251

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).

40
  • 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, 2016 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). Oct 17, 2016 at 17:03
  • 12
    +1 for better duplicate handling and for improvements with flagging. Oct 17, 2016 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, 2016 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, 2016 at 18:48
  • 6
    Did anything come of the post characteristics as images tests?
    – Andy
    Oct 17, 2016 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, 2016 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, 2016 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, 2016 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, 2016 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, 2016 at 7:31
  • 8
    When will we hear about the outcome of this initiative/?
    – Raphael
    Feb 5, 2017 at 19:29
  • 23
    @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. Jul 5, 2017 at 6:16
  • 13
    And another full year went by. Do you by your comment "nobody wasted time", Tim?
    – Raphael
    Jul 15, 2018 at 14:32
  • 6
    Any update on this @TimPost? And I actually mean any.
    – Script47
    Jul 24, 2019 at 12:37

81 Answers 81

489
+150

New users need better awareness of what is expected of them when they ask their first question.

This is the single most important thing that needs to happen on the Stack Exchange network, in my opinion. If the new Question Ask page will facilitate that, then I am enthusiastically supportive. However, there are historical barriers, some of which SE itself has erected, and some which are due to conflicting goals.

Currently, we rely on a system of Help Center articles that new users never see, a patchwork quilt of Meta articles, and a Tour Page that focuses on site mechanics. We carefully bury any information that might help new users shed their bad forum habits, allowing them instead to crash headlong into a "hostile" user community that expects new users to already know our rules, and we call that being inclusive.

Good software systems always follow the Principle of Least Surprise. If your system's behavior regularly surprises, then it fails its purpose.

It is unfortunate that online behavior has been co-opted by the Facebooks, Twitters and YouTubes of the world, but if we are genuinely interested in making the world (or at least the internet) a better place, then we must also believe that we have (at least in some measure) a responsibility to educate people on how good communication really works.

This isn't about being negative. It is about being proactive. Nor is it about providing better flagging tools, although we should do that as well. Rather, it is a mindset. The best off-topic question is the one that is never asked. Allowing sites to run off the rails by providing the lowest possible friction to ask off-topic questions and then saddling the community with cleaning up those questions... well, that's not what most of us signed up for, nor is it the best use of our time. It is probably the least-friendly way to welcome new users.

Most of the professionals who are here come here to contribute and to help others, not to spend all of their time sweeping the floor.

26
  • 51
    Completely agreed: the best way to "improve the experience of our existing users" and moderators is not filtering the view after the fact. It's catching (and ejecting) bad stuff before it even makes it through the pipeline. That said, it seems that the two bullets under "improving the new user experience" actually are aimed at this aspect, and I'm heartened to see that.
    – jscs
    Oct 17, 2016 at 18:46
  • 3
    @TravisJ: I have lots. You can browse my activity on Meta.Programmers.SE in the new-site-name tag if you want the sordid details. tl;dr: meta.programmers.stackexchange.com/questions/8254/…
    – user102937
    Oct 17, 2016 at 20:02
  • 8
    @TravisJ: I think I'll wait and see what the new Ask Question page looks like. I don't view this as a technical problem; I view it as a philosophical one: how can we tell people what they need to know without being negative? I am claiming that this assertion is a non-sequitur, and that it needs to be worked out before we even begin discussing a technical solution.
    – user102937
    Oct 17, 2016 at 20:16
  • 3
    A possibility here - meta.stackexchange.com/q/278444/193412 some GH repos have a similar thing for filing a bug report and I suggest that the average dev filing a bug report is a lot more experienced than the average SO asker.
    – enderland
    Oct 17, 2016 at 20:22
  • 39
    @TravisJ: I have seen some non-SE sites partially solve the problem like this: "Before you ask your first question here, you should read the rules. You don't have to read the rules, but you will be treated by the user community as if you have read them. So read them."
    – user102937
    Oct 17, 2016 at 21:05
  • 20
    And related to this, can we stop caring how people who come here, don't read the rules, and drop off topic posts feel about things? If a question is off topic, poor quality, etc, just get rid of it - I will downvote and vote to close. I'm not a mean person. I will just devote minimal time to someone who can't be bothered to help themselves. The reason we have this "SO is so mean!" is because people have this attitude but the guidance Robert is suggesting is poor - for fun, open an incognito tab and ask a question on SO, observing what you have to read in order to post.
    – enderland
    Oct 18, 2016 at 0:00
  • 5
    @TravisJ: Since you did ask, this is the proposal I floated specifically for Programmers. It didn't get much traction, so I narrowed it down to the single "what we're not" paragraph. Eventually, I got it down to a single phrase of 4 words in the tagline of the site: "but not code troubleshooting." I thought I got agreement from SE that this phrase would be adopted, but it didn't make the final cut.
    – user102937
    Oct 18, 2016 at 1:48
  • 6
    I've found I magically acquire powers normally reserved for 20k users if I politely point first time posters to the how-to-ask, on-topic pages as well as a breakdown of what's on topic on other sites. If I can delete questions faster than we can close them by asking politely doesn't it suggest that many users are simply lost? Yeah, we need to educate new users better. +1. Oct 18, 2016 at 4:23
  • 4
    Why isn't there a direct link to the "what topics can I ask about?" and "what is considered off-topic?" help pages on the "Ask Question" page?
    – tonysdg
    Oct 18, 2016 at 4:28
  • 53
    Speaking as someone who did read the Help Center before asking their first question...it's still useless. I got plenty of information about how to interact with the SE software, but nothing about how to interact with the community, and very little about what sorts of things were on-topic.
    – Mark
    Oct 18, 2016 at 8:32
  • 14
    +1 for help center articles that new users never see. SOOOOooo true.
    – Cullub
    Oct 18, 2016 at 21:20
  • 7
    If someone has a '?' in an answer field, a big-ass warning should be 'do you realise you're asking a question in response to a question?'. The number of answers we have to delete because people post questions there is sad :(
    – Mark Mayo
    Oct 19, 2016 at 1:38
  • 10
    Robert for president :) It boils down to whether we want quality over quantity. If you want quality questions, you need quality users, or you need to turn bad quality users into good quality users, which requires 'education', as you suggest.
    – Benjol
    Oct 19, 2016 at 11:48
  • 3
    @MarkMayo Just don't make it a hard rule. There are valid reasons to include question marks or even questions in answers, not even limited to code blocks. A warning is probably a good idea, but an outright ban would be a horrible idea.
    – user
    Oct 20, 2016 at 11:37
  • 5
    I'll be announcing some tests within the next 3 or so weeks, as well as a survey that's shown to brand new users after they ask their first question. tl;dr; - we think we know how to optimize the ask page to get good results, but we want to confirm that we're correct here - or a really good idea might get shelved due to a series of poorly constructed tests.
    – user50049
    Oct 25, 2016 at 19:04
161

Implement a tool for editing tags without bumping

Mass tag edits are disruptive to both new and established users, especially on small- and medium-size sites. There's no good way to edit all the questions of a nontrivial tag without thoroughly burying new posts on the front page.

Tag edits without any review are understandably problematic, but other options are available – one possible tool is described in Tool for peer-reviewed no-bump mass retagging.

10
  • I believe at 10k the tag-only edit privilege already does not bump the question... I could be wrong.
    – TylerH
    Oct 17, 2016 at 20:09
  • 28
    @TylerH Tag edits by 10k users, and even moderators, always bump the question. See Disabling bumping for retags by hi-rep users (a declined feature request). Oct 17, 2016 at 20:12
  • I'm not sure that covers the case of inline tag edits which could well be a newer privilege than that 5-year-old MSE request.
    – TylerH
    Oct 17, 2016 at 20:30
  • 12
    @TylerH, when I do inline tag edits as a moderator, the question is always bumped. As far as I know, the bump is only absent when tags are merged or when a tag is automatically deleted due to low popularity. Oct 17, 2016 at 20:47
  • 31
    This should be a moderator- or community team-only feature, and it should be limited to mass retaggings that have been discussed and agreed upon on the site's meta. For regular users, and in all other cases, editing should bump the question for audit purposes. Oct 18, 2016 at 4:05
  • @CodyGray What auditing purpose is achieved by bumping that isn't already achieved when a person edits it and leaves behind a new revision?
    – TylerH
    Oct 19, 2016 at 1:27
  • 4
    @CodyGray - +1 to this. I personally think that site mods should have a button they can press to turn off bumping for the duration of a mass tag edit. This button should have some kind of audit trail to show a) When it's been used and b) what it's been used for.
    – Richard
    Oct 20, 2016 at 17:00
  • 2
    Please, can we have this feature. Recently on ELL there has been something like ninety, (maybe more?) tag edits which dominated over four active pages. And the majority of these necessary and legitimate tag edits were performed by a mod. What's the big deal, when the newest page is not affected? A huge one on smaller sites which receives a new question hourly or every two/three hours, and when someone posts a new answer on an old question, it gets buried on the active page. Oct 21, 2016 at 5:24
  • Change "editing tags" to "minor edits (including tag only edits)", and I'll give you a +1. Seriously, the bumping has been used as an argument for rejecting small improvements to questions.
    – jpmc26
    Oct 22, 2016 at 20:20
  • 3
    I think that completely hiding tag edits is bad because of the whole principle of being transparent, etc. Perhaps there could be a "mass tag edits" heading in the sidebar where moderators can mass edit tags? Or the sidebar on meta? Oct 23, 2016 at 22:49
141
+50

Two more minor requests:

Allow unregistered accounts to delete their own posts: this one should be fairly obvious. This occurs a lot when some unregistered user finds out that they can't delete their own post, and decides to just vandalize it. This is a restriction for no good reason ("Cookie-based accounts don't support voting, therefore they don't support deletion" doesn't help an unregistered user remove their post), so I'd support removing it.


What is the reputation required to edit one's own deleted question?: That's not a feature request, but it details the problem:

One of my friends who is new to the SE recently posted a question on SO and then quickly deleted it after I pointed out some issues.

However, he finds that he cannot edit the question while it is deleted.

Even high rep, experienced Stack Overflow users believe that this is possible. However, it is quite obviously not. I understand not wanting hidden spam, but bumping on undeletion would solve this problem. Further, I doubt that many spammers would even think of such a complex way to spam Stack Exchange, but many legitimate users want to fix their post if it was accidentally posted or a huge flaw was found. Allowing (at least some) users to edit their self-deleted post would be helpful.

2
  • 8
    Both are good suggestions. Thank you!
    – user50049
    Oct 17, 2016 at 18:16
  • 14
    This occurs a lot when some unregistered user finds out that they can't delete their own post and wastes 3 golden precious 20k delete votes in turn. stackoverflow.com/a/39943048/4099593 Really need this feature, Please, Please. Oct 17, 2016 at 20:12
129

Better automatic guidance when questions are closed

Thousands of questions get closed every day over the network, and newer users are more likely to get their questions closed. Yet little guidance is given to new users, so to be helpful, close voters often resort to canned comments! The information in these canned comments should really be provided automatically. There are several things that would improve user experience a lot and are technically easy to do.

Notify askers when their question is closed

If somebody comments on a question, the asker is notified. If somebody answers, the asker is notified. If somebody edits, the asker is usually notified. But if the question is closed, the asker isn't notified…

Helpful close voters may leave a comment, solely to notify the asker. Moderators on beta sites even get a nagging flag for that (“question closed without comments”). That's either busy work for closers or leaving the asker in the dark.

Provide better guidance about closure in general

The notice under a closed question reads:

closed as reason summary by close voters date of closure

detailed close reason

If this question can be reworded to fit the rules in the help center, please edit your question.

That's not too bad, but it can be improved. The guidance suggests editing, but it doesn't explain what happens next. There are many small features around this which I haven't taken the time to collect, including:

Oh, and get rid of the term “on hold”, which nobody understands. There is no need for a different word to mean “closed less than 5 days ago”.

Reason-specific closure guidance

“If this question can be reworded to fit the rules” is so generic that it isn't really helpful. The guidance would probably depend on the type of closure — closing as unclear usually requires filling in missing information, for instance.

There should be more community-edited close reason, including not just a reason but also close-reason-specific guidance. And of course those shouldn't be shoved under off-topic, which is an endless source of confusion (“it's a programming question, why are you saying it's off-topic” — “Because you haven't provided the full code.” “What does this have to do with being on-topic? Is Javascript off-topic?” …)

Better heuristics to enter reopen review

A question reenters reopen review if it is edited within 5 days of being closed, but only after the first edit.

I'm sure there's room for improvement, even though I don't have specifics to offer. It's very common that someone retags a question after closure, for example, and that doesn't make the question worthy of reopening; but when the asker comes back and adds missing information, the question has already accumulated “leave closed” votes and it's too late for the reopen queue to do its job. Edits by the asker should be more likely to trigger reopening, but edits made by someone who adds information that the asker provided in a comment should also count somehow.

Make migrations sensible

Instead of making migration a reason to close, make migration what happens if a question gets closed and there is another suitable site. “It belongs on site Y” is not a reason to close a question posted on site X if it also belongs on site X. On the other hand, if site X has closed a question then denizens of site Y should be the ones deciding whether to accept the question on Y, not denizens of site X.

8
  • 3
  • Very different items here, but I agree with all of them.
    – Raphael
    Oct 18, 2016 at 15:02
  • Sadly, the % of cases where the asker edits after closure are not high. And the % of these which are reopened tend to be far lower still. For this reason, I like Wrzlprmft's idea: side-step the whole cycle if at all possible.
    – Shog9
    Oct 19, 2016 at 3:15
  • 8
    @Shog9 How often the asker edits after closure isn't really relevant. The reopening procedures aren't for the unsalvageable crap that gets closed, they're for the salvageable questions that needed a second iteration from the asker to get to a workable state. The important thing is how often those edits should lead to reopening, and how often the edits do lead to reopening if they should. Oct 19, 2016 at 7:06
  • 2
    I disagree strongly about the use of "on hold". It's vitally important not to discourage new users. Too often I've seen a cycle like: New user asks question which is interesting, but technically violates a rule of the se in question --> Question closed without discussion --> new user leaves and is never seen again. "On hold" conveys the need to edit and tidy up, without telling the new user they've been shut down.
    – Tynam
    Oct 21, 2016 at 9:28
  • 3
    @Tynam How on earth does “on hold” convey the need to edit? “Awaiting edits” or “closed pending edits” would convey the need to edit. But “on hold”? That just says “wait for someone to deal with it”... Besides, I've rarely if ever seen an interesting question that gets closed without any discussion. Crap questions, sure, but if the question is “do my homework kthxbye” then I don't mind if the user leaves. Oct 21, 2016 at 20:02
  • @Gilles: I'm not saying "on hold" is the best possible terminology; I agree with you. I'm just saying "Closed" is terrible. And if you haven't seen an interesting question closed - maybe you're not on the same stack as I; I have. I raised this precisely because I just spent time following up with a new user who asked something interesting and useful (but against the local stack rules in a non-obvious way) who'd had their question closed without comment. They won't be back, and it's our loss.
    – Tynam
    Oct 22, 2016 at 8:57
  • @Tynam I've seen interesting questions get closed, sure. But rarely without any comments. Oct 22, 2016 at 10:14
125

New users need better awareness of what is expected of them when they answer their first question, especially when it is old.

TL;DR We are speaking about the NAA (Not an Answer) Problem

Robert explained about the issues created by uninformed users while asking their first question, so let us also extend that to answers.

Background

The New Answers to Old questions shows all the answers on questions that have been asked at least 30 days earlier. This is a hot bed of NAAs (Not-an-answer's), as well as VLQs (Very Low Quality posts). Most of the bad answers there follow a general pattern, and hence, we decided to write a small application that tries to figure out those types of answers and reports it to us after filtering a lot of data. Some interesting revelations were

  • Almost all the posts that were small and ended with a question mark were NAAs!
  • Almost all the posts that had wordings like "I have the same problem", etc., were NAAs!
  • ALL the posts that had a combination of more than one of these factors were NAAs!

There are more than 200 reports in a day on Stack Overflow, and about 100 are reviewed by us and nearly 90% of them are NAAs (or VLQs). We can see the results of the individual filters here: (tps are NAAs and fps are falsely detected posts)

An image of the analyzed results - 1

An image of the analyzed results - 2

The dataset of all the posts with feedback are here.

What makes people write a question in place of an answer?

Getting to the root of the issue, is what we need to do. While answering a question a new user sees this:

Answer Pop Up

The answer pop up links to How do I write a good answer?, but, that page never mentions about not writing a bad answer. That's the main issue that we have at our hand.

Requests to the SE Team

Pre Answering: Make the user know what a BAD ANSWER is!

  • Split the trio, i.e., help, clarification and responding, into three different bullet points (and provide examples). Lists are more helpful than writing them in a single line.
  • Prevent the users from stating that I don't have rep, I answer by adding a link to Why is the 50 rep limit present?
  • Inform the users about link only answers and their ill effects and Why should I not add a link only answer?
  • Scare them by speaking about answer bans

Post Answer Writing: Make the user know that they did not read the rulez.

  • Detect the answers as NAAs by using simple techniques like, Post ending with question mark, etc. Add a warning label to those answers just like the pop up we get when a code is badly formatted and inform the user that their post might be a NAA
  • Prevent the users from entering gibberish and non-English posts. Add an error label that mentions them that their post is not accurate for the site.

Post Answer Submitting: Sigh, can't help, but help the reviewers.

  • Reduce the number of reviews needed to complete an LQP Queue review from 6.
  • Have binding flags, like the close votes. 3 flags soft delete a post.
  • Add a "move to comment" privilege, with 3 votes needed (similar to delete votes). Let the high rep users also decide.
  • Educate the 3k+ to review.

Another personal request is to Enhance the "New Answers to Old Questions" moderator tool. It's really hard to use that at the state it is in, now.

We can be sure that the users will continue to post NAAs even after these are implemented, but these should help us reduce the number of NAAs by a bit at least.

15
  • 5
    "Detect the answers as NAAs by using simple techniques like, Post ending with question mark, etc." Actually this is status-almost-completed. All we have to do is to integrate NATOBot to the system as well as SmokeDetector!
    – SE is dead
    Oct 17, 2016 at 22:18
  • 3
    @dorukayhan While NAAs are being detected with a pretty good accuracy, The bot has a lot of fps with VLQs. Hence I feel that it's still a bit premature. Smokey on the other hand, has been tested a lot. Oct 18, 2016 at 7:54
  • 3
    This problem won't go away for good until comments cease to be treated as second-class and ephemeral, but one thing that would really help is a way to explicitly ask the OP a complicated question, one that won't fit into the comment box. "In order to answer this question I need you to run the following test program and report its output: [code...]" is the prototypical case, but there are others.
    – zwol
    Oct 18, 2016 at 14:06
  • If we make someone rewrite their “not an answer” before they think it has been posted, it will become harder to find them. Oct 18, 2016 at 14:14
  • Maybe make ever use that has had an answer deleted as "not an answer" complete and pass a test before they can post anther answer. Oct 18, 2016 at 14:14
  • @Ian, For your first comment, The answers will still pass through the first post (and/or late answer) review queues. So, those posts might still be discovered. Another way might be to alter the Warning label in such a way that it informs the users that they are asking a question and not an answer. For your second comment, Yes, That's a better idea. Oct 18, 2016 at 14:19
  • @zwol Yeah, I hate those types of NAAs. Those which are too long to fit into a comment. (Like these). But, in my experience, They're usually good enough to be another question or bad enough to be deleted with no loss of useful material. There has been no in between. Thanks anyway, I'll be monitoring the late answers 10k tools and try to find out such posts. Oct 18, 2016 at 14:32
  • 1
    @zwol we could mitigate a lot of that problem by allowing some basic formatting in comments and increasing the character limit (or not counting code blocks toward the character limit)
    – TylerH
    Oct 19, 2016 at 1:36
  • 3
    "What makes people write a question in place of an answer?" -- attempt to circumvent a question ban might be a popular reason for that. As far as I know system keeps an (internal) record of users bumping into the block - that makes it possible to, say, trigger a warning if it detects attempt to answer after a recent block. "Wait! You've recently been blocked to ask questions. Consider that if you are posting question into the answer it will be deleted. Please select the check box 'I understand that and I am going to post an answer, not a question' in order to proceed answering"
    – gnat
    Oct 19, 2016 at 8:18
  • 2
    Interesting point @gnat, I remember one such user who had asked a question in place of an answer. They later confessed that they were question banned. After that answer, they were answer banned too. (Perhaps). But I haven't seen that as a regular occurrence. Do you have any idea as to how to monitor those users? (as a non-mod). If so, we can certainly find out if it occurs more often and add that as a new request to Tim. Oct 19, 2016 at 8:58
  • @BhargavRao non mods can't reliably monitor that and as far as I know even mods can't, bumps into blocks are only kept internally in the system and one needs dev access to see. Regular users can only guess by checking questions in the user profile and mods can only make a bit better informed guess because they can additionally see the history of user deleted questions. If you want to learn more then probably the only way is to request SE team for stats on how many NAAs are posted by recently blocked askers (FWIW there was similar request at Programmers meta and they gave stats on it)...
    – gnat
    Oct 19, 2016 at 9:04
  • ...you can use that stats request as example if you want: How many questions do we get from users recently blocked at SO, how many of these are closed / deleted? (I am not sure if MSE would be the right place for it though, probably MSO is a better fit)
    – gnat
    Oct 19, 2016 at 9:07
  • Mabe a first-class, voted and scored follow-on question, with full answer-box editing. Could look like a comment but be blown open as needed. Answers could be moderator-demoted to such a thing.
    – Clay
    Oct 23, 2016 at 12:38
  • "move to comment" is quite an elegant idea to solve part of this problem. It allows moderation to prevent poor comments being posted as well as educates the new user a bit. It has some potential for abuse, but no more the current system at first glance.
    – user343082
    Oct 30, 2016 at 14:44
  • 1
    @DoritoStyle Yep. ATM moving to comments is one way only. So there needs to be some way to reverse this to prevent users from abusing this feature. But unfortunately, Jon doesn't like increased comment moderation. Oct 30, 2016 at 14:50
116

I'm recommending Categories other than "off-topic" should allow custom close reasons be considered for this project. See Nathan Tuggy's excellent answer there to see how widely useful this would be to a variety of sites.

This is an issue for new users who have their questions closed and what they see first is "off topic" which is confusing to them because they can't understand how their question is off topic (which it's not). This often requires a knowing user to explain that the question isn't actually off topic, it's ______. If they think the question is off topic, they may be less inclined to fix it, while if the main close header says "unclear what you're asking" instead, they may be more interested in improving their question.

It's also an issue for new close voters/flaggers who may not be aware that these close reasons exist or have difficulty finding them because they don't fit the "off topic" category.

6
  • 7
    Yes, the custom reason should be at the top level, not buried under the Off-Topic reason, especially since you can modify the text to say whatever you want anyway.
    – TylerH
    Oct 17, 2016 at 20:11
  • 3
    Agree with Catija. Disagree with TylerH. I think giving a groupbyable primary reason to which you can add your custom text will keep whatever stat-keeping is available to the moderators. Oct 17, 2016 at 21:11
  • 8
  • @Michael I'm fine with a custom reason in every category so long as there's also one at the root level, where it is most useful for the purpose of closing questions.
    – TylerH
    Oct 19, 2016 at 1:31
  • Using a custom reasons does increase the possible of a revenge down vote as it shows clearly who is "to blame". Oct 31, 2016 at 14:54
  • 1
    @IanRingrose not sure what you mean. This is not about the "other" box that creates a comment, it is about the site-specific close reasons. They work just like all of the other close reasons.
    – Catija
    Oct 31, 2016 at 14:57
85

A minor suggestion, but we can help make new askers more likely to accept an answer by pointing them to the "accept" button when they try to upvote and don't have sufficient reputation.

7
  • 1
    Care to explain the anonymous downvote? What Tim Post said is "we're going to put the highest priority on small-ish things that could potentially make a big impact first" and he also emphasized improving the new user experience. This would align perfectly with both of those.
    – Wildcard
    Oct 17, 2016 at 20:35
  • @hichris123, not the same thing. That only works if the asker can upvote, and it reminds them that they can also accept an answer. It doesn't educate new users who can accept an answer but cannot upvote, which is even more important IMO.
    – Wildcard
    Oct 17, 2016 at 21:01
  • Oh, oops. I didn't read enough I guess. Does this actually not happen? I'm surprised, seems fairly obvious.
    – hichris123
    Oct 17, 2016 at 21:02
  • 39
    @hichris123, I know it's surprising, but you can judge from the number of comment conversations that go along these lines: "Thanks, it worked! But I can't click the 'up' arrow on your answer for some reason." "You can only vote after you get enough reputation, but if my answer solved your question you can accept it by clicking the checkmark." "What checkmark? I only see voting buttons." "It should be just to the left of the answer, below the voting buttons." "Thanks, I see it now. Accepted!" :)
    – Wildcard
    Oct 17, 2016 at 21:05
  • 4
    A big floating flashing arrow, scrolling text saying "award best answer"; +1 Oct 18, 2016 at 11:36
  • @axsvl77 Accepted answer is not necessarily the best. Also 'award' is not exactly accurate either (in common parlance, at least).
    – TylerH
    Oct 19, 2016 at 14:46
  • 1
    Ha! You're right; in this instance, I meant award = designate. Not regular usage. Oct 19, 2016 at 14:53
80

On some sites, there are simple heuristics to detect problematic questions. By detecting a few problematic keywords, we can easily alert posters to potential problems before the question is even submitted.

For example, Code Review has been requesting these automatic alerts for over half a year:

The last two, in particular, would also benefit Stack Overflow.

As another example, English.SE considers the use of monospace text for emphasis to be a faux pas that should be edited out. We have the technology to detect backticks and advise the poster to use italics instead, right? That would be preferable to bastardizing Markdown.

By giving automated instantaneous feedback, we should be able to improve question quality and increase user satisfaction. That would be better than review queues, comments, waiting for the original poster to respond, downvotes, and close votes.

10
  • I'll say that some of this is most likely going to be covered in some of our /ask page redesign, etc. We're looking at a variety of things that we can do to encourage users to include everything they should to make the question better.
    – Taryn
    Oct 17, 2016 at 21:07
  • 32
    @bluefeet that's great, but #PeopleDontRead. The idea is to pop a red in-your-face warning that you can't miss, e.g. when your question title says "Why isn't this code working?" it's 100% guaranteed your question is off-topic. Having yet another page to link to after we've closed their question isn't going to help, if they can post such a question without the system telling them "WAIT! Your question is likely OFF-TOPIC!", then it's too late already. 60% of CR's closures involve broken code, if pointing them to [help/on-topic] isn't enough, I don't see how [ask] is going to change it. Oct 17, 2016 at 21:17
  • 9
    @bluefeet It'd be important for the triggers in this functionality to be configurable on a per-site basis, so that the different communities can define different standards for acceptable questions.
    – Thriggle
    Oct 17, 2016 at 21:19
  • As you said @Meta'sMug People Don't Read, and it's entirely possible they would not read a giant red box yelling at them either, but as I mentioned we're looking at a variety of ways to offer better guidance to users when they ask their question.
    – Taryn
    Oct 17, 2016 at 21:28
  • 17
    @bluefeet they might not read it, but then at least their first clue for being off-topic wouldn't be a close vote, a downvote, and/or a friendly canned comment kindly explaining why their post is off-topic and linking them to the help center (or all of the above): their first clue would be a very short, obvious notification they'd have to actively ignore. Basically it boils down to Robert Harvey's answer. Oct 17, 2016 at 21:32
  • 3
    If people don't read then delete their question.
    – Miles Rout
    Oct 18, 2016 at 7:23
  • 6
    @MilesRout Wherever possible, the goal should be nudge the marginal questions into being respectable, or to help people post their question to the right site. "Then delete their question" is like saying "spam isn't that bad — is it really that hard to just press Delete?" And, to many new users, reading the Help Center pages is only slightly less onerous than reading the Terms of Service. We can make things better for everyone — we have the technology. Oct 18, 2016 at 7:40
  • @Meta'sMug Not to go down a rabbit hole, but that topic title is not at all a guarantee that the question is off-topic.
    – TylerH
    Oct 19, 2016 at 1:29
  • 5
    @TylerH On CodeReview, I'd take 99:1 odds on any question with that title being Off-Topic.
    – Kaz
    Oct 20, 2016 at 9:19
  • For a while now I've been wanting to build a model for predicting self-study questions and off-topic programming questions on Cross Validated. It would greatly cut down on review load during midterms and finals season. Oct 27, 2016 at 23:47
76

Improve communication between flaggers and moderators, or, "what's obsolete about that?"

Some sites are very chatty. And that means flags on comments. My biggest moderation time sink is comment flags, and, within that, obsolete flags. I don't know why it's obsolete -- was the post edited? Was the comment it replied to removed? Or is it not obsolete, but the guy arguing with that other guy is trying to remove comments he doesn't like?

Can we add more information to "obsolete" flags? requests a little bit of tooling that would help flaggers tell moderators what the problem is, so we can handle the flags efficiently. We all have better things to do than to read through 20 comments and 6 revisions to figure out what the history is.

On one of my sites we've asked our users to mostly stop using the obsolete flag and use custom flags instead, but this only works for people who've seen the meta post and remember. Also, it means we're giving the opposite advice for comments ("please use custom flags if you have any doubts") as for posts ("please don't use custom flags if the built-in ones apply"). That's confusing for users who aren't SE fanatics keeping up on every detail ("wait, am I supposed to use 'other' or avoid it here?").

8
  • 32
    Whenever I mark a comment as obsolete I always wonder if the moderator knows what I am talking about. It would be nice to leave a simple comment.
    – Suragch
    Oct 18, 2016 at 7:11
  • 22
    Alternatively, let 10k users handle obsolete/chatty comment flags to lighten the moderator load. I have read some discussions here and there, but have not taken the time yet to really dive into this and write a feature request.
    – Jan Doggen
    Oct 18, 2016 at 7:26
  • 15
    Also let me mark many comments on the same post as obsolete along with providing the additional information to the mods in one package, instead of lots of unrelated flags. Oct 18, 2016 at 13:55
  • @IanRingrose if you really need to flag multiple comments on one post, the recommendation is to flag the post with a custom flag explaining why you think the comments need to be cleared.
    – Catija
    Oct 18, 2016 at 22:18
  • 4
    @Catija that's the recommendation, but it's still a hassle for either the flagger (collecting links to specific comments to bundle together) or the mod (to dig through it). Some sort of "these comments go together and should have the same disposition" affordance could be helpful. Oct 18, 2016 at 22:29
  • 2
    I’ve drastically cut down using obsolete flags for precisely this reason. Most of the time, I flag for mod attention stating obsolete - followed by a reason.
    – Jan
    Oct 19, 2016 at 11:51
  • 1
    @JanDoggen: Alternatively, give the comment’s author a chance to decide about the flag first.
    – Wrzlprmft
    Dec 26, 2016 at 18:57
  • @JanDoggen true. I love it when users can sort out their own obsolete threads without involving mods. Dec 26, 2016 at 19:09
65

One of the minor-ish things that I would like, and have been meaning to check on the feasibility of would be to add custom filtering for certain sites in the "Hot Topics" bar down the side.

I don't really care at all about certain network sites, but often see a lot of their stuff show up. For example, I routinely have 3 or 4 from puzzling.stackexchange.com in my sidebar, no offense to those guys. However, every now and then I see stuff from other things like DIY, which I do find interesting, but I'm sure they're more often drowned in the noise from other sites.

Perhaps other people find this useful too. If not, no worries!

16
  • 4
    ahem Esperanto. ahem physics/math. ahem worldbuilding +1
    – Jeutnarg
    Oct 17, 2016 at 19:42
  • 2
    @Jeutnarg physics/math is one of those that I occasionally do check out when the title sounds very interesting. Therefore, I made my request a little more clear that we should have the ability individually to say "I never want to see these 20 sites", or something like that.
    – krillgar
    Oct 17, 2016 at 19:45
  • I already read it that way, but I can see why they wouldn't have. FWIW, I just listed mine, not everybody else's. I'm sure that scifi and rpg exchange would be on some people's lists, but I like those.
    – Jeutnarg
    Oct 17, 2016 at 19:48
  • 8
    @Jeutnarg Worldbuilding is actually one of the niche sites with better content out there... at least it started out that way. Now... puzzles, codegolf, lifehacks... I wouldn't mind seeing those sites del--er... ignorable.
    – TylerH
    Oct 17, 2016 at 20:03
  • 16
    I find Worldbuilding super interesting, but I tend to be on SO at work... it's more of a 'I like you, but I can't care right now' filter. If this was a thing, I'd probably make a script to change my preferences during work hours.
    – Jeutnarg
    Oct 17, 2016 at 20:05
  • That feature is currently available in SOX.
    – fixer1234
    Oct 17, 2016 at 21:04
  • 1
    @fixer1234 Thanks for the heads up. However, it would get tedious installing that on every computer I would use to access the network. However, that could be used as a template for how to implement it making the SE Devs' jobs easier. I'll probably install it on a few as well.
    – krillgar
    Oct 17, 2016 at 21:24
  • 3
    it would be great if ignored tags were honored for hot network questions too (applicable when one is a member of the network site)
    – Keith Hall
    Oct 18, 2016 at 6:51
  • 1
    the only reliable way to achieve what you ask for is to Increase the Hot Network Questions randomness. This is because hot questions currently stick for too long and that annoys those who want more diverse list to pick the sites they prefer. But this is unlikely to happen, two or three months ago SE team folks told that they are perfectly happy with the way how hot questions work now
    – gnat
    Oct 18, 2016 at 11:57
  • 3
    @gnat Then it sounds like SE could use a reminder that some users aren't.
    – krillgar
    Oct 18, 2016 at 12:15
  • 2
    they seem to ignore complaints about that. The only thing that worried them was meta noise when Stack Overflow questions stuck in hot list for too long ("croissants") but after they figured a trick to quickly drop them off the list they don't worry anymore
    – gnat
    Oct 18, 2016 at 12:25
  • 5
    That would be an amazing feature, but was declined in the past. I truly don't understand why the hell they refuse to turn a spam-like advertising of other sites into targeted advertising. It would be a win-win situation; we are served content we like, they promote content we are willing to consume.
    – user
    Oct 18, 2016 at 17:24
  • 3
  • @gnat While I agree with your historical analysis, Q&As like this seem to point directly at the opposite conclusion.
    – TylerH
    Oct 19, 2016 at 1:32
  • How will this improve the quality of questions and answers? Oct 31, 2016 at 14:56
58

Give new users better advice on how to write question titles.

There are a lot of terrible question titles out there - and I'm not just talking about ones along the lines of "Plz send me teh codez for my C++ porgam". Many ignore the advice given in How do I write a good title?, and are

  • Not specific enough
  • Grammatically incorrect to the extreme
  • Not specific at all
  • Not concise enough
  • Not specific in the slightest
  • Actually part of the body, rather than a succinct summary
  • Not specific

This is a problem on a lot of Stack Exchange sites, from what I've seen, and I think that helping new users could solve it. I propose linking to the above question somewhere near the input box for the question title in the first few questions from a new user, which can hopefully make titles, at the least, a bit better.

7
  • 3
    With the option of linking to a site-specific how-to, maybe!
    – Raphael
    Oct 18, 2016 at 9:21
  • 13
    (1) I’m bothered when I see a question where the author composes a multi-sentence interrogative, and then puts the first sentence into the title box and the rest into the body — rather than putting the entire text into the body, and then adding a title (summary).  I expect question bodies to be self-contained, so I find these hard to read.  Am I alone in this?  Would you add it to your list? (2) Do we have any automated tools for evaluating titles for these criteria?  Something like Microsoft Word’s grammar checker would be better than nothing. Oct 18, 2016 at 21:36
  • 6
    @G-Man Honestly, the easiest way to address that issue--and something that I'd actually appreciate even as an experienced asker--would be to put the title field below the question-body field. It's fairly well known that the title is often the last thing you should write when writing an essay, journal paper, novel, etc; question-asking seems to work the same way, I think. (And if we had the machine-learning capabilities to do so, it would be awesome to auto-generate a proposed question title based on the content, the way we auto-suggest tags...) Oct 19, 2016 at 17:49
  • 2
    @G-Man I like your point (1). A related pet peeve of mine is when the asker just copies the single-sentence body into the title, or vice versa.
    – HDE 226868
    Oct 19, 2016 at 17:51
  • 1
    @Kyle, I think the reason the title field is on top is so the "Questions that may already have your answer" box that pops up has something to work from. I agree with you though.
    – user299767
    Oct 20, 2016 at 0:44
  • 1
    Some regulars could also use something like this. I have been known to downvote questions even from reasonably high-rep users simply because the question title is not a meaningful summary of the question, particularly if edits to try to improve are rejected or rolled back (as opposed to being used as a starting point for the OP to adjust further). Yep, guilty as charged!
    – user
    Oct 20, 2016 at 11:45
  • Some veteran users could use a question-phrasing workshop too. +1.
    – einpoklum
    Dec 9, 2017 at 11:14
57

Spam

We still get too much of it. The built-in filters do an awesome job keeping a lot of it out of our faces, but they're not catching everything.

SmokeDetector is reporting spam posts for people to flag, which is generating about 100 flags a day. That's around 50-60 posts per day, across the network.

That's decreasing, but it's still 50-60 posts more than we need.

A couple of months ago, Pops came around and did a feasibility study on integration between the two systems. The results of that seemed pretty favourable - to quote Pops:

Seriously though, that's crazy impressive.

As part of this quality project, I'd like to see us do something about spam - that may be working out an integration strategy, it may be something else on SE's end, or it may be something that doesn't need SE involvement - but let's see how low we can get those spam numbers.

29
  • 9
    Yup, the reason I was looking at Smokey then was ahead of this project. Be back with you soon!
    – Pops
    Oct 17, 2016 at 17:20
  • 5
    That's precisely what we're going to look at. I designed the original system and it has been holding, but we need to keep it modern. The problem with the system now is 100 spam posts come in, they're blocked within seconds, but there's still 100 spam posts to clean up which gives the perception of swiss cheese on the front end. Integration with Smokey at the first instance could conceivably prevent 70 of them from being posted, we just need to figure out how we can make it happen.
    – user50049
    Oct 17, 2016 at 17:22
  • 1
    @TimPost Yep, makes sense. We've got some ideas about how things can be interested - just yell when you want to get discussing :)
    – ArtOfCode
    Oct 17, 2016 at 17:24
  • This was already discussed in the question... "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."
    – hichris123
    Oct 17, 2016 at 17:45
  • @hichris123 Well, he added a bit more context to it (while enthusiastically agreeing that we need to do more) - I think he's just pretty happy that we're getting close to getting to it :)
    – user50049
    Oct 17, 2016 at 17:54
  • 12
    @TimPost I'd much rather see other stuff prioritized rather than spam (like bullets 1 & 2). I say this even being one of those who works on SmokeDetector. :)
    – hichris123
    Oct 17, 2016 at 17:56
  • 2
    By "spam" do you mean "unsolicited commercial advertisements?" Or do you subscribe to the looser form of the term which basically means "anything unwanted?"
    – user102937
    Oct 17, 2016 at 17:58
  • 7
    That definition appears to mean "anything unwanted." I ask, because UCE seems to be well-controlled on the Stack Exchange sites, while question quality is still somewhat in "The Wild West" territory. If we're going to talk about these things intelligently, I think it's important to make the distinction.
    – user102937
    Oct 17, 2016 at 18:01
  • 4
    On Stack Exchange, the term "spam" has, historically anyway, been limited to unsolicited commercial advertisements. A good example of that is the spam flag. This is very different than other communities and forums on the Internet, which often use the term "spam" to refer to any undesirable or unwanted content. I agree with @RobertHarvey that we should be consistent. I think favoring the narrow definition is better, since it's a particular type of flag. I've rejected spam flags unwanted posts, but not unsolicited commercial advertisements because it doesn't meet the SE definition of spam. Oct 17, 2016 at 18:08
  • 1
    @Robert I was going to start a thread on "spam". It seems some (younger?) people take the definition from the verb form "to spam" that apparently means "to pepper someone with a lot of junk" and use the flag to mark the posts as a thing used to spam the site. ie About ¾ of the "spam" flags I see are not spam, but just "annoying junk" with no "advertising" content (which I decline, usually with an explanatory custom message). Maybe there should be a custom "This is not spam! Just downvote or vote to delete" canned decline response
    – Bohemian
    Oct 17, 2016 at 18:34
  • 1
    @TimPost Be careful, if you give feedback to the spammer to quickly about their post, they can learn how to bypass SmokeDetector, at present they don’t know they have been detected, as most spamming systems don’t check for results after getting conformation the post has been made. Oct 17, 2016 at 20:22
  • 1
    @IanRingrose They can already figure out how to pass Smokey if they want to; the code is freely available on GitHub. They don't, though.
    – ArtOfCode
    Oct 17, 2016 at 20:36
  • 1
    @ArtOfCode It's not about whether it's possible to circumvent, it's about whether they realize that they need to circumvent it, and how quickly they're able to figure out how. Rapid feedback (e.g. rejecting posts) does make both more likely. I do know a thing or two about SEO spam, and can tell you that feedback speed there is a very genuine concern. It may still not be an issue here, because those spammers are probably also spamming a ton of other places and may not notice/care too much about rejection here, but that's a different reason.
    – Cascabel
    Oct 17, 2016 at 23:12
  • 1
    @Jefromi The built-in system already rejects posts immediately. The only thing that would change with integration would be that Smokey feeds the system to tell it what to reject.
    – ArtOfCode
    Oct 17, 2016 at 23:38
  • 2
    @IanRingrose Not reject the post - act as if the post has been accepted, but delete it immediately after posting. Most spam systems won't go back and check.
    – ArtOfCode
    Oct 18, 2016 at 15:02
56
+200

Make close votes scale with reputation

Right now, users with 10,000 or more reputation get to vote to delete 5 closed questions per day. This number increases as users get more rep, to a maximum of 30 daily delete votes.

Why on earth do we not apply the same logic to close votes that are a lot less destructive potent than delete votes?

3
  • 30
    This is planned. What we don't want to do is make our most avid users feel like they have even more responsibility (and feel bad if they don't have the time). Scaling with rep works a bit, but also scaling with the efficacy of their previous votes. Related discussion here
    – user50049
    Oct 17, 2016 at 19:11
  • 3
    @TimPost what happened to that dashboard thingy? related message on the tavern
    – Braiam
    Oct 18, 2016 at 0:31
  • Overzealous closing by high-reputation users of valid or arguably-valid questions is a problem IMHO. Super-high-reputation users are too far removed from the general population to be given that extra power. So, -1.
    – einpoklum
    Dec 9, 2017 at 11:13
55

Help new users understand that answers aren't for comments.

This has been an issue for years and years and years. There have been loads of proposals to fix it - by getting rid of the 50-rep comment threshold, having low-rep users' comments reviewed, and so on and so forth. These specific solutions have been , for reasons I agree with, but I still think we need to do something to explain to users that they can't use answers as comments.

A close I found is Low-rep users posting comments as answers, which suggests adding a pop-up in certain cases when new users try to post answers. A reasonable objection was that the Low Quality Posts queue often deals with these, but I think we can still do something to cut the issue off before answers reach the queue.

Perhaps the pop-up could

  • Explain to these new users that they need 50 rep to comment, and link to the site's Privileges page.
  • Explain why answers shouldn't be used for comments.
  • Explain that comments posted as answers will likely be downvoted and deleted. It could contain examples (posts with the gist of "Me too!", "Nice question!", "I had the same problem", etc.).

The pop-up could be shown to users the first X times they answer a question and have less than Y rep (should Y=50?), with X hopefully fairly small. To be effective, it would have to be highly visible; putting it off to the side, as Blender suggested, might not be enough.

When I flag answers on Stack Overflow, a lot follow certain patterns, so it's easy to find them via queries after the fact. The ones I find are generally from people who have the same problem as the asker, so maybe a second pop-up could contain additional information explaining that comments should not be used for this purpose, triggered for newish users when the answer text contains phrases like

  • "Please help me"
  • "I had the same problem"
  • "Did you solve this?"

and so on.

6
  • 7
    Yes, absolutely this. Even if it leaves answers that request some kind of clarification (which they may insist on doing anyway), then so be it. At least helps in some way. But these three answers that you mention at the bottom just need to go away forever.
    – Jamal
    Oct 17, 2016 at 22:07
  • 1
    I always thought this was a very silly rule. We expect far higher standards from a posted answer than a posted comment. The solution is to get rid of the rule. I believe the main reason why the rule exists is some moderator tool implementation fiasco that makes it harder for mods to moderate comments than answers, and to protect against spam.
    – Lundin
    Oct 21, 2016 at 11:14
  • Since this is an issue which will impact new users it needs to be considered as part of that wider review. In particular, new users could usefully receive some guidance/reminder on positive strategies for gaining reputation so they can participate more fully in the site. Overpolicing of this rule can contribute to a less than optimal early experience for new users. Oct 23, 2016 at 15:10
  • I don't see the point of this. What is so awful about giving a comment in an answer or an answer in a comment? Just because the software makes a formatting distinction, that doesn't mean there actually is a clear distinction.
    – Greg Lee
    Oct 29, 2016 at 21:37
  • 3
    @GregLee Comments are meant to be temporary; answers are not. Plus, giving them equality would induce more of a forum-like atmosphere, which isn't the point of Stack Exchange.
    – HDE 226868
    Oct 29, 2016 at 21:45
  • @Jamal, actually, I could easily imagine someone stumbling across an unanswered question page on their way to finding a solution, then finding a solution, then going back to the question page as an unregistered user and posting an answer that starts out, "I had the same problem, and what solved it for me was...." And if that gets blocked, do you think they'll stick around to figure it out? No, they'll be gone, and so will the knowledge they would have shared.
    – Wildcard
    May 12, 2017 at 2:42
53

Can't believe no one has posted this one yet... Allow badge holders to have binding close and reopen votes for other close reasons.

The duplicate "hammer" has been... Wildly successful. But perhaps most surprising is how popular instant reopen votes have been: over the past 90 days, just a hair under 60% of all reopened questions have involved a gold-badge holder swinging that hammer of mercy.

Turns out, folks who've answered a crapload of questions to great success are as likely to call out incorrectly closed questions as they are to close those that need it.

...So why are we still wasting all this time with voting? Let's at least extend it to "Unclear", the reason I'd wager folks who've done a lot of answering are more than adept at gauging. Maybe disable it for questions that've already been answered (our definition: +1 answer or accept) just as a safety check.

And just make reopening available across the board.

20
  • I think that a lot of that comes from gold badge holders not really disagreeing with the duplicate closure but wanting to provide their own answers, dupe or not. I don't think thats a measure of success personally, but one of an open avenue to abuse the Mjolnir privileges. There are even gold badge holders who disagree with closing duplicates as a concept, do you really want to give those people even more unchecked authority?
    – Magisch
    Oct 19, 2016 at 7:10
  • Or vote to reopen to close as another duplicate. Happens often too. I would like to see some more statistics. Oct 19, 2016 at 7:50
  • 11
    I totally disagree with "disable it for questions that've already been answered", on SO the answers that gain upvotes the quickest are wild guesses to terrible questions. Why shouldn't we be able to close those? Oct 19, 2016 at 7:52
  • 9
    maybe nobody posted this because this feature is strictly specific to Stack Overflow? how many other sites have gold tag badge holders. I think that even silver tag badges are rare outside of SO, otherwise somebody would probably already asked here about ideas discussed in Empowering tag-badge holders part II - let's look at silver?
    – gnat
    Oct 19, 2016 at 8:25
  • Is there good number-crunch data that supports this? If this is indeed a good idea, I would expect correlations like 'dupehammer holders are unlikely to be countermanded on reopen votes on unclear-question closures' to be reasonably strong. This would be interesting data to see.
    – E.P.
    Oct 19, 2016 at 13:07
  • @gnat We have several gold-badge holders on Computer Science and it's by no means one of the largest sites.
    – Raphael
    Oct 19, 2016 at 14:33
  • @Magisch The reopen-answer pattern seems easy to detect. We could include it into the suspicious behaviour detection cronjob.
    – Raphael
    Oct 19, 2016 at 14:34
  • 3
    I have one fundemental problem with this idea. Having a gold badge means that you have contributed a lot and/or good posts. It does not say anything at all about whether the user is a worthwhile moderator. I'd rather see these privileges handed out based on constructive community moderation contributions. (Of course, the whole rep-driven privilege system has the same problem. Oh wait, let me make that an answer.)
    – Raphael
    Oct 19, 2016 at 14:36
  • 1
    @Raphael your last comment is also an explanation why having several gold-badge holders on Computer Science doesn't help in making this feature relevant for smaller site. This is strictly specific thing of SO that they have not "several" but many such users (hundreds if not thousands I think). And that makes it so that even small fraction of these users who got interested in moderation can make substantial contribution. When (smaller) site has 10 holders and 1 of them is interested, this doesn't help. When SO has 10,000 holders and 100 of them are interested, they can make a difference
    – gnat
    Oct 19, 2016 at 14:49
  • @gnat You have do account for the number of full-fledged moderators and traffic, as well. On a small site, having an additional moderator (effectively) on one of the few very active tags does help!
    – Raphael
    Oct 19, 2016 at 14:51
  • 1
    @gnat, Raphael... A good bit of this data is in SEDE: overall stats for the last year, reopen voters.
    – Shog9
    Oct 19, 2016 at 14:59
  • @Raphael that assumes that site indeed has them, "additional moderators". Of 10 holders, it statistically can happen that no one is interested, or that one interested is away etc etc. That's what makes a radical difference at SO, where they are guaranteed to always have large enough number of interested holders
    – gnat
    Oct 19, 2016 at 15:00
  • @Shog9 I see, thanks. ~300 hammered at AU, ~450 at SU, ~1,450 at Math.SE and (drumroll) 73,479 at SO. Helps to feel the difference
    – gnat
    Oct 19, 2016 at 15:05
  • @gnat You were proposing 1 effective semi-mod, not me. My point is, don't look only at the absolute numbers. It's the percentages that count. If we have 4 instead of 3 mods on a high-traffic tag (by our standards) and the gold-tag guy hammers 10% of the questions that need to be closed, that is helping.
    – Raphael
    Oct 19, 2016 at 15:09
  • 12
    I like this suggestion a lot more than scaling the number of votes we have. Adding more votes just increases fatigue, if users even decide to use more of them. Adding more weight to votes means we're more effective with the votes we're already using anyway.
    – davidism
    Oct 19, 2016 at 17:59
53

Allow site specific placeholder texts to be added to the question box

Many users evidently ignore the side bar which explains how to ask questions, and what we expect to be present in our questions. A placeholder text would be seen by almost everyone. (Would it work in mobile? I don't know.)

Here is a quick mockup I made for ELU: (If this was implemented sites like ELU would need to spend considerably more time deciding what a good placeholder text would be.)

Image showing the new question page with a placeholder text in the question body

This would pair very well with Tag specific question requirements

1
  • 10
    This might be a good idea for the answer box, too.
    – user
    Oct 20, 2016 at 11:54
52

Make every closure, answer, and comment deletion work like closing as duplicate

I really like the “new” UI when somebody votes or flags to close a question as a duplicate for several reasons:

  • The author is directly informed what people consider to be problematic about the post. (While they may be totally oblivious if their question is closed for another reason.) Thus they can act before there is any effective action on the post (resulting in disappointment and the difficulty to undo this action).

  • The author is presented a reasonable choice of options how to proceed – either to accept the verdict or edit the post to improve it. Should they accept the verdict, there is no pointless waste of reviewer or moderator time.

  • Reviewers, moderators, and other users are informed as to what the original flagger/voter thinks is problematic about the post.

  • All of the above is done with the UI, not with a custom comment or similar.

Note how I did not use the words duplicate, question, or closure to describe these advantages? I see no reason why a similar UI should not be applied to closing for another reason, impending deletion on an answer, and even comments.

More specifically, I suggest:

  • Adapt the mechanism for other close reasons.

  • Introduce answer-deletion reasons (and get rid of the confusing and pointless distinction between VLQ and NAA (1, 2)); inform the author directly as to what is considered problematic with the answer; make it obvious to the author that they can edit or delete their answer. Give flaggers the option to choose a reason (instead of having only reviewers select canned comments). The UI of the low-quality queue badly needs an overhaul anyway (1, 2).

  • Inform authors when a comment is flagged and why. This particularly makes sense when the reason is obsolete. It may also work if the discussion in the comments digresses or if answers are posted as comments. Authors of comments flagged for other reasons may not be so understanding, but at least they may take the hint after a series of their comments was flagged. Right now, persistent comment chatterers may not even notice tons of their comments getting deleted.

Of course, in most if not all cases, flagged posts should still be subject to review if the author does not agree with the flag/closure. Also note that there are some cases where this is probably not a good idea, e.g., spam flags.

6
  • 15
    Part of this proposal effectively suggests we get rid of the 250-rep view close votes on your posts privilege, making it available to everyone. In reviewing this proposal, it may be worth considering why this privilege was ever necessary. Oct 18, 2016 at 6:19
  • @JeffreyBosboom, or just hide the user names of the voters when below 250 reps. Oct 18, 2016 at 13:56
  • 5
    Trivia: that privilege predates the "vote-to-close" system, and was once called "Close your questions", @Jeffrey. Prior to early 2009, you could (at 250 rep) close your own question, preventing further answers from being posted (given there was no automatic deletion, no delete votes, and no moderators, this was pretty much all it did unless Jeff caught you doing it). Once the vote-to-close system landed, that particular privilege pretty much lost any utility it might have had, hence the current focus on being able to view close votes.
    – Shog9
    Oct 19, 2016 at 0:34
  • 16
    To Wrzlprmft's point: the duplicate UI change had a fairly striking effect on asker editing: it doubled, pretty much instantly, and mostly stayed there. Granted, that took it from about 6% to about 12%, but that's still quite a few questions. I could see this being a HUGE win for stuff like "Unclear".
    – Shog9
    Oct 19, 2016 at 0:48
  • 6
    The longer I look at this, the more I like it.
    – Jan
    Oct 19, 2016 at 11:59
  • @Shog9: Granted, that took it from about 6% to about 12%, but that's still quite a few questions. – Also, we can presume that a considerable portion of questions flagged as duplicate actually are duplicates, in which case it is fine for the asker to leave the question as it is. This does not apply to any other closure or deletion reason, where a correct flag means that the author should edit or delete the post.
    – Wrzlprmft
    Dec 5, 2016 at 20:53
51

Implement a working FAQ system

It is a truly painful process to find a duplicate of a question you know has been asked many times before.

The current system with the "frequent tab" is not working well. It is based on close-as-duplicates and similar activities, rather than community consensus. A number of "canonical FAQs" eventually spawn from this system over time but they drown in the number of less canonical questions present on the "frequent tab". And we always end up with several "canonical duplicates" to the same question.

After speaking with many other users who do lots of user moderation in their favourite tags (gold badge/dupe hammer), it turns out that almost every such user uses the site's "favourite" feature as their own private "canonical duplicate" list for moderation purposes. Essentially every such user has created their own FAQ. There must be so many great links gathered in each such private FAQ, but they aren't shared with other user moderators.

On SO there exists one case where the C++ community has taken the matter in their own hands and created a tag (link). The site has allowed this because their FAQ system is heavily user-moderated. As a result, this system works far better than the "featured tab".

Instead of having all these "private FAQs", shouldn't we gather all this knowledge into creating an official FAQ system that actually works? Which is tag-based and points at FAQs/canonical duplicates for each tag. Which is created explicitly by the community, rather than implicitly/by chance.

10
  • 7
    (The Documentation project will not solve this, since you can't close something as a duplicate while pointing at a Documentation page.)
    – Lundin
    Oct 18, 2016 at 7:03
  • The tag wiki could be used for this, or let gold tab badge holders mark (or vote for) questions as FAQ and have a new FAQ tag. Oct 18, 2016 at 14:01
  • @IanRingrose It is often used for that too. But it isn't consistent across tags. Plus there tends to be a lot of clutter edits to the tag wikis. A system where gold badge tag holders can nominate a question for FAQ sounds like a very good solution. A special kind of votes "agree/disagree that this should be FAQ" could be introduced.
    – Lundin
    Oct 18, 2016 at 14:14
  • 6
    So instead of a system based on frequently-asked questions, you want a system based on what folks think are frequently-asked questions?
    – Shog9
    Oct 19, 2016 at 3:17
  • 3
    @Shog9 The main problem is that there often doesn't exist a canonical duplicate for a frequent question, or it is too hard to find for anyone to bother. So people just keep answering the same thing over and over. Now if you make an effort into creating a canonical duplicate, like making a community wiki together with others who read the same topic, it will take ages before that one makes it to the "frequent tab". Another problem is that there are hundreds of FAQs per topic, so even if the frequent tab is relevant, it can be very hard to find what you are looking for.
    – Lundin
    Oct 19, 2016 at 6:51
  • 3
    Veteran users often complain about lack of research from new users. But since the veteran users themselves can't easily find a topic, then what hope is there fore new users?
    – Lundin
    Oct 19, 2016 at 6:54
  • The problem with this is that most of the time the system already knows what is frequently asked, based on hard data, rather than individuals impressions. Example rust tag faq generated includes 5 out 6 rust-faq tagged questions, r tag the first page is filled with faq tagged questions, 2 out the most linked c++ questions doesn't have the faq variant.
    – Braiam
    Oct 19, 2016 at 20:12
  • @Braiam It would be interesting to know the algorithm that determines what shows on that tab. My impression is that the "frequent tab" also includes frequently linked questions, and not just frequently-closed-as-duplicate. Looking at the C tag, most of the top ranked posts are purely frequently linked, such as a book recommendation list, which would never be used for close-as-duplicate. If this is true and the page lists "frequently linked", then that's one major design mistake. When looking for canonical duplicates, I'm not in the slightest interested in "generic good advise" questions.
    – Lundin
    Oct 20, 2016 at 10:30
  • 4
    @Shog9 I think you missed the point. The point is that the system fails at automatically identifying FAQs and users are doing a better (if imperfect) job of identifying them. I tend to agree with the sentiment here; the site is generally very bad at helping you find potential duplicates. In the situations where I'm confident there's a dupe, I always turn to Google before any site tools.
    – jpmc26
    Oct 22, 2016 at 20:33
  • 2
    Well, that's a different issue, @jpmc26: it's hard to beat Google's search; they're kinda good at it. Tons of folks have their own personal lists of common duplicates, tons of tags maintain "frequently-asked" lists in their wikis, and all of that is useful... But that all takes a great deal of effort to maintain and use, and falls apart as soon as something changes; there's no magic bullet to be found in curation - if there was, Yahoo! would still be a big list of sites organized by category.
    – Shog9
    Oct 22, 2016 at 22:23
50

Let mods (and 10k?) know when questions go "hot"

Because on small(er) sites, influx from the Hot Questions list can mean a disproportionate amount of not-quite-productive posts. A heads-up for community moderators would be great.

3
  • 3
    We already have the moderator tools privilege for 10k+ users. It can be used to see most viewed, commented, edited, and voted posts, but you're right, it's quite buried in the menus. Some sort of easily visible menu or toolbar might really help bring experienced attention to the hot spots.
    – Anko
    Oct 21, 2016 at 12:00
  • 9
    @Anko Yes. "Data is accessible somewhere" != "notification".
    – Raphael
    Oct 21, 2016 at 16:07
  • 1
    An option to remove the question from the HNQ list would be good, too. Nov 5, 2016 at 14:35
49

Meta Request: Get rid of "Jeff said so"

Over the years, I have encountered several instances of "Jeff said so" reasoning, that is design decisions were justified with the opinion of the founder of SO but little else. Well, he'll soon have been gone for longer than he was here.

Please keep an open mind regarding ideas that violate policies/principles that "have always been there". The network has come a long way since the days of Jeff. Once reasonable assumptions do not fit anymore.

(To be clear, I'm using "Jeff" as a shorthand for "any kind of dogma".)

16
  • 3
    "The network has come a long way since the days of Jeff." and thank god we followed those advise, imagine what would happen if everyone got whatever they wanted... madness.
    – Braiam
    Oct 18, 2016 at 1:54
  • 16
    @Braiam: This is not about giving everybody what they want; this is about avoiding unnecessary dogmatism. Consider, e.g., the request to inform askers when their question is closed. The only arguments against it came from Jeff (now the answer is deleted but I mostly quote it in my answer). I am pretty confident that this would have long been implented if Jeff hadn’t said so.
    – Wrzlprmft
    Oct 18, 2016 at 4:40
  • 35
    Maybe get rid of 'Joel said so', given that Jeff's ideas were a lot better than Joel's.
    – Miles Rout
    Oct 18, 2016 at 7:27
  • @MilesRout and the 'Joel said yes' too...
    – Braiam
    Oct 18, 2016 at 13:03
  • 8
    As a fellow Jeff, I disapprove of this message...
    – JeffC
    Oct 18, 2016 at 19:23
  • Which specific assumptions do not fit anymore?
    – Travis J
    Oct 18, 2016 at 20:17
  • 10
    @TravisJ To name just one example: "Every question is ontopic on exactly one site." That said, this is more about the general mindset. I have seen many feature requests closed as duplicates of status-declined posts that are a) old and b) declined without much discussion but a healthy dose of dogma instead. This has to stop.
    – Raphael
    Oct 18, 2016 at 22:15
  • 1
    @Wrzlprmft many of Jeff worries were more or less well founded except for a handful ones, but many more were critical for the correct grow of SE sites.
    – Braiam
    Oct 19, 2016 at 19:55
  • 6
    @Braiam: If you are doing something because it is well-founded, you are not doing things because Jeff said so and you are not dogmatic. Ideas should be judged by their usefulness and not by their source. That’s the entire point of this answer.
    – Wrzlprmft
    Oct 19, 2016 at 20:00
  • @Wrzlprmft remember that many things that Jeff said weren't meant for us... SE could say "Jeff said so" without elaborating, but Jeff may have good reasons, or no reasons at all.
    – Braiam
    Oct 19, 2016 at 20:13
  • 3
    @Braiam which seems to be the point of this answer; 'stop giving us "Jeff said so" as justification for certain things'. If Jeff said so because X, Y, Z, then use X, Y, Z as the justification. If he just said so to say so, then say "because we want to and don't have a better reason or alternative".
    – TylerH
    Oct 24, 2016 at 17:57
  • 1
    Jeff Atwood stopped being an influencing factor ages ago. As to "any kind of dogma," we're never in short supply of that. When you say "stop being dogmatic" to some people, you might as well say "stop being a dog" to your dog.
    – user102937
    Oct 25, 2016 at 18:46
  • 1
    @RobertHarvey I distincly recall having (and seeing) seeing questions closed as duplicates of such that were answered (and "closed") by one Jeff saying "no, because" ages ago.
    – Raphael
    Oct 25, 2016 at 22:07
  • 1
    @RobertHarvey Yea, except these get closed as duplicate as well.
    – Raphael
    Oct 25, 2016 at 22:23
  • 1
    Does StackExchange (as a company) have a set of mission statements? Is "what Jeff said" still relevant to those? I think being able to directly quote mission statements or even guidelines would be a fine way to combat this thinking.
    – user343082
    Oct 30, 2016 at 15:14
46

Tag specific question requirements

As proposed in Tag specific question requirements

We currently have some handy tag-specific alerts when asking questions with certain tags. These are for certain problem questions that are only a good fit for the site if they follow certain requirements. The alerts list these guidelines and specific requirements.

enter image description here

These alerts are great, but they don't go far enough and people simply ignore them a lot of the time. I propose we make these more prominent and permanent. They are currently (I believe) only implemented by devs on a case by case basis, let's give that power to the people who care about them (certainly mods and maybe high rep users).

A quick mockup:

enter image description here

I'm not totally sold on my mockup, it's just a quick idea. Maybe the check boxes on each point are too much, I don't know—but there needs to be some interaction.

5
  • 6
    people simply ignore them a lot of the time – Which could be due to the fact that people do not see them most of the time.
    – Wrzlprmft
    Oct 20, 2016 at 4:51
  • Ah yes, I do mention that in the feature-req, didn't realise it was already reported.
    – Cai
    Oct 20, 2016 at 7:35
  • And soon you'll have walls of text like it.wikipedia.org/w/… before the submit button-
    – Nemo
    Oct 30, 2016 at 18:20
  • But #peopledontread even the text they get already... :-(
    – einpoklum
    Dec 9, 2017 at 11:16
  • @einpoklum sure, that’s the point. Interaction at least forces them to look and if they still haven’t read then we know they intentionally didn’t read.
    – Cai
    Dec 9, 2017 at 11:21
43

Some low-hanging fruit:

Give moderators one-click declines on auto flags

I mentioned this a few days ago in SO's mod room:

On auto flags, we should have two one-click dismiss buttons - 'helpful' and 'not helpful'

That'd let us get meaningful numbers on how useful auto flags are - right now I tend to dismiss them as helpful whether they are or not, just because it's two clicks vs. four clicks.

I don't know what, exactly, we'll find from this... but I know we'd find something. There are plenty of auto flags that aren't really necessary, and mod time wasted on those can't be used in other (more valuable) places.

This would let staff look for / verify patterns on a deeper, more objective level than we currently can.

7
  • 2
    In particular, the "Too Many Comments" flags. Comments don't need to be scrutinized by a moderator at all unless someone raises their hand.
    – user102937
    Oct 17, 2016 at 21:09
  • 4
    Side note: If you think an auto flag isn't necessary, then poke one of us in the mod room. I'm always looking at ways to adjust some of these auto-flags and will dig into any of them to see if they are beneficial.
    – Taryn
    Oct 17, 2016 at 21:09
  • @RobertHarvey Those flags actually age away after a period of time if there are no additional comments being added to the conversation.
    – Taryn
    Oct 17, 2016 at 21:10
  • 2
    @RobertHarvey for what it's worth, every time I see one of those on Workplace it almost always includes some, ah, less than useful content.
    – enderland
    Oct 17, 2016 at 23:41
  • @enderland Could definitely see that on the more soft-subject sites, but SO and others tend to raise cases like stackoverflow.com/a/40050748/1849664 (not worried about revealing the flag, you could SEDE it up anyway) where the comments might not need to stick around, but it'd be a prohibitively large amount of work to evaluate every one, in every case.
    – Undo
    Oct 17, 2016 at 23:44
  • @Undo alternatively just make it so mods on any site can turn off any community flag...
    – enderland
    Oct 17, 2016 at 23:47
  • Honestly, when I see the too many comments flag, I ignore it and leave it for a fellow mod (sorry guys). Part of me wants to do something to see if there's any useless or low value comments to delete, but it takes time that I often don't have when I jump on in a short 5 minute break at work. Sometimes, the comment thread can actually add value to a future reader and it's a lot more visible than a chat room, so I don't want to blindly delete, but I also can't immediately read and review. Oct 17, 2016 at 23:52
43

The biggest quality problem I see in the site where I spend most time (PPCG) is the Hot Network Questions sidebar. It's set up to promote cheap questions which anyone can answer without thinking, and because it channels in people with association rep allowing upvotes but not downvotes it self-reinforces the promotion of those questions and gives all newcomers (those who arrive via HNQ and those who simply find the site and look at highly voted questions) the impression that that is the best type of question to ask.

There are various feature requests relating to reducing the promotion of poor quality questions via the HNQ. The two I would particularly like to see are:

Another one which would help to mitigate the anti-quality effect of HNQ is:

10
  • 14
    Here's one less-extreme way to cut that feedback loop: allow 'outsiders' to vote, but stop counting those votes as a signal for the HNQ algorithm.
    – E.P.
    Oct 19, 2016 at 12:48
  • 3
    That cuts the HNQ feedback loop, but it doesn't cut the feedback loop which drags the site's culture towards questions which are optimised for the HNQ rather than the site. Oct 19, 2016 at 18:22
  • @PeterTaylor: If clickbait can't stay in HNQ for very long, it's relatively less rewarding, and therefore requires more work to get the same abnormal boost… work which could probably be put to make the post better in a less unnatural way to improve HNQ and organic voting response. The result is that there is less reason to optimize for HNQ, because HNQ-only optimization doesn't work as well. Oct 20, 2016 at 7:44
  • It may be that questions are more likely to accumulate close votes after getting on the HNQ list. When you increase a question's exposure, it becomes more likely that someone seeing that question is going to think it's close-worthy. Removing a question from HNQ if it has close votes would have to take that into account.
    – David Z
    Oct 22, 2016 at 8:33
  • 1
    What if instead of “hot” questions it listed “well answered” questions (has a highly upvoted answer, many views, and if it gets many more answers that pushes it off the list)? That would still highlight interesting content of general interest.
    – Kevin Reid
    Oct 29, 2016 at 19:55
  • 1
    @KevinReid, there may be some subtle problem with that which escapes me, but my initial reaction is that it's an excellent idea. Oct 29, 2016 at 20:21
  • I use Stack Overflow Extras (SOX) to hide the Hot Network Questions sidebar. Jan 4, 2017 at 22:32
  • @FranckDernoncourt, I don't see the relevance of that comment. Jan 5, 2017 at 13:10
  • @PeterTaylor hiding the sidebar may be a solution for people who do not like its content. Jan 5, 2017 at 15:39
  • 3
    @FranckDernoncourt, but the problem that this answer is about is the content shown to other people and the consequences of that. Jan 6, 2017 at 20:58
41

As a "meta" quality improvement, it would be great if the existing quality requests on Meta.SE and Meta.SO got an update from the SO team so we know where things stand. I understand that, in general, responding to each feature request is not an efficient use of time, but since there are relative few requests related to the quality project it might be more reasonable. For example:

0
37

For those few of us who are heavy voters…

That's a very small target demographic, but the requested changes are pretty minor.

Bring back the indicator of the number of votes cast in the current quota period.

Relax the voting quotas. I shouldn't have to care that it's almost midnight and I need to get my voting done or else I'll run out every day until next week-end.

7
  • 3
    Another pertaining feature request
    – Wrzlprmft
    Oct 18, 2016 at 4:33
  • 7
    Maybe have a "pot" of votes, that have a max value of N, with 1 vote added to the pot every 1440/N minutes. (N being the current daily limit) That way voting at the start or end of a "day" will make no difference to the number of votes you can use. Oct 18, 2016 at 14:06
  • 3
    Pretty sure that indicator was added to encourage folks to vote more. Which... Is still a good idea.
    – Shog9
    Oct 19, 2016 at 3:26
  • 1
    @IanRingrose That would be a good answer for my feature request. I don't care exactly how it's done, what I care is that it doesn't make a difference whether I do my daily trawl around 23:30 or around 0:30. Oct 19, 2016 at 7:03
  • @Shog9 Any idea why there's a voting cap in the first place? (I'm not an active enough user to have ever come close to hitting the cap, I expect, but I'm still curious.) Oct 19, 2016 at 17:51
  • 3
    Someone got the bright idea of writing a script to up-vote everything on the site back in the private beta, @Kyle. Kinda made the need obvious.
    – Shog9
    Oct 19, 2016 at 17:53
  • 2
    It would also be nice if the voting cap was just a little higher - it doesn't even have to be a lot more, but a cap that was even ten votes higher would help a lot. Especially if there's a lot of activity on some sites on some days, and there are lots of posts worth voting on - which shouldn't be a bad thing, but it can be annoying when I will be playing catch-up for a week or more because most days I use most of the votes anyway.
    – Megha
    Oct 23, 2016 at 12:08
36

Make it easier for people to remove bad and incorrect answers. Some possibilities for discussion:

  1. Allow delete votes on positively scoring answers - obviously we need to scale the delete votes along the same lines as question delete votes are scaled so that it's next to impossible to delete something that's highly upvoted. If a bad answer that's got a couple of upvotes starts to attract delete votes then it might prompt the OP to fix it up, others to fix it up or add a new answer themselves. Obviously delete votes should age away, perhaps more rapidly if the post has been edited since the vote was cast.
  2. Educate users more to remind them that they'll get their precious 1 point of rep back when the offending answer is deleted.
  3. Make downvotes on answers free to the caster - or perhaps at least after a certain reputation level/length of time on the site.

    I think that this will have a big impact on quality, as I am sure many people are reluctant to hit that down arrow because of the -1.

    "What!" I hear you say, "people are reluctant to lose one point of reputation when they see something bad?!?!. That's just crazy". It may be crazy but I fear it's true, especially for low rep users who are struggling to reach the next privilege threshold.

    This results in:

    1. Poor quality answers being left on the system with no indication to say that they are poor quality.
    2. Users flagging for moderators to delete stuff because they can't do it themselves as the post is not negatively scoring.

    All-in-all not useful. A word of warning though - if down-votes are free it becomes easier for people to down-vote all the competing answers on a question. We'd need to have a way of spotting that and either preventing it (the -1 remains if you've answered the question, or add an answer later), reversing it automatically or flagging it up to the moderators.

If the community. properly peer reviewed, can do the cleanup there's less work for the janitors to do.

13
  • You think that in order to improve the quality of Stack Exchange we need less answers? I just don't see that being the case. The largest problem is with low quality questions, especially from certain subsets of the community, as explained elaborately in this post.
    – Travis J
    Oct 17, 2016 at 20:03
  • 17
    @TravisJ we need fewer low quality, or down-right wrong answers. We need people to believe that when they come to a Stack Exchange site they're getting good answers that provide correct information.
    – ChrisF Mod
    Oct 17, 2016 at 20:05
  • Perhaps for small traffic sites, but for the larger SE sites, users are getting good answers that provide correct information. If you were to look at metas across the network, there are basically no posts stating "answers need to be of higher quality" anywhere. Question quality, on the other hand.. those complaints are everywhere.
    – Travis J
    Oct 17, 2016 at 20:09
  • 8
    I would say that every post on Meta.SO about NAA flags is a post about answer quality, @TravisJ. And there are plenty of those.
    – jscs
    Oct 17, 2016 at 20:13
  • @JoshCaswell - I think there is some truth to that, although users are also constantly confused with the flagging mechanism. Many of those NAA flag questions are "why was my flag declined".
    – Travis J
    Oct 17, 2016 at 20:17
  • 4
    @TravisJ - probably because they used a NAA flag instead of a down-vote.
    – ChrisF Mod
    Oct 17, 2016 at 20:18
  • 5
    I think there's too much potential to abuse a completely free downvote -- especially among inexperienced users.
    – Glen_b
    Oct 17, 2016 at 20:51
  • 12
    @Glen_b there was concern that free down-votes on questions would lead to chaos, but that didn't happen.
    – ChrisF Mod
    Oct 17, 2016 at 20:51
  • 5
    @ChrisF I'm not forecasting chaos, just an unwelcome degree of abuse among a subset of users; e.g. the temptation to downvote some or all the other answers - regardless of merit - when you post an answer may be quite high for a new user. There's no corresponding moral hazard with questions.
    – Glen_b
    Oct 17, 2016 at 20:55
  • 11
    @Glen_b - you'd still need 125 rep to be able to down-vote, so you wouldn't be that new a user. Also I did raise that concern myself.
    – ChrisF Mod
    Oct 17, 2016 at 20:56
  • 3
    @ChrisF Well, relatively new. Some people break 125 their first day on site. Even after I hit the downvote level of reputation (a long time after my first day), I don't know that I'd have had the sense to know better before I'd been on site a good while longer. It was fortunate that the cost of a downvote was there. By the time I had enough reputation not to care that it costs reputation, I understood enough not to misuse it. Some people won't need that, but some will, and it's the ones that will who break things.
    – Glen_b
    Oct 17, 2016 at 21:18
  • 9
    Many people (on the sites I frequent) refuse to downvote because they don't want to be mean. I don't know how to address this, but removing the -1 rep will certainly not do so.
    – Raphael
    Oct 17, 2016 at 22:59
  • 2
    somewhat related: When there are many answers already, help me check that mine won't repeat others (there would be less of bad / repetitive answers if system would help answerers check that)
    – gnat
    Oct 18, 2016 at 6:26
36

Cater for other screen sizes

If you use the full site, then a lot of things require horizontal scrolling if your screen is less than the prescribed width, and some things are simply placed off-screen. I'm not sure what the prescribed width for ordinary users actually is; I know that my preferred width (which, guess what, depends which computer I'm using) usually isn't enough. For moderators, there's an indication of the presence of handled flags or deleted comments on a post in the left margin¹. It's provides information which otherwise can only be seen behind a lot of clicks and manual URL modifications. To see all moderator tools, a minimum of 1400 pixels is required (more on a site that gets more than a double-digit number of deleted comments on a post, down to 1361px if you don't mind seeing how many deleted comments are on a post). That gives a handy empty ~150px bar on the right for… um… why would you want to have an empty margin in a window? I don't know but I'm sure that studies show that 93% of users prefer more than 100px of empty margins in their windows. Or whatever.

Instead of making it worse to fit more ads, make it better. Remember, your money may come from advertisers, but your content comes from users. Without content, you wouldn't get advertising revenue in the first place.

If you use the mobile interface then a lot of functionality is missing. The apps are even worse (at least the Android app is so buggy it isn't ready for prime time if you want to do more than read and vote).

This isn't the 1980s. This is the era of browsers using hundreds of megabytes to display a web page on a wide variety of different hardware. There's no excuse for designing for a particular screen width.

¹ Unless you move that with ArtOfCode's userscript.

13
  • 12
    +1 : The Android app leaves a lot to be desired
    – Zaid
    Oct 18, 2016 at 13:26
  • 2
    The android app can't do half of what the mobile site can do. I wish it would just give me notifications about responses and take me to the website. Get out of my way, cripple! Oct 18, 2016 at 16:00
  • 1
    Preferred width is at least 1060px, which is how wide the site content is.
    – TylerH
    Oct 19, 2016 at 1:38
  • @TylerH Is that what it is today? On my main PC my preferred width is 800px. Oct 19, 2016 at 7:08
  • If the Android app offers less than the mobile version of SE, why is it still around? Wouldn't dev time be better spent on other things?
    – Raphael
    Oct 19, 2016 at 14:38
  • @Gilles Yes, the <div class="content"> that contains the site's content column (including the sidebar on the right) is 1060px wide. Since 2010 this means roughly 80% or more of web users use monitors wide enough to cover all the content. Since January 2016 that number is ~94%
    – TylerH
    Oct 19, 2016 at 14:38
  • Yeah, I stopped using the Android app within...days, probably. The mobile-web version is tolerable. Oct 19, 2016 at 17:33
  • 1
    @Zaid: The Android app is great for reading (Hot Network Questions is great entertainment for when I'm on the go). For answerting or asking questions... not so much. Posting comments is fine though.
    – user299767
    Oct 20, 2016 at 1:01
  • I've noticed this too. I actually have to zoom out of the site to see all the relevant content on at least one of the devices I frequent SE on, in order to see everything.
    – enderland
    Oct 21, 2016 at 12:56
  • @TylerH For moderators, it actually takes a minimum of 1361px, otherwise some of the tools are off-page. Oct 21, 2016 at 19:58
  • Art's userscript is linked in an answer to meta.stackexchange.com/q/188628/162102. (For your future discoverability, you placed a bounty on it, as did I.) Oct 21, 2016 at 22:13
  • The only thing I use the Android app for is getting notifications of inbox events. I've never figured out how to do basic site-by-site navigation like I can do on the web. That's...pretty basic. Oct 21, 2016 at 22:15
  • I actually like the Android app, I use it daily when I'm not in front of a PC. Typing long answers, especially when code is involved, is a bit tedious, but manageable. But the main thing I don't like about the app is a lack of access to my favorite Tags. I have to type them by hand in the search box to see filtered questions. Oct 29, 2016 at 0:44
35

Give moderators tools to engage the community

Moderators can not always be sure about whether a question fits site policy, especially if they are not experts in the area the question is about. Also, they may want to defer certain decision to the community, i.e. promote a more democratic approach to community moderation. In particular, they may have to "train" their users to use the tools they have as the site grows.

Currently, there is no tool to do this, short of opening a second account for casting regular votes: every action a moderator takes is immediate and binding.

I think moderators need tools to prompt regular users into action.

Related proposals:

14
  • 2
    Overall, I think this fits into how moderators behave. I can't speak for every site, but the theory is that moderators are really exception handlers and not community leaders. I think that if you look at the majority of the sites in the network, you'll find that moderators tend to be among the higher reputation users and also subject matter experts who can use their knowledge and influence to help do more than just handle exceptions. I'd like to see more at giving moderators community growth tools, since they are trusted. However, there would need to be some level of monitoring over this. Oct 17, 2016 at 23:43
  • @ThomasOwens 1) I don't know about others, but I myself consider myself an expert in some areas of CS but certainly not in all or even most of them. 2) Why would softer feature than we already have need more monitoring? Or have you other features in mind besides those I link?
    – Raphael
    Oct 17, 2016 at 23:48
  • 2
    I think your suggestions are a good start, for sure. But I think the whole role of moderator in engaging and developing the community should be looked at in more detail over a longer time. I don't think a moderator can be an expert in all the subject matter of a site (can anyone really be an expert in all of computer science or all of software engineering or all of programming?), but they do tend to be people who are involved in the subject matter and may be in positions to help grow the site more. There are lots of different features to look at in this direction as a long-term vision. Oct 17, 2016 at 23:52
  • 17
    Functionality exists to allow developers to move posts to review queues (that's how we test). My first reaction to making that bit of UI available to moderators is "why didn't we do that yet?" - I'm going to take a look at it and make sure there's nothing horrible that could happen if we exposed it (when something that useful isn't exposed, there's very often a very subtle but important reason why not) and see about getting it done.
    – user50049
    Oct 18, 2016 at 4:47
  • @TimPost Cool, thanks!
    – Raphael
    Oct 18, 2016 at 9:17
  • Would be great if the mod could also add a comment that is only visible in the review queue. Oct 18, 2016 at 14:08
  • 2
    @IanRingrose I don't think I've ever felt the need to post a comment that only reviewers would be able to read. Our moderation decisions should be as transparent as possible, so let the users read it!
    – Raphael
    Oct 18, 2016 at 15:06
  • @Raphael, it could also be a comment you want the reviewers to read, when there are 101 other comments on the item. Oct 18, 2016 at 15:19
  • That's a different issue with different possible implementations (highlight comments by mods and flags; special message for this purpose; clean up comments before sending to queue). Still, I don't think the comment has to be hidden from everybody.
    – Raphael
    Oct 18, 2016 at 15:21
  • That's what the site's primary chat room is good for. Use it to develop a healthy community. Oct 20, 2016 at 5:44
  • @200_success Are you proposing to use chat as general-purpose review-"queue"? I don't think that's a good approach.
    – Raphael
    Oct 20, 2016 at 8:51
  • Why wouldn't you use chat to solicit opinions on how to handle problematic questions? Especially if the chat room subscribes to the questions feed, it's perfectly natural to "reply" to posts there. ("I'm going to close this one as off-topic— second opinions, anyone?" "Can anyone figure out what this guy is asking?") As long as it's done with a respectful tone (no brigading or shaming), it's fine. Oct 20, 2016 at 11:43
  • 1
    @200_success I could do that but it's a lot less democratic than moving it into a queue. I don't know about other sites, but we have far more reviewers than chatters. Also, I want to get more people into the queues, not fewer. I would say chat is a decent workaround, but not a replacement for being able to move stuff to review.
    – Raphael
    Oct 20, 2016 at 12:14
  • 1
    @TimPost Any news?
    – Raphael
    Apr 24, 2017 at 15:05
34

Give new users better guidance on how to answer

The Help Center on each site has guidance about what to ask and what not to ask, but no obvious guidance about what makes a good answer. (There's some tucked away in places people won't think to look, like the help about flags. That doesn't count.)

On many sites the biggest new-user problem isn't bad questions but bad answers. People see a question, see a textbox, and type their personal opinion into it, or tell a related story, or ask another question. Yes, people will probably flag them and they'll go to the review queue, but that's later. It would be better if we could get out in front of it.

We should develop a template "how to answer" topic for the Help Center and allow sites to edit it for their own particular rules. A new user (first answer? rep below a certain threshold?) should see this information when he starts to answer a question, along with the other guidance that's given.

Let's help newcomers meet answer guidelines from the outset, rather than finding out only later when the downvotes and comments start rolling in.

(By the way, while we're talking about the Help Center, please fix this: Why is "what can I ask about here" not the first entry in the help center?)

2
34

Crossover questions!

My pet popular-but-ignored feature request is "Crossover Questions".

I realize that this would probably be very, very far from trivial to implement, but there really is enough overlap between many of the SE sites that I think this would be pretty useful.

While I do not think that crossover-questions would replace question-migration (since some questions really are off topic on the original SE to which they're posted but on topic at a different one), I do think that since the question-migration feature is currently the only feature that provides some sort of cross-site per-question "sharing", the lack of the proposed crossover-question feature is probably not entirely unrelated to the fact that the way we think about the purpose and motivation of question-migration is currently a hot topic.


NOTE: The top-voted answer on the linked meta-question fleshes out the proposal and is very worth reading. In particular, it discusses how questions will be nominated and accetped as crossover questions. It also discusses how voting, moderation, and so on should work.

18
  • 3
    Similar request.
    – Raphael
    Oct 17, 2016 at 23:05
  • 7
    It's common to have the same question on different sites, such as Why does Kylo Ren wear a mask? on both Science Fiction & Fantasy and Movies. Crossover questions would be a nice way to merge the information instead of having it fragmented between two sites. Oct 18, 2016 at 23:11
  • Sometimes the same question fits multiple sites but the answers will be different, so this proposal needs to account for that. We sometimes see the same question on more than one of the religion sites, but answers that are appropriated on one would be offensive on another. So SFF/Movies is at one end of the spectrum, religion sites at the other, and I don't have a good sense of the space in between. Would the same SQL Q&A work on both SO and DBA? The same webserver Q&A on both Webmasters and Server Fault? How many workable site pairs are there? Oct 20, 2016 at 13:03
  • @MonicaCellio Are you thinking of particular religious questions? I'd be interested in an example. I would think that the implicit grounding for religion SE questions is an assumed "What do adherents of religion X believe about..." even if that's not explicitly stated in every question. If so, then the questions actually aren't the same. For non-faith-based sites I'm not sure why there would be any question about whether the same question is "really" the same or not; in fact, database and webserver questions seem like excellent crossover question topics. Oct 20, 2016 at 14:10
  • @KyleStrand "What does X in (some book in the prophets) mean?" "Oh, that's about Jesus". Even with explicit religious scope we get those kinds of answers on Mi Yodeya at times from drive-by Christians & Muslims. If the question were somehow marked as multi-site it would be even worse. We don't make people phrase those questions as "according to $religion what does X mean" because the "according to $religion" part is implicit based on the site you asked on. I asked about databases and web servers because I don't know if those sites also have implicit scoping not obvious from outside. Oct 20, 2016 at 14:30
  • @MonicaCellio One of the important aspects of the proposal in the top-voted answer on the linked meta question is that making a cross-site question would require the users at the receiving site(s) to nominate and vote for potential cross-site questions. As long as both sites are involved in the decision, I don't think it would be too easy to abuse. Oct 20, 2016 at 14:57
  • @KyleStrand if all sites involved agree then that's different. It still has challenges (different editorial norms, voting, comments, moderation...), but that would address the one I brought up here. (Err, with that clarification, if you were to edit this I could remove my DV.) Oct 20, 2016 at 15:30
  • @MonicaCellio Thanks for your feedback! I've gone ahead and added a couple sentences mentioning that the top-rated answer addresses some of the complications with the proposal. Oct 20, 2016 at 18:30
  • 1
    Would totally support that, since to many sites partially or even completely overlaps with others. Questions on SharePoint are mostly on-topic on StackOverflow, but the people there will get almost no recognition of their efforts due to the low traffic. Sites like Science&Fiction, Movie, Anime and Arqade have questions that would probably benefit from being cross-posted on multiple sites (a question about the plot of a game based on a sci-fi anime is on topic on at least three sites), and those are only some examples.
    – SPArcheon
    Oct 22, 2016 at 18:20
  • I'm still trying to figure out how this would improve the quality of SE content. Isn't this the very reason why migrations exists?
    – Braiam
    Oct 27, 2016 at 13:53
  • @Braiam I don't think so. Migration is for questions that explicitly don't belong on the original site. This proposal is for questions that do belong on multiple sites. Examples are given on the proposal itself, but here's another one: askubuntu.com/q/694031/236343 Oct 27, 2016 at 16:18
  • That's why I fail to see it. For me migration is for questions were they are best answered, and have a swat of users able to curate the content as it ages.
    – Braiam
    Oct 27, 2016 at 17:08
  • @Braiam I'm not sure I understand what you're trying to say. It sounds like you're saying that for every question, there is exactly one site where it is "best answered", and it also sounds like you're assuming that cross-site questions would somehow be un-curated. Am I understanding you correctly on either of these points? If so, I disagree that all questions are necessarily best-answered on a single site (based on the examples shown) and am not sure why you would assume that this proposal somehow limits the curation capabilities of the communities involved. Oct 27, 2016 at 17:51
  • It's because there is actually one person that is able to provide the best answer, and it usually only hangs out in a single site, the site where they can use their knowledge at its fullest.
    – Braiam
    Oct 27, 2016 at 19:54
  • @Braiam I think I'm going to have to respectfully disagree. I don't see anything to support your observation, and my experience has clearly been different. Oct 27, 2016 at 20:54