Should we stop closing questions?

The primary reason for closing a question is to stop it from receiving new answers

… so the question can be improved without having to tip toe around these premature answers.

We're not supposed to edit a question in a way that invalidates answers, but if we allow premature, incomplete questions to receive answers then it's likely to make actually meaningfully improving the question impossible without violating that rule.

So no removing closure without finding a different solution to premature, 'shot in the dark' answers is not something I think we should do.

All sites need moderation.

The problem with SO's system isn't that questions are removed, but that questions are removed as slowly as possible, as publicly as possible, with as much drama as possible. Instead of having it instantly removed and taken elsewhere for potential question recovery.

Overall, public shaming as a moderator tool was a bad idea to begin with. Everything I've to say about the matter can be read here - Giving question feedback in private - a moderating system to reduce conflicts. A summary:

The main problems of SO's model:

• Humans often simply don’t take kindly even to constructive criticism, especially not when given in public for the world to see.

  The basics of leadership & keeping people motivated is to give praise loudly in public but to give criticism discreetly in private. This makes people far more likely to actually listen to the criticism and change.

  Solve this by removing the question from the public eye and then give private feedback to the poster.

• Deleting posts “as slowly as possible”. Bad questions get slowly ground down into dust by downvotes, comments, close votes, all in public, really rubbing it in. And even when it sits there with 5 close votes and -10 score, it is still published for everyone to see.

  Solve this by giving trusted users privileges to instantaneously remove a bad question from the public eye. This also minimizes friction as the question is moved away from those who haven't the slightest interest in helping new users.

• “Bandwagon moderation”. The first veteran user who encounters a bad question and is willing to help out, often gives constructive criticism with links to help pages etc. So far, so good - that initial polite comment is often all that’s actually needed. Yet we have subsequent users arriving later, piling on further comments or repeating what's already been said. It stops being constructive and derails into what the poster might interpret as “you are bad”. And it creates a negative atmosphere for everyone stumbling over that post too.

  Lots of such comments come from veteran users who are simply fed up by viewing the same endless flood of bad questions day after day. They actually don’t have much interest in helping the OP at all, they just want the crap question gone.

  Solve this by not forcing regular users to view bad content, again by quickly removing such questions away from the public eye to a “quarantine” area.

• SO's “crap hugging” policy of “we must preserve and publish all the crap ever posted and polish it until the end of time” is harmful. Similarly, when a question is closed since it can’t be answered and needs to be corrected by the OP alone, it is senseless to keep on displaying that question to the public.

  It is much more important for the community to reduce negative criticism and low quality content than to preserve some unsalvagable homework dump for all eternity.

• In addition, do not force users who just want to use the site to become moderators, by having a messy rep system that assumes that people with good domain knowledge automatically make good moderators as well. This simply isn’t true. A better reputation & moderator privilege system than the one at SO is needed.

Can this be fixed and implemented on SO? I don't think so, it's a lost cause. But a new site replacing SO could implement this and learn from SO's many mistakes.

I disagree with almost all of this. Closure needs to be improved, not removed.

Concern: Closing questions gets them deleted after some time.
And so will unanswered questions. If a question cannot receive answers, then the system will remove it automatically, too.

Roomba needs to stay. If a question hasn't been edited and reopened within a certain amount of time then clearly the community think it's not useful. Same goes for a question which isn't getting any interest (or actual disinterest). These should be deleted so we can focus on curating the other two and half million questions, no matter how old they are.

Concern: Closing prevents bad answers from new users.
Historically, mods never accepted flags about wrong/low-quality answers because we cannot judge their correctness and the volume would drown us. But 20k+ users have answer-delete votes and they can use them on poor answers. With limited number of flags, mods probably wouldn't mind an occasional flag about a totally rubbish answer.

The problem with bad answers in these case isn't that they're wrong like normal bad answers. It's that they often don't answer the question because they're answering a different question. The point of closure to prevent bad answers is to stop people misunderstanding the question and answering something which makes no sense.

So closure in many cases isn't helping the mods' workload, it's helping the users'. It's saves people from going to the effort of writing an answer to then find out that the question wasn't answerable in the first place or was actually something entirely different.

I actually think there should be a Wrong Answer flag, but only for highly upvoted answers which are blatantly and misleadingly wrong, or the circumstances/product have changed. These should not be handled by the mods, but by gold tag holders, who are the subject matter experts.

Concern: A closed question offers guidance to the user on how to improve the question.
People who cared and asked the question to the best of their effort and still got it closed, probably wouldn't be able to edit it even with the guidance to get it reopened. Those who didn't put enough effort, probably couldn't be bothered to read and apply the instructions. We can offer personalized guidance to the askers now that the question volume is down. If something can be improved in the question, we can just say what in the comments.

Telling the questioner what the issue is and giving personalized guidance in comments has always been possible, and we are capable of giving detailed guidance how to improve their question. Either they listen or they don't. It's not true in many cases that they can't edit it to get it reopened, they just give up because they can't be bothered to follow the rules. So the first type of people we should help to improve the question before it gets answered, the second type should just go away.

The solution for this one would be to make much more and more specific closure options, and make closure reasons for specific tags. For example, the sql tag could have a reason "Needs Query Plan" or "Needs Table/Index Definitions". An ef-core question could have "Needs Model Definitions". And a regex question could have "Needs Input Text". We can also have non-tag-specific things like "Needs to Define All Edge Cases", for those please-split-first-names-and-surnames questions.

To me, this reads a lot like: "should we stop fixing broken windows?"

It would be nice to not have to put so much effort into closing questions, so that attention can be focused on improving the existing collection. The obvious fix is a reversed posture — require questions to be opened from a default-closed state. This is what the Staging Ground fundamentally accomplishes, and it solves a lot of social problems: the vetting can feel like a conversation with auditors rather than putting the question trial on the front page; and the question is not continuously subject to downvotes (since there's no motivation to downrate the question as an emergency signal while waiting for other close votes); there are stock message templates provided and an explicit interface communicating the process. In this model, the conventional closure process is an emergency backup for those new questions, when someone opens a question inappropriately. And it only takes one user (not the OP) to "publish" the question, so it's not as painful as the reopening process.

However, there's also a serious misalignment of incentives here: people who have an answer in mind for the question are incentivized to open it without heed to policy, while people who are concerned with policy have their hands full with existing Q&A curation.

In a general sense... Yes, I think we should stop closing questions.

But I don't want to see closure just disappear, rather, I want to see it re-invented. I want the closure reasons to become statuses we apply to a post that indicate what the user can do or needs to do to improve their post in a way that doesn't get in the way of answering. Maybe then, down the line, posts that aren't positively answered that have these statuses can be deleted at a lower threshold/sooner than those that don't.

I think there's roughly 3 types of closure:

• Question that are completely unsuitable for the site because it's in the wrong language or is about some off-topic subject like gardening. These questions are expected to be closed and never re-opened. We should probably keep the ability to close these (unless you want them to be handled entirely by mod flags?). Personally, I think the new "advice" questions should also be close-able like this.

  @Dharman has a comment on this saying

  An off-topic question is one where we cannot provide an answer because it's out of our expertise area. Whether it's closed or not, chances are it's not going to get a good answer here.

  I disagree here - mainly because people will answer off-topic questions and that kind of encourages others. Dropping this type of closure essentially makes all subjects on-topic.

• Questions are considered unanswerable because they need an MRE or something similar. The hope is that these will be improved and reopened, however if those are closed then it can be a frustrating experience getting them re-opened.

  I also think people are sometimes too insistent on an MRE and in a lot of cases it is possible for a subject expert to answer as-is. (I'd argue this also applies to the staging ground to an extent and maybe we're overly strict about questions needing to be "perfect" to graduate.)

  I'd be OK with these being left open to be answered or age into non-existence based on their merits. It removes the frustration of trying to get them reopened, and if they are actually answerable then they'll get answered.

• Duplicates. Again can be a frustrating experience if the asker believes that the question is not a duplicate (either correctly or incorrectly).

  I'm less clear on what to do about these - they fall somewhere between the two other cases. I think that accurate closure of duplicates is a good thing but also that it can be frustrating.

I always operated on the belief that it's way more discouraging to get your question looked at but not voted on, commented on or answered by anyone, than it getting closed.

When your question gets closed, it means the community is encouraging you to edit it so it can get an answer, or encouraging you to seek help where you will actually find it. It legitimizes your attempt to salvage a question and bring it back up the queue, whereas editing a question for no clear reason other than the fact nobody touched it feels like abusing the system.

I think reinforcing that sentiment would be better than leaving more questions opened only for them to never get answered.

I don't think we need to fundamentally change the rules here. But we are at less than 5% question volume compared to the old days. We have completely different problems now than when most of the rules were designed. So I do think we should carefully reconsider many rules, including when we close questions.

We should focus on closing questions that have fundamental problem that prevent good answers. But question volume is low enough now that SO could do the same thing as smaller sites have often done, and just try to get the asker to improve the question before voting to close. Theoretically closing is reversible, but in practice that is not something new users understand nor will they actually do all the right steps to trigger a review to reopen.

It also might be worth it to discuss the topic boundaries of SO, and what kind of subjective questions could work on SO. The latter is something the company is experimenting with anyway, but I don't think they're doing a good job here. I do think that there are a lot of potential good questions that aren't in the usual "here's some code, how do I fix this" pattern that currently have a hard time on SO or just get closed immediately. It's tricky to figure out good boundaries here, but a lot of the issues we had here in the old days stemmed from the large volume of posts on SO, and we don't have that anymore.

For as long as the site existed, users have complained that the most toxic part of Stack Overflow is the question closure. We have defended it saying that it is crucial to maintaining high quality of questions and answers on the site.

This is not true. No one (to my knowledge) has ever defended Stack Overflow's model by conceding that closing questions is toxic or by saying that being toxic is a cost of doing business. Instead they attempted to explain how curating a high-quality resource works. That these attempts usually fell on deaf ears does not mean such a model is "toxic".

Actual toxic behavior has been banned from the site for most if not all of its existence and users, moderators, and employees have always encouraged users to flag any toxic behavior they see.

But if closing questions doesn't help in keeping the site clean anymore then it remains only to punish users.

Faulty premise; closing questions does help keep the site clean. Also, closing questions doesn't 'punish' users and never has.

Concern: Closing questions gets them deleted after some time. And so will unanswered questions. If a question cannot receive answers, then the system will remove it automatically, too.

Also incorrect. Closing questions does not necessarily get them deleted after some time. There are over 1 million closed questions today. Likewise the system does not remove all questions with no answers. Both upvoted questions with no answers, as well as questions with 2+ comments, are never automatically removed by the system.

There's a myriad of reasons why I think this specific idea here would not be an improvement. Yet, going through the motions of reciting and countering and counter-countering them all has been done to death already. However, I'll take this opportunity to say something that needs to be said to all these ideas lately...

Stop Frankenstein'ing a Dead Horse

What you describe just isn't Stack Overflow.

Outside of the Meta SO bubble, and outside of the Hate SO bubble, SO is appreciated for exactly what it is. That includes double-edged features that lead to perceived toxicity on one hand but also the high quality on the other. And the high signal-to-noise that this brings is still why the silent majority values SO.

That there is something like SO – with this and that radically changed – that might work better these days frankly does not matter for SO itself. When cars gained traction and started competing against horses, you would not have brought your trusted old crock to the chop shop to replace his legs with wheels. When airplanes took off, you would not have bolted wings to your ride.*

There is still some mileage left in that nag as it is. Let those that still find value in this accompany it on its last chores.
Let the poor thing go out with peace, grace, style, and all the legs it was meant to have.

You are not here to serve SO|SE [Inc]

What you describe won't save Stack Overflow.

At least not the SO why we were all here. The SO that is not The Brand, or The Company Formerly Known as SE Inc but now SO or maybe Stack or Whatever, or StAIck Overflow, or ….
No, the SO that had actual experts (almost all gone), and enthusiast programmers (GIMME! is not enthusiasm), and desirable knowledge (poor, poor data dumps), and a high signal to noise (noise! Noise! N̶̤̂ō̷̹̗i̵͚̚s̴̡̛͂e̶̖͙͛͝!), and whathaveyou. What is left of that is barely enough to limp on. It's definitely not enough to go for a sprint in even a vaguely different direction.

What you maybe can save is SO the Prosus investment. What you maybe can save is SO the symbol of toxicity. What you maybe can save is SO the DNS entry.
And I really do not see why you would want to save those unless you are paid for it.

So please – please, please, please – remember that you here because you are all competent, and diligent, and motivated people. You are here because SO/SE served you.
You can do great stuff. You can do useful stuff. Don't waste this on trying to revive a dead horse by chopping off its legs and bolting airplane wings on it.

*I realise that some people did and do. Just look at the result and let's be done with that.

The question's title Should we stop closing questions? is click-bait to me.

The actual question is hidden way below and I'm very unsure if everyone who answered has read it:

Do you think the current system of closing new questions is still useful? Do you think it should be reworked?

As such, the answers and votes refer to substantially different interpretations of the question. I think the interpretations can be categorized like this:

1. Stop closing questions, no other changes
2. Change the mechanism, keep the scope
3. Keep the mechanism, change the scope

"Mechanism" meaning how questions get closed, "scope" meaning which questions get closed. Above is basically a matrix of the two orthogonal ideas.

1. Simply Stop Closing

I interpret the votes on cafce's answer as a pretty clear "no".

Charlieface's answer goes in a similar direction.

2. Change Mechanism

• Lundin suggests to reduce public humilation.
• user400654 suggests to make the reasons more useful.

Answers largely depend on which mechanism will replace it, this post may collect a few ideas. There are a few more upvoted posts in this category.

3. Change Scope

• DavidW's answer does some categorization and suggests a little more leeway on answerable questions.
• Mad Scientist suggests to rework the rules, learn from smaller sites.

In general, posts in this direction seem to receive fewer votes, some are even downvoted, e.g. ravenspoint's idea. Seems like there's no idea on a scope change that meets the masses taste currently.

Closing Notes

The community answer to: Should we stop closing questions? seems to be "no".

However, for reworking closure there are ideas around, and have been around for a long time. It seems we're more comfortable with changing the mechanism, than the scope. But then we need the company to conduct the implementation.

IMO, start with making the closure process, especially of duplicates, more friendly: Make Stack Overflow more friendly: remove the stigma of duplicates?

AFAIK, a lot on scope and mechanism has been discussed in the past, so a focused question that contains research of said past would make more sense in my eyes.

If I misinterpreted any answer or missed an important one, please let me know.

I think most closures still apply for the original mission of creating a Q+A resource but I think duplicate closures are the main gripe for most as evident by the comments any reddit post about StackOverflow will receive.

It would make sense to me that instead of closing a question as a duplicate, the answer on that duplicate that someone believes applies to the new post can be linked to as an answer on the new post, linking back to the original question and applying any gained reputation to the original poster.

Should we stop closing questions?

No, but we do need to tweak the closure process.

Some of the asked questions will inevitably be completely off topic (non programming related) for the site. We need to be able to close those and we need to close them fast.

But as far as programming related questions are concerned, current closure system is not working well.

Some of the problems come from rather narrow scope of what we consider on-topic questions. It is not just that we don't allow some kinds of questions, but also there is a tendency to interpret closure guidelines in extremely strict manner and many questions which would otherwise be answerable and could become a good knowledge base artifact get closed on technicality. Such closures are even worse when they happen on old, programming related questions with good answers which have clearly been helpful to many.

This is one of the main problems we have with closures. Even if we don't touch the scope, we do have a problem with overzealous moderation which at the end does more harm than good.

So I would like to ask curators to really think twice before casting a close vote, especially when it comes to older questions with seemingly good answers and in the area outside their expertise.

Before casting a close vote ask yourself whether you are fine with such question being deleted? Because all closed questions are eligible for deletion. Are you fine with those answers being deleted, too?

Many times I have heard from curators that they just want question to be closed, and that closure does not mean deletion. And while we do have plenty of closed questions which are not deleted, some of them do get deleted. And some of them shouldn't have been closed in the first place.

We do need to have quality controls and we do have to be able to get rid of the unproductive and unanswerable questions without long term potential. But we have to remember that questions are seeds here. Seeds which may yield great answers. I know that we would all like to have high quality questions, but let's be real. Not all questions are and will be stellar and we have absolutely great answers given on meh questions.

Another problem is that question ban is really extreme. Six month wait period is way too long. This made sense in times where we had more questions than we were able to deal with. Nowadays this is no longer true. We can deal with users asking few poor questions too many.

Staging Ground should have been a solution to that problem, but at the moment it failed to achieve its goal.

One of the solutions would be to fix and improve SG workflow, but if we lower the question ban time period, to a month or even less, then maybe SG would not be needed at all. Or at least it could stay as an opt-in feature for those users who really need and want help with drafting their questions.

If the SG stays, then those questions would have to be shown to all users, mixed with regular questions. They would also have to stay visible even after being reviewed because too many of them are not properly reviewed. Sometimes due to bad reviewers, but often reviewers who did the initial review are not subject matter experts and objectively cannot always properly determine whether question is improved enough to be published or not.

Being visible in the question list for all users would make them visible to subject matter experts who can either give more precise feedback, publish, and answer questions if they are good enough.

Besides allowing a tad broader scope of questions even under current rules, just by being slightly less strict when judging questions, there are still some programming related questions which would fall out of Stack Overflow scope.

For those we had opinion based questions experiment, but one of the major problems with that experiment is the threaded reply format where everything is treated as answer, even those replies which are meant to be comments.

Such format is also incompatible with regular Q/A and we have no ability to move questions between regular Q/A and opinion-based ones.

Now, there is a question whether we could change some of the closing rules and allow asking some questions which would otherwise be deemed unsuitable for Stack Overflow in regular Q/A. Another option would be changing the format of those questions to become compatible with regular Q/A which would allow us to move questions back and forth but keeping those in opinion-based section outside or current reputation system.

And of course, we would need a closure system for those kind of questions, too. Not everything asked there is programming related and mod flagging everything is not a good long term curation strategy.

No matter which way we go, one thing is clear. If nothing changes the future will be bleak.

Oh boy, this posts gave me pause, until I read the answers. We shouldn't lower the bar, we should actually be raising the bar. The bar has been so low for so long that we don't even know if we have a bar. Askers that don't even attempt to explain what they are trying to do, nor the issues they are facing is literally the butter and bread of the site which was supposed to get the asker to do as much as possible so that the answerers, which time is more valuable, only focuses on the practical ways to solve the problem.

Removing closure doesn't fix the problem, it just exacerbates it. Ludin is on track with a potential solution, something that I've been asking several times: we need to delete more and doing it faster. If nobody cared about any of the half a million questions with only 2 comments asking for clarification for a month, we as collective shouldn't care either.

A great deal of the frustration by new users comes not just by getting their question closed or marked as duplicate. The real frustration happens when non-SME vote to close a question or mark them as duplicate and it ain't helpful to the user and leave them with the issue on their own.

As an example: Flow Layout with customtkinter adds extra padding between widgets

The question was closed by non-SME's because they found it not reproducible and wasn't aware of the distinction between the related tech's, before I had the chance to figure out that it was an implementation detail in the related techs that caused their issue.

The solution to this in my opinion is that only people with badges in a specific tech or a moderator has the right to close a question. Same applies for the staging ground in my opinion.

In the times we received hundreds of questions a day it was overwhelming for the SME to keep their tech clean and I valued the preview of other people. But as of today it is relatively easy to keep the site clean with just handful of questions.

So no - I don't think we should stop closing questions. I think we should leave it to the communities in the techs to do that and require them to an justification of closing to the user in the comment section.

Let's boiler this down.

You are sitting in an office and someone comes to you and explain to you their issue. How do you react in the real world?:

1. Put up a sign and close the curtain.
2. Start a dialog and give them a handout.

I personally go with "2", because I find "1" rude and wouldn't like to find myself on the other side of the curtain. I do appreciate the staging ground in that regard, because there is now an instrument for it.

A "handout" can be:

• A pointer to a different question where their issue is solved.
• An answer to their question
• A pointer to formalities that are needed to proceed.

I don't agree with people when they say "We are not a help-desk". Indeed we are a help-desk! Just because we already created a bunch of "handouts" doesn't mean people stopped struggling and they deserve to find answers as well.

Then there is the argument of "They need to do their research, it's not our end of the deal". Okay fair enough, but the same people arguing this way created links and multiple pointers to similar questions because it's hard to find the duplicate sometimes, since you need the fitting "terms".

That's why duplicate questions are valuable as well, you find more entry points to the desired "handout". That's why those "AI" are getting preferred, they formulate the question x-times in a fractions of a second and come back to you.

However we, the people, can still give context and advice, which an "AI" isn't able to provide. Because we don't just formulate the question differently, we can formulate an entire different path which an "AI" doesn't even consider.

So, I don't mind and never really had to, to close a question, as long as a valuable "handout" is provided.

We can achieve this by:

• widening the staging ground
• ensure the quality standards of "asking"/"answering" and of "reviewing"
• put up better signs

I'm tentatively in favor of this. In theory, closing a question is supposed to act as an enforced buffer so that the user can improve it. The problem is, this is often done inaccurately (questions that aren't close-worthy get closed), and once a question is closed it is very difficult to get it reopened. People instantly associate the dreaded [Closed] label with low-quality questions, and tend not to look deeper.

Summary: Close more volume + allow more power for new comers

Details:

While it is no secret not many new questions are being posted, this does not mean we stop closing questions. In fact these high standards are the very reason why it has become successful.

As with any system it is a matter of evolution in changeed circumstances - in our case - it is the LLM's age.

To me it appears the logical next step is to acknowledge the situation and find a way and we have done the first part and trying to find route for the next phase.

In the same modus operandi as before we just need to have more activity - meaning, close more and allow more power for new comers to comment or post. This means relaxing rules a bit interms of min reputation needs.

Also, while having a rule saying new questions must be posted on stack overflow when LLMs are using SO content may not be possible, there should be no-exploitation and/or attribution clauses in SO data license.

How about something else than closing?

I used to work on a dead tree computer magazine back in the '90s. When we opened our BBS, we had a policy how to deal with crap some users wanted to post. We didn't delete the messges, we returned them to the sender by moving it was back to the user's inbox. A followup was included that it didn't pass our ToS and thus the message was handed back. (The system, First Class, was able to move messages in that way.)

Could we use somehting like that in closed questions? Instead of closing, those will go back to the user's whatever inbox-styled system we have (I don't know, I haven't asked a question in 10 years...), or staging ground, or whatever, and give the user a (curated?) option to edit it before re-posting.

Of course, the dupe hammer closing should be extempt from this system.

Would like to highlight two concepts: toxicity and activity in relation to specifically duplicate closures.

For as long as the site existed, users have complained that the most toxic part of Stack Overflow is the question closure.

Another perspective on this: asking a question on SO in many cases requires hours of effort. If not on research, then on edits to improve quality. This also makes a clear difference with early years, when questions could have been asked even without upper letters.
The most toxic sentiment is received from people who expect that asking a question should take mins and minimal effort.
Allowing this will not only make SO a copy of Reddit, but also losing strategy because it was not designed to compete in that niche.

1. I believe sites like Medium and Reddit actively use fake activity and fake accounts as both part of market strategy and income source. This is one of the reasons why it looks like SO is less used.
2. Another reason of low activity is that answering a question under current circumstances means just helping to train LLM.

Both 1) and 2) look to me like very hard to solve details, but having them in mind is useful to avoid breaking changes on SO while trying to compete with other sites.

Questionable, but valid solution for adding room for low effort activity, while still keeping site quality:
Introduce full scale archive (archive.stackoverflow.com?) with lower priority in search and start actively versioning questions.
This way, for example, a question on Python 3.12 won't be closed as a duplicate of a question on Python 3.11, but instead 3.11 will be moved to archive.
The cost will be users to start copying best answers from 3.11 to 3.12, but if SO requires CPR, that direction can be considered.

So the main change from current will be: questions on ..., 3.9, 3.10, 3.11, 3.12 are dropped to archive when similar question on 3.14 is asked.
Upd: a set of rules and conditions is implied, for example, the top answer on the archived question should be outdated and the new question must explicitly contain why the previous top answer is no longer optimal.