderator activism. (The only people who can delete comments are the moderators and the poster.) – 200_success Oct 18 '16 at 7:50
  • 1
    @200_success CR is the exception to the rule, then. On many sites across the network, moderators don't remove answers-as-comments, or worse, post their own answers-as-comments. – user307833 Oct 18 '16 at 7:52
  • 2
    Often a "link only answer" solve the problem, but I don't have time to write a full answer. Therefore I leave the link as a comment. – Ian Ringrose Oct 18 '16 at 14:02
  • 4
    I do this a lot on SO, usually when I expect that a new questioner is about to get steamrolled by a hostile hivemind and I won't have time to write a proper answer before the post is closed. Fix that first. – zwol Oct 18 '16 at 14:08
  • 2
    I also do this occasionally. I'm not sure how it harms the site. – Kyle Strand Oct 18 '16 at 14:24
  • 2
    @IanRingrose Then just don't answer or comment. It's not a race. – user307833 Oct 18 '16 at 14:24
  • @KyleStrand You don't see the problem with blatantly disregarding the official policy on what comments are meant to be used for? – user307833 Oct 18 '16 at 14:27
  • 2
    @Mego, it is about helping the person, even if you are not going to write an answer that will help google. – Ian Ringrose Oct 18 '16 at 14:32
  • 2
    @IanRingrose You can help them (and yourself) more by writing a proper answer. If you don't, either someone else will answer, or you can write a full answer when you have time. Like I said before - it's not a race. – user307833 Oct 18 '16 at 14:52
  • 2
    I tend to do this if I have a suspicion I have an answer, but either lack the time or interest to actually form it into a full answer, and know that posting that as an answer will end up with me pounded with downvotes. So I think @IanRingrose has something of a point. Fix the motivations for people doing this in the first place - or provide some alternative - which is easier said than done. It's telling you would rather have me say nothing than actually address the issue. – user295616 Oct 20 '16 at 17:22
  • 1
    @IanRingrose re: "it is about helping the person" I disagree. SE is about forming constructive and useful Q/A. Helping individual/specific cases is not really the end goal. Digging thru comments to find answers does not help make this a good Q/A site. – user343082 Oct 30 '16 at 16:00
  • @WilliamKappler I think you are over-inflating the role of the individual here. It doesn't matter who answers the question - just as long as it gets answered. If you aren't the one who answers first, it's not a big deal. Move on. – user307833 Oct 30 '16 at 16:02

Help prevent new users from asking questions as answers

The number of times new users think SE is like a forum and post a question as an answer is upsetting.

Can a simple parser look for a '?' in their answer and say something like 'If you have a question, please use the new question link!!!'.

In fact, I'd recommend this just if it's their first answer, regardless of the existence of a '?' in the text...

  • 4
    This is pretty well covered by Bhargav Rao's existing answer here. Also see one of Monica Cellio's and HDE's. – jscs Oct 19 '16 at 1:47
  • If that is added it should be site-specific. Consider e.g. Code Review: it should be common for a good answer to include questions to the OP, because a significant part of code review is asking the author to justify decisions whose justification isn't adequately given. – Peter Taylor Oct 19 '16 at 9:54

We need a clear, defined process by which we can request the ability to embed YouTube videos on a SE site

We've been requesting this for eons on Mechanics.SE (questions related to strange sounds, part identification, etc.), written meta.mechanics.SE posts about it, discussed it with Jon Ericson in chat and nothing ever seems to happen.

  • 1
    Not only Youtube embedding but any site-specific feature – Cai Oct 20 '16 at 9:49
  • 4
    But your site is not even big enough to be a rounding error compared to Stackoverflow. – Ian Ringrose Oct 20 '16 at 11:45
  • 3
    @IanRingrose it's about quality, not quantity :) – Zaid Oct 20 '16 at 12:38
  • 1
    I can imagine videos being useful on Arqade as well... but I'd prefer not to see Stack Overflow questions containing videos of code. So this really needs a site-specific system, not just turned on globally. – Jeffrey Bosboom Oct 21 '16 at 2:35
  • FWIW, Arqade already has this, @JeffreyBosboom – Shog9 Oct 21 '16 at 22:05
  • 7
    We're a little bit reluctant to turn this on willy-nilly, based on years of experience with forums where answers that contain videos often contain zero additional information... and thus are useless if the video is taken down for some reason. That said, I do see the value; if y'all are dedicated to policing "video-only posts", this is trivial to enable. – Shog9 Oct 21 '16 at 22:06
  • 4
    @Shog9 That is such a reasonable position. The meta-meta question, though, is: why is there no record of the SE staff response on the feature request? The lack of feedback or transparency means that the feature-request process is broken. – 200_success Oct 21 '16 at 22:21
  • 1
    I guess I overlooked it, @200_success – Shog9 Oct 21 '16 at 22:46
  • 1
    @Shog9 your concern is also applicable to link-only answers, and Mechanics does a fairly good job of policing those. – Zaid Oct 22 '16 at 2:10
  • 2
    And @200_success is on the money; the primary concern I wanted to highlight in this post was the perceived absence of process to request this feature. – Zaid Oct 22 '16 at 2:16

Can the tag suggestions get some love? Specifically, when the system suggest tags it should offer JIT help what the tag is about, probably show the excerpt below the tag.


Improvements to tags:

Allow tags to be marked as “obsolete” along with a message explaining why.

  • An “obsolete” tag will not be allowed to be used on a new question, or added to the question that does not already have the tag.
  • The removal of an obsolete tag from a question would not show the question on the front page.
  • The person that asked the question will get an inbox request to remove the tag from their question.
  • Maybe marking a tag as “obsolete” with a moderator-only power.
  • This would make the process of removing tags less painful.

Allow a tag to require that one of a set of related tag is also used.

  • For example, when SQL is used, we could require that a tag is used to say what data vendor is being used.
  • A regular expression is another example, when it has to know the version of the regular expression engine to write a useful answer.
  • 2
    Which problem does that solve? We already can delete tags without bumping any questions. – Raphael Oct 18 '16 at 15:13
  • @Raphael The problem is when it takes some time to retag all the questions, but while doing so the old tag is still being used on new questions. Common when the old tag is not clear, so the retagging has to choose between a few different options. – Ian Ringrose Oct 18 '16 at 15:16
  • 3
    If there's a canonical retag, SE devs can do it for you (without bumps). Or a synonym may do it. In cases as you mention, wouldn't blacklisting work? – Raphael Oct 18 '16 at 15:18
  • 4
    @Raphael, blacklisting seem to never be allowed in real life for some reason. There are retags on Stackoverlow that takes many weeks due to the number of questions that need cleaning up as part of the process, on other site I expect there is no issue. – Ian Ringrose Oct 18 '16 at 15:21
  • 5
    My problem with most tag cleanups is that they tend to be... Really hit & miss as far as finding actual problems. Folks love to fixate on tags they personally don't see a use for, which tends to cause a lot of collateral damage; meanwhile, synonyms and merging (which are already accessible to privileged users and moderators respectively) are often neglected. See also: meta.stackexchange.com/questions/239189/… – Shog9 Oct 19 '16 at 0:17
  • 1
    Either suitably-privileged users should be able to deprecate a tag (and have the system enforce this), or we should stop adding "DO NOT USE" to tag excerpts and just let people tag as they will. – Jeffrey Bosboom Oct 28 '16 at 2:34

We need to make the migration feature more robust and intelligent for users and moderators

I don't think there is a topic more discussed on SE than this topic. A basic search for questions about migration lie in the 4,000 questions asked range. To me this indicates that the current way of handling migrations is insufficient and should be addressed and not simply rationalized away. Due to the length and amount of arguments in the community I will simply list a few posts for example, and then a proposed solution. I am posting this not so much for my idea to be used over others but that we can actually make a step forward in this area. I feel like the justifications to date are "its too much work", and instead we should be making at a minimum smalls feature steps to test if any of the ideas proposed to date would be beneficial to the community at large.

There are currently some open feature discussions on SO and SE on this Flagging migration should include more options and This question may belong to <Another site>, consider migrating and Propose Close -> Migration, as well as some that were rejected such as Migrate based on tags and More options when flagging for migration, but I don't think that the answers have been satisfactory given that the question keeps arising.

I think that we could potentially solve this issue by having a few criteria work together in concert. I propose using the current migration stats like SE already does, use the question tags to make more relevant suggestions, use the flaggers criteria such as reputation (in both SE sites) to expand their list options as well as if the flagger has recieved tag badges (arguably subject experts) to make this system more robust and intelligent for the community as a whole.

In many of the posts the argument is that community members of one community will not know the scope of the recieving community; or more generically that experts are few and limited and would not be best able to judge the correct action. I think this is a short sighted argument. This means the only option is to go to a moderator. To me this burdens a moderator both in time and in needing to confirm the destination community. In addition a queue would be adding to the slew of already exisiting cleanup the community endures (although such questions already go to a queue with some other flag applied in some cases such as deletion).


  1. We leave the current commonly migrated based list as it is for those in the community below a certain rep level, such as new users.
  2. If you have a tag badge (meaning you are an expert in the topic) and/or sufficient rep level in both the current community as well as the (proposed) destination SE community, then your list includes those sites as options for migration in addition to the ones based on migration stats. This would handle the situation of community scope, because the user knows the scope of both sites due to rep level that they have in both communities.
  3. Over time we will see stat trends in tag groupings and can make the system wieghted to list the most likely migration at the top of the list based on historical stats. This is different than todays stats which is an after the fact aggregation. A great example of this is Android or Wordpress which have full communities on these topics.
  4. After 3 or 5 flags applied(?) to a question by the experts defined in #2, the question gets migrated automatically OR to prevent all out anarchy, it goes into a queue on the recieving site for that community to make the decision to accept or reject the question. Although I would not want another queue, it would arguably remove this decision from the mods and at least allow BOTH communities to benefit in a sense that the FROM community can boot a question and the TO community can benefit and expand its question and knowledge base.

I know this might not cover any technical hurdles with the current system, but once again I think it's important to take one or two of the proposed options and make headway because this is clearly a painful item in the community that isn't going away and just keeps getting rejected based on opinions of difficulty to implement.


Make close votes on own questions binding, even for under-3k users.

A under-3k user can freely edit his question. He can see close and reopen votes on his question. His delete votes on his own question is binding - the question is deleted right away. So why aren't close votes to his own questions binding, insta-closing? I had this problem when I found out that my question is a dupe, but though I (an under-3k user) voted to close as duplicate, the question did not get closed.

  • 3
    For dupes specifically, this isn't actually a problem. Just vote to close, reload the page, and click the This Answered My Question button in the dupe banner at the top of the question, and it will (using the existing binding confirmation mechanism) automatically finish closing the question as a duplicate. It's only other close reasons that are limited this way. – Nathan Tuggy Oct 20 '16 at 7:40
  • 1
    Which this would solve. – Cai Oct 20 '16 at 9:50
  • 4
    Aha, "reload the page", @Nathan, I would never have thought about that... But I guess the banner could be shown right after voting. – Arjan Oct 23 '16 at 12:21

Order answers by most votes instead of accepted answers

A good percentage of answers I end up using are not the accepted answer.

But I see a strong correlation between upvotes for an answer and the its usefulness.

Would be great to see the most upvoted ones on top. In most cases, those are the ones that have been used by others and have helped others, so it makes sense to sort answers this way.

  • 3
    Quite often the accepted answer happened to be the first, or just tells the asker what they wanted to hear instead of facts they needed to, or is simply not as good (I've seen some that fit into two or even all three groups). "Accepted" only means one person in particular thought it was useful, and they already have an upvote to give, too. This is an easy but sensible suggestion. – Nij Oct 23 '16 at 12:27
  • Basically, this is status-bydesign. – Jan Oct 23 '16 at 16:19
  • 1
    Perhaps. But it's all kinds of wrongness. If the question is here for all of us and those who come afterward, as is often claimed and a fuss made over, the best answer as determined by the community will rise up - it doesn't need acceptance for that. And if the best answer is not the one accepted - because that happens often enough now - then the entire idea of an accepted answer is not worth keeping. – Nij Oct 24 '16 at 10:28
  • So I think its a question of "does Stackoverflow want to bubble up useful , but possible less relevant answers. Or does Stackoverflow say, "The guy who asks the question should determine which is the best answer". Also there's a risk of people voting up stuff just to make an answer popular, so not sure how that would play out. – pragman Oct 24 '16 at 16:24
  • How is that risk not already present? 319k questions on Math SE do not have an accepted answer, compared to 359k that do. People can upvote in order to move it higher on the page already - and that's actually the entire point of upvoting and answer ordering. If there was suspicion of someone doing it just to have their answer higher up, there's algorithm to catch sockpuppetry and to catch abnormal voting, and moderation to stop people being egotistical halfwits. – Nij Oct 25 '16 at 10:06
  • @Jan meta.stackexchange.com/a/285938/279306 Kinda plays into this one then, doesn't it? – Sidney Oct 25 '16 at 18:21
  • 4
    I often see mediocre first answers gather a few votes and then see a better answer being posted and getting accepted (with fewer votes). Thus I would suggest something like 'order by votes, where the accept mark counts for 3~5 votes'. -- A less intuitive alternative could be to only look at votes after an answer has most recently been accepted. – Dennis Jaheruddin Oct 26 '16 at 15:16

Not the answer you're looking for? Browse other questions tagged .