-8

This is a feature request that addresses the problem (at least on Math.SE) that low-quality questions are often quick-shot answered while the closing votes are still piling up. This often makes several quality problems out of one, given that there has been so little time for composing an answer.

To address such problems, simply block early answers. In detail:

  • for the first, say, 20 minutes (configurable per site) that a question is publicly visible (outside special queues), no answers to it can be submitted. Even at the API level, the server rejects attempts to post answers. Everything else, be it voting, flagging, comments, edits, or moderation, is not hindered.
  • During that period, the "post an answer" button on the question page gets replaced with a message such as

    This question is recent; it cannot receive answers yet. You can vote on it, or help improve the question by commenting or editing where appropriate. You can also prepare an answer to be posted if the question gets opened.

  • No countdown-to-acceptance is displayed, because such a thing would suggest that speed is of the essence, which is against the goal of setting the focus on quality.

  • The editor functionality (textarea plus preview) for entering an answer is left intact. Addendum: This includes the auto-saving of drafts, so users can still prepare and preview an answer and return later.

Note that answers, as opposed to questions, already have this property, even permanently: You can vote on them, and improve them with comments or edits, but you cannot answer an answer, and that is a good thing here. Basically, this drives users to rate the content that is already there, instead of adding to the existing chaos.

The intended effects of this feature are

  • More attention to the question itself, and more focus on its quality
  • More time to close bad questions without collateral damage in the form of already-posted answers
  • More time for early spotters to compose a decent answer, knowing that they cannot be outrun by the competition as long as the block is in effect.
  • Nice side effect: Real-time support for exam questions is slowed down.

Tuning the duration of the early answer block period may require experimentation. In the range of small values, an increase of the early block duration should correlate directly with an increase in quality. For longer durations, perhaps 30 minutes and above, that correlation should disappear.

Related topics: On Meta.Math.SE, there is a discussion about rate-limiting answerers. I think that an early answer block might be more effective and less intrusive.

  • 2
    Would this be for all questions or only questions posted by low reputation/new users or first time askers or... ? If the former, why? Is there some point at which we can trust users to ask good questions without putting their question immediately in limbo? – Catija Jun 11 '18 at 19:28
  • @Catija: All users. This is about quality. Quality needs time. For everyone. – ccorn Jun 11 '18 at 19:29
  • 2
    i mean... the best time for a question to receive answers is while people are looking at it the most, which is within the first 10-15 minutes. Why not allow answers during that period? it doesn't take 10 minutes to answer most questions. – user400654 Jun 11 '18 at 19:29
  • Eh, @KevinB I can understand wanting to prevent FGITW answers to questions that may be only half-explained... heck, I've thought of something similar for sites I use that have strict question quality limitations to start the question in the "on hold" state until a couple of users review it and say "yep, looks good"... but I've only ever talked about doing it for new users' questions... most experienced users can be trusted to know the rules. It's not guaranteed but it seems less necessary to make them wait if they've shown they can do it well in the past. – Catija Jun 11 '18 at 19:32
  • @ccorn I'd recommend you to read about @Shog9's story to the end: chat.meta.stackexchange.com/transcript/message/7019855#7019855 Read all of the messages please! – πάντα ῥεῖ Jun 11 '18 at 19:36
  • @ccorn Have you considered proposing this on Math Meta instead of main meta? I know y'all have concern, particularly about homework questions that show no effort, so it might be an easier sell for a single site than for the network as a whole. Whether the Staff would implement it is still a question... but if you can work it through on the Math meta site, they may try it out on Math.SE and then shop it around to other sites that are interested in something similar. – Catija Jun 11 '18 at 19:37
  • @Catija: While configuration is site-wide, implementation is very probably network-wide. Sites that want to do without it can have the early block duration configured to zero. That might even be a default. – ccorn Jun 11 '18 at 19:43
  • That's not really my point. Based on the voting here, and maybe due to how you've explained what you're asking for, I doubt there will be any support for it to be implemented at all. Feel free to clarify or explain better if you think it will improve things... but if Math.SE finds this really valuable and really wants it and sells it really well on that site, the staff may be more willing to try it than they would a negatively-scoring feature request here. Does that make sense? – Catija Jun 11 '18 at 19:45
  • I'd be viewing questions I can't answer. I have time now; I won't in 20 minutes. Then I'll want an option to not show me questions that I won't be able to answer right now. – 1201ProgramAlarm Jun 11 '18 at 20:38
  • @ccorn - So as an experienced StackExchange user, and let's assume all my experience is from answering questions at Math.SE so I am aware of what is expected out of an answer, you expect me to wait 20 minutes if I know the answer to the question. Why would I ever come back to the question? If this is implemented, I will simply stop answering questions, since my time is clearly not appreciated. – Ramhound Jun 11 '18 at 21:45
  • About your paragraph "Note that answers...": I don't think that questions and answers can be compared that way. Answers aren't meant to be answered, so allowing to answer them wouldn't make much sense. On the other hand, the whole purpose of asking a question is to get answers, so it's natural to allow answers to questions. A question without answers isn't usually very useful, but an answer which itself isn't answered can be very useful. Note that I don't necessarily think that your feature request as a whole is a bad idea, I just don't think that paragraph makes much sense. – Donald Duck Jun 11 '18 at 22:49
  • @Ramhound: It is indeed already the case that your time for reading the question and voting accordingly is not appreciated. (You don't get rep for voting. At most, you get a badge or two.) That explains why users want to answer. But it also explains that there must be some mechanism that ensures that questions are properly rated. Currently, voting on questions does not affect you at all. However, introducing incentives for voting per se would probably have nasty side effects, so I'd rather have a temporary block on answers until the question has been exposed enough. – ccorn Jun 11 '18 at 22:55
  • @DonaldDuck: If answers were just labelled responses, then responding to responses would make perfect sense. Clearly, that is not what we want. A noteworthy effect is however that answers attract votes, and I suppose that that is the more so because users cannot leave a more worded imprint (and gain rep in the course). In that sense, the scheme already works for answers. I want to see it work for questions as well. – ccorn Jun 11 '18 at 23:06
  • @1201ProgramAlarm: Note that the editor is left intact. It is already the case that your drafts are saved automatically, so when you later return to the question, your answer draft is already awaiting you, and you just need to click the newly appeared "Post your answer" button. (Plus, for answers that take longer, or to questions that you just have not spotted immediately, chances are that the block has been lifted by the time your answer is complete.) – ccorn Jun 11 '18 at 23:41
6

I've half-joked about a feature like this for at least six months, maybe almost a year... that said, my idea was always slightly different

  • It was limited to first questions and/or low-reputation users (under 200 rep or so)
  • It was a queue where the question would go before ever being posted rather than a live question.

To explain my thinking, I want to talk about why these questions (and their answers) are troublesome.

Many sites (if not all of them) have quality requirements for questions. We require questions show some effort at finding a solution, explain what they've already tried, and describe the problem in sufficient detail to be answered well. Some sites may have specific additional requirements for how to ask a question or ways of asking that are considered primarily opinion based rather than objective questions.

Many sites also have users who are more than happy to answer questions that don't meet these quality guidelines. Considering that I'm not a user of Math.SE and I'm still aware of their discussion about homework questions, I'm guessing that one specific example for that site is that such homework questions must show more effort than the asker simply posting the problem out of their assignment and asking for the answer.

Another example I'm very familiar with are the rules surrounding proofreading on English Language Learners. Questions asking "is the grammar in this sentence/paragraph/essay/novel correct" with no specific point of concern are closed as proofreading.

So, why is it a problem when these questions are answered before being closed?

Because all that many users care about is getting an answer, not about whether their post is closed or downvoted or even deleted. Shog even said that earlier in a somewhat unrelated discussion in the Tavern (yeah, I'm taking this totally out of context, but it's still true):

new askers want an answer to their question. If they get it, generally they're happy. If they don't, generally they're not. Downvotes have some small effect, but... It's probably about the same as the color of the bikeshed on the attentiveness of the safety monitoring team at the power plant.

I've seen it happen several times that users will ask a question, be told in comments that their question fails to meet some guideline and be asked for explanations or other improvements but, in the time it takes to close the question, it's answered. This completely invalidates and undercuts the users asking for more details and tells the asker that they can ask their question however they like and still get an answer. If the question is later deleted - they don't care. They have what they came for.

So, what happens? Well, sometimes someone will choose to complain at the person who answered the question, commenting that it's a type of question that is not acceptable on the site, maybe linking to a meta post about it... the answerer may then delete their answer or, if it's accepted or they don't care, they'll ignore the comment. More often, nothing will happen including the OP editing their question to make it better for the site.

The only real carrot we have to encourage improvement of questions is the question getting answered and we have no sticks for users who choose to prematurely answer questions. I've actually seen people say "I'll do better next time but can you please answer my question now?" or something similar rather than being willing to edit their existing question!

This is not to say that what amount of detail is sufficient for an answer isn't somewhat subjective - it is! And that's completely fine. What we do need to keep in mind is that more detailed questions tend towards more detailed answers because we have a better understanding of what the asker is trying to do and why they're failing. It also means that searchers are more likely to find similar questions when there are more keywords in either the question or the answer that will match their own problem - possibly helping with duplicate questions.


If our goal is to have high quality questions and answers, we definitely should at least think about whether improving them before they're live on the network is a viable solution.

Some sites have a "Sandbox" (here's the one on Worldbuilding). While the one on MSE is used for testing (and occasionally breaking) the site, other sites use them for question vetting before posting. They're a place for users to ask

  • Does my question make sense?
  • Is it clear?
  • What can I add to make it better before I post it as a question?

One site even has the 5 reputation restriction lifted so that all users can post on meta to get help writing a good question.

The problem with this method is finding it... knowing that Meta exists at all is difficult, particularly for new users, let alone finding one post out of hundreds or thousands. I'm guessing they're directed to the site's sandbox after they ask a first question that fails to meet site guidelines but wouldn't it be at least worth testing what would happen if we put them there first - by default?

Even on Stack Overflow there have been many attempts to do something like this for example, the question mentorship process that was tested in 2017. So, the question is, how can we help users, particularly newer ones, write better questions before people try to answer them - or yell at them for posting bad questions?

A review-type queue for questions before they are posted could be a possible solution.

This would have a few benefits:

  1. This queue would act as a speed bump. New users would be given specific guidance for how to improve their question and only people willing to help them would (hopefully) be active in that queue.
    This takes some of the weight of bad question content off the users who are unwilling or tired of helping to improve new questions and possibly keep the site neater and less stressful since there would be fewer low-quality questions showing up in the new question list. It could/would also filter out spam questions and questions that are blatantly off topic.
  2. This could also be an option for all users. A tick-box on the Ask Question page could let users specifically request their question to be put into this queue.
    This would make it possible for sites with sandboxes to go to a more findable and official format that makes it easier for all users to get improvement suggestions for their questions before making them live on the site and risking downvotes or closure.
  3. {WILD IDEA} Questions migrated between sites could be put in this queue automatically before being completely migrated. While in this queue, the question would be "on hold" on the originating site (not moved) and, if the target site accepts it, it's actually migrated. Otherwise, it's declined and stays where it was.
    This gives users on the target site the chance to improve (or at least vet) the question before it's plopped on their front page.
  4. Users who have asked questions before (and had them well-received) will skip the queue and be posted immediately on the site - perhaps it's a 200 reputation privilege so that the association bonus won't allow skipping it. Having a previous question (or several) closed/deleted might trigger reversion to the queue before asking in some situations.
    This reduces the headache on high-reputation users who should know better but doesn't exempt them from the possibility of asking low-quality questions.

I think that something along these lines could be - at least - interesting to try. I grant that it'd probably be somewhat difficult to implement but it has the possible benefits of increasing question quality, the opportunity to educate new users without downvotes or close votes or delete votes, chance to reduce user fatigue from seeing an endless onslaught of poor questions... and maybe even an interesting solution to migration issues.

There are some hurdles that may need to be overcome. While most sites could probably get through the volume relatively quickly, Stack Overflow may end up with an extreme backlog (close vote queue count anyone?), so this solution may not work for SO - or it may need some fine tuning to work well. But on many sites, it may help quite a bit.

  • Nice answer. The problem I see is that queues should be reserved for exceptional cases -- they just don't work well with massive inflow. Therefore I propose to expose the question to the full-size crowd, but restrict the ability to skip over the review straight to answering immediately. – ccorn Jun 11 '18 at 22:41
  • Isn't what you have described in (the end of) your post somewhat similar to triage and help & improvement? – Martin Jun 12 '18 at 8:00
  • Since the OP was probably motivated by problems on Mathematics Stack Exchange, I will point out that recently there was some discussion about this also on the local meta of this site. I will link to my post What are Triage and Help and Improvement review queue? - where I have also tried to collect some links to other recent discussions related to this idea. – Martin Jun 12 '18 at 8:01
  • @Martin Thanks for that. The flow chart does clarify that to a degree. One thing that I think that differs is that triage (as far as I can tell) doesn't prevent answers from being posted. – Catija Jun 12 '18 at 14:36
  • For what it's worth, in July when TeamDAG finally gets to the question improvement process, they are gonna start focusing on new user question wizards, which may help a lot. See meta.stackoverflow.com/questions/358600/… and meta.stackoverflow.com/questions/363051/… These are Stack Overflow-specific, but I can't imagine they would be kept there after a successful test run. This would help address the root cause of the OP's complaint rather than trying to fix it by blood letting/guessing. – TylerH Jun 12 '18 at 16:24
  • That would be helpful... @TylerH if they get used. If they're not used, I worry that the tempers towards users who clearly haven't used them will increase in a big way. It'd also be something that has to be developed for each site individually, which will take time. – Catija Jun 12 '18 at 16:40
1

Experts' time is valuable. They click on answers that pique their interest. If you prevent them from providing an answer when they are there, reading the question, and then don't even tell them when to come back in order to try again, you are effectively sending the signal "we don't want your contribution, please stop trying to answer".

Off-topic questions receiving answers is a common problem but it is not a severe one; many questions, once closed, will be cleaned up by the system automatically. Those that aren't can still be deleted manually by moderators and users with delete-vote privileges (of which there are currently 718). That should be plenty of people to handle closing and deleting even a third of the ~500 questions a day that Math.SE receives.

  • 3
    Users can only delete downvoted answers... so that doesn't necessarily make it a solution... and I'd avoid using the term "off-topic"... many of these questions are probably perfectly on topic but simply aren't sufficiently explained to garner good answers. – Catija Jun 11 '18 at 20:01
  • The time of the experts who vote to close is also valuable. The signal sent is: First, we want your opinion on the question, and for that you get a sneak peek of what is coming up. Nothing bad there. – ccorn Jun 11 '18 at 20:21
  • @Catija I'm talking about deleting closed questions, which will delete the answers, too. – TylerH Jun 11 '18 at 20:40
  • @ccorn Voting to close takes two seconds, writing an answer takes orders of magnitude longer; the two are not comparable in terms of time. – TylerH Jun 11 '18 at 20:41
  • Is it a realistic assumption that there are (1) representative users who (2) find a question not worth returning to and yet (3) can produce a quality answer in less than the time it takes the question to gather a decent amount of votes? – ccorn Jun 11 '18 at 20:53
  • 1
    @ccorn I think so. It's such a silly thing to prevent someone from answering because they "saw the question too soon" that if I ever saw such a thing I would just close the tab and not bother trying to answer it again. What you should do is edit your request above to show us some data on how this is actually a problem for Math.SE. Show us how many questions each day are getting answers while close votes are accruing and/or that get closed shortly after an answer. – TylerH Jun 11 '18 at 21:21
  • The question is not being "seen too soon". It is seen because it requires votes. That's not silly. It's actually more natural than allowing answers while the question aspects are still in flow. – ccorn Jun 11 '18 at 22:12
  • @ccorn Question visibility is not maximized so that people close vote it. Question visibility is maximized so that they can be seen and answered. It seems like you have a different perspective on what Stack Exchange sites are for than what the organization intended. Also, the default position is not "the question is in flow", the default position is "the question is posted". A question that is not ready to be answered should not be posted yet. If it needs revisions, fine, but you should not assume questions, when first posted, are or should somehow be "in flow" or "not ready for answering". – TylerH Jun 12 '18 at 13:44
0

Regarding the FGITW problem, I don't see any reasons why this should be restricted or throttled.

Users come here to get answers, if these are appropriate and helpful they'll gain reputation, no matter of the time taken to write them was short or not.

Even with obvious duplicates, these might be considered helpful, so I don't see any real reason for restriction.

In fact, I answered questions myself, while dupe hammering them at the same time within minutes (seconds).
My answer might still carry value to explain about the duplicate and clarify the situation.

If those answers aren't considered being useful in the long term, those would be downvoted and probably removed in the longer term anyways.
So I can't spot any harm regarding the overall quality.

-2

A relatively well known fact, that preliminary moderation has catastrophic effect to the new user activity. The user should see, that what he wrote, is on the site, on the spot as they posted it.

The MathSE is a very friendly community, held together by the love of the Math. This is highly atypical. Despite how are they looking, yes they take care the VLQ content as it should be. This resulted its growth and wonderful stats, too.

The MathSE is a flagship of the Stack Exchange. It is because of the other side of your dislike. Because it is a friendly community where everybody loves math. This is unique between the big SE sites. The MathSE is imho a diamond of the network, yes with its admittedly higher crap rate of new posts. Because the people there is not deterred by the moderation cruelty. Some smaller sites with similar mentality are, for example, the Retrocomputing SE.

About the realtime support for exam questions: I agree that making the site an abettor in academical cheating should be handled on the possible hardest way. What I can't see that how could the MathSE check, who is on an exam, and who is just curious. While any teacher could simply ban using phones/tables, they are even forbidden to take into the exam rooms.

  • New users (first posts) are moderated in the first posts queue already. The change would only affect established users giving quick answers soon after the question has been posted. Lemma (check with your experience): An answer that can be given quickly is an answer that can be given by many. Many later visitors can give such an answer en passant as well. But the early moments of a question should quickly decide whether more moderation is warranted or not. It follows that the most needed contribution from the quick onlooker is not an answer, but a judgement on the question. – ccorn Jun 16 '18 at 5:35
  • @ccom You are right, I will delete this answer. But I still think, the MathSE is very wonderful even because its freedom, and the current system is enough to kill LQ posts. I think your suggestion is equivalent of that answers will be invisible 20 minutes after posting the question. I think it will lead to more essentially same answers; but not for sure a bad idea. I see also that strong, dark powers want currently to kill the freedom of the MathSE and I find it frightening. Not the MathSE should be changed to be lesser free, the other sites should be changed into the MathSE direction... – peterh - Reinstate Monica Jun 19 '18 at 5:18
  • @ccom Currently, there is often a race for being the first answerers below new questions, it goes clearly against quality... but the MSE is still diamond :-( – peterh - Reinstate Monica Jun 19 '18 at 5:21
  • @ccom Btw, the voters of the MSE are typically favorizing the more restrictions, the more downs, the lesser freedom. Thus, they should have voted your initiative up. The reason of the downs were that you had the spine to suggest something. Unfortunately, they seem lacking the knowledge, that then they should have given a rejective answer and they should have voted that answer up. I did not vote your question because I condemn their this practice and won't be one of them. – peterh - Reinstate Monica Jun 19 '18 at 5:27
  • Better keep this answer; it reflects an important sentiment. The strong dark powers are neither strong nor dark I'm afraid; those are experienced users confronted with the 1000st PSQ and having to deal with it as if it were the first one, without sign of things getting better. To me, that simply points to the need for a workflow that avoids some of the subjectively worst conflicts. This proposal is about such a workflow. As to invisibility: We have full visibility, and IMHO that is the only way to attract votes on the question soon enough. – ccorn Jun 19 '18 at 8:01
  • @ccorn Sorry what is a "PSQ"? – peterh - Reinstate Monica Jun 19 '18 at 8:56
  • PSQ = problem statement question. Copy-pasted, no own text, no own thought. Early answers prevent the asker from actually learning anything. – ccorn Jun 19 '18 at 9:23
  • @ccorn Also the answerers need interesting problems. The worst sin what most SE site commits: they increase the quality requirement for new questions to the top. With it, they expel not only the VLQ first-posters, but they expel also the answerers. If a site is ruled by such people, it is much worse as if it has too many crap (particularly if the crap is handled roughly fairly even after it). – peterh - Reinstate Monica Jun 19 '18 at 10:59
  • It's not about ruling a site (which would be bad); it is about getting a fair chance to handle the egregious stuff. Currently, both LQ askers and early answerers do not need to care; in fact the system rewards them for not caring. With a configurable early answer block duration, the system can mill on as before, except that focus is unconditionally forced on the question itself during that time, thus more people will actually care about that. The block will be lifted automatically and unconditionally, and then everything is back to how we know it. – ccorn Jun 19 '18 at 11:50
  • @ccorn Against your reasoning stays my experience: I've seen too much sites already, which were essentially expropriated by a small minority in the name of "keep it HQ". They expelled the askers, and they've expelled also the people who could have (and wanted to) answer their question. This mechanism made the growth of the SO stagnating since 2014. The Stack Exchange is also a game, but most sites are ruled by a narrow minority, a quasi-criminal gang, wanting exclusive right to play on the playground. The MathSE is an exception, and I feel myself there much better, in its current state. – peterh - Reinstate Monica Jun 19 '18 at 12:10
  • @ccom They always say the same: that in order to preserve the quality, they need to expel exactly those questions what the majority of the site members could answer. Somehow they never want to expel the questions what they can answer. No, it is long not about quality. It is about power. And it significantly worsens the enjoyability of the whole SE. Fighting braindamaged psychos playing the nice guy wanting only to "preserve quality" is a far lesser enjoyable game, as racing with others, who gets more rep today. Despite that the site quality is a little bit really worser (but not much). – peterh - Reinstate Monica Jun 19 '18 at 12:20
  • @ccom Try some other sites, too, particularly in the topics where you are not a professional, just a well-educated layman. Try to ask your questions, and answer others'. You will feel on the spot, what I am talking about. – peterh - Reinstate Monica Jun 19 '18 at 12:31
  • I'd agree with you if it were a proposal like "Reducing the required number of close votes", particularly if the given reason were to be more efficient. Such things reek of politics; we all know what's bad about it. (Therefore, the number of close votes must be high enough that clique ruling is hard.) Now note that this proposal does not convey permanent advantages to any side; and it deals with the problem in the most open way (no queues, no exceptions, no privileges). – ccorn Jun 19 '18 at 12:31

You must log in to answer this question.

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