Ideas for incentivizing curation of old questions

The good old pastime is gone. Here is one that is inferior (old questions in bad shape probably aren't very valuable, we didn't ignore questions randomly) and less likely to result in a useful result (the original asker is often gone too, unable to give any clarification or confirmation) and likely mostly helpful for AI (which currently and for the foreseeable future doesn't give attributions). Why don't you do this?

Reading it like that I admit I'm not that bored. I actually want to do something really meaningful. Wading through the long tail of lowly voted, unanswered questions sounds like a Herculean task. Surely some of them are pearls waiting to be polished but most are not. The pearls to sand ratio would probably be most unfortunately, mostly downvoting and closing. Simply deleting them all would be faster.

In my eyes the original mission failed in that activity came to a premature halt and future knowledge will not be contained in the library. Likewise I don't want to be the last one looking through the remains of what is left from the past.

And what would the purpose be? Just to help AI crawl the content better and serve it on other sites without giving any attribution? This is more and more happening and might not be enough motivation for some people. I don't want to be the nameless helper of AI applications.

Nevertheless there are the usual things you can do:

Gamification Give rep and badges for the activities that you want to push. Like the excavator badge only much more so. And edits that add or update content are really badly paid in rep.

Tools Implement (by the company) better tools so that the work can be done more efficiently. That also increases motivation. Like better lists of questions I'm really interested in. More support in finding duplicates. Labels for outdated answers. Or else

Money If the only reason is AI, surely AI companies can pay people to do this work for them. AI is said to massively lower entry possibilities for software engineers, the money saved can be spent on data collection instead.

Statistics Try to gather information about what needs to be done and what is done and visualize it. If only quality could be measured somehow. But surely some quantitative stuff can be given like unvoted upon questions/answers. People like to see their progress.

Collaboration See for example the burnination process where the community coordinates tag removal. The same way for example clean up operations could be coordinated.

I've been thinking about this for some time as well, but writing a quality post takes more time than I currently have available.

The help I can offer, however, is the perspective of someone active in low-traffic tags. My feeling is, there aren't as many people from those tags on meta, compared to e.g. python or web development areas.

What makes Stackoverflow useful for me is that I save time. For basically any objective, on-topic question that multiple programmers face, there's one or more Q&As. We have the close reasons to remove questions that don't meet this criteria. Cleanup, in the best case results in a canonical Q&A, either by editing or posting a new question. For the tags I'm on, this doesn't really work. There are a few popular questions, but they aren't put in shape. But they still help people, because they contain some answer, often from experts on the topic. Only, those experts sometimes don't want to spend their time to make the Q&A shine, and not many others do so. Also the author of the question is often long-gone.

So we ended up with lots of "useful to a handful of people, but not very polished" questions.

Fixing them in the current landscape is often impossible because of three main reasons:

1. Few people vote on them
2. We discourage edits that "change the authors intent"
3. The question is (close to being) off-topic

I'll ignore 1., because the company already has its own agenda. Still, up and down voting is viable curation, of course.

For 2., I think we could be a bit more lenient, especially for questions. I'm not sure, though, how to achieve this without creating chaos. Existing voting has taken place on the existing answer. Major changes decrease, how much I can rely on a post's score. Still, I believe if done carefully, this will improve the site. Many of the top voted questions have very little resemblance to their original form. To me, I don't care if the author asked "X". If thousands more people find a solution for "Y" in a question, than to "X", the question should be about solving "Y". Alternatively, the issue could be handled by better meta data on posts, as suggested by TylerH in the comments.

For 3. I see the rationale to clean up. However, I consider myself lucky that many tags haven't been cleaned up, because I rather have an off-topic answer than having no answer at all. I know I won't win anyone over to keep those questions, but if we decide on more aggressive off-topic closing, I hope to see a way to preserve the knowledge. Migrations are not the solution, because they are blocked after 60 days and the process doesn't scale (AFAIK, there's e.g. no queue for migrated questions). Not sure how to find a balance here.

Shove old content under my nose to look at

I agree that SO is transitioning into more of a "maintenance/sustaining" mode. The Site Analytics graph clearly shows that the number of new questions per week is steadily and rapidly declining towards 0.

I view questions on my tags of interest on the Active tab, and I will make improvements when a question shows up due to recent activity. The Community bot bumping old questions is one automated way of showing me old questions that could use some attention.

Expanding this automation would make it more likely for me to do more curation to improve old content: editing questions and answers, flagging comments, etc.

I know people generally dislike the Community bot kicking old questions because they also show up on the Home page. It would be better if any expanded automation would not show up on the Home page.

Here are some ideas for this:

• Bump a random question.
• Have a new mode where a random answer could be bumped, similar to how added or updated answers are shown in the Active tab.
• Allow a user to opt-in to this new behavior.

This would be more useful if I could choose which tags are bumped. For example, my list of Watched Tags would be a good starting point. I don't find the general Community bot bumping to be useful because I rarely look at all questions; I almost always filter on tags of interest.

Extend my privileges

I agree with you that SO has changed dramatically in recent years, and now drastic changes are needed to achieve this goal.

I realize this is a tougher sell than my previous answer, but the best incentive for me is extended privileges.

I am not interested in earning more reputation. Reputation was a way for me to earn more privileges, but I have no new privileges to gain. To be clear, I am not asking for another privilege level based on more reputation. I am asking for added privileges based on the current levels.

Users who accumulate 20k rep are considered "trusted users". Place more trust in us.

Perhaps start out with doling more powers out to users with gold tag badges, similar to dupehammer powers. For example, powers to single-handedly:

• delete comments
• delete duplicate answers
• delete wrong answers

The rationale is that gold tag badge holders are subject matter experts in those tags and can decide whether the content should be removed, just as they can decide if a question is a duplicate.

Maybe make it a requirement that a user needs 20k rep and be a gold tag badge holder.

Perhaps these types of changes could be rolled out gradually as an experiment.

You asked what would keep me interested, this would be a huge step in the right direction. I can't think of any small changes that would make a difference; big changes are needed.

I think there's something to be said about making space for these kinds of conversations.

Making intentional space(s) for experts in particular topics to talk to each other about curation of high-visibility content in their topic spaces could help. I.e. enablement (it still depends on how many experts there are that are interested in doing this). Doing some appropriately placed/timed promotion of those spaces would probably help (Ex. based on your watched tags / what tags you answer in or your tag badges).

Chatrooms are the first mechanism/platform that came to mind (Ex. how there are language chatrooms like the Python, JS, and C++ rooms). There's also Collectives as a potential option, but the community doesn't have control on creation of Collectives. I don't know how well Meta works for this kind of discussion (it's not really designed for back and forth conversation- Chat is). The Discussions space could be an option too. There are probably a couple relevant feature-requests floating around (I'm guessing there's probably at least one with reference to Wikipedia talk pages).

It's not like people can't already make these spaces by spinning up a chatroom today, but there are some things that maybe aren't ideal about it, like auto-freezing, and visibility of chat on the main site. I think the thing that needs thinking about is how to make it so that these spaces come to the attention of people who could be interested in participating without annoying them if they're not interested (or maybe there's already a really good proposal out there waiting to get staff attention; I don't consider myself well-read on the feature-requests in this area).

It's important to make sure the repository is of high quality.

10 years ago I would have agreed instantly. Nowadays I am not sure maintaining a high quality repository is possible anymore as the company is not prioritizing that goal and from my perspective has been actively working against it on some fronts by focusing on quantity over quality.

You seem to be mainly looking at this from the angle of improving old posts, but I'm not sure that's the correct thing to focus on. The site has accumulated lots and lots of posts over the years that, while not neccessarily worthless individually (although IMO a huge amount of the repository is trash), add nothing of value to SO as a whole as the topic is already covered in many other posts. Similarily, improving any individual post makes that page better but does not neccessarily improve the repository as a whole.

Personally one of my biggest issues in recent years is that google search results from SO are littered with tons of basic posts - and yes, google enshittification is part of that but it's still leagues better than SO search. For example, searching for advanced JPA topics or nontrivial Spring errors returns multiple variations of superficially similar entry-level errors, and often google-fu and some luck is needed to find something that's actually relevant to my situation. Nightmare difficulty would be searching for an NPE thrown internally by a 3rd party library; even if a relevant post exists on SO, it will be drowned by hundreds of "my own code threw an NPE" beginner questions (once upon a time there were dedicated search engines such as samebug and symbolhound which were great for these cases, but they don't exists anymore).

If I do find a relevant and answered SO post I usually don't have any issues dealing with other quality issues. Bad formatting and presentation is an inconvenience at most, and if the accepted/highest-voted answer is not the best to use in 2025 I usually catch that as I try to understand the offered solutions and will look at multiple answers by default. For someone who just copypastes the first code snippet they see it would be important to ensure that's the best solution, but personally I'm pretty certain I'll find the best answer unless it has a negative score or is otherwise obscured.

So to improve my situation, lots of basic posts would need to disappear from the search results, which probably means deleting them (and consolidating any useful answers that don't yet exist elsewhere). I'm not sure if there is another mechanism e.g. if negative score or duplicate closure affects the google index; even if dupehammering works it would require a good target and could be controversial for similar-but-maybe-not-similar-enough posts. And for any questions which do not warrant closure on their own AFAIK there is no valid path to deletion by the community (except for the roomba for questions without accepted answers that can be downvoted to a negative score).

So from my PoV any broader improvement of the repository as a whole would mostly turn into a excercise of removing the chaff from SO, and I can't see that happening for lots of different reasons - for starters it would be very hard to decide which posts to keep and to remove, would require subject matter expertise to understand what is similar enough to consolidate and which parts of which answers add value, would require broad community consensus to not cause lots of drama, and would require SE to not hinder any of the efforts to remove huge parts of the repository.

So TL;DR improving old questions is a noble goal but might be futile in the grand scheme of things, as the most amazing QA pair is worthless if you can't find it.

Interesting question. I read “It's no secret that Stack Overflow has pretty much stopped receiving new questions and answers,” then I go back to the title, “Ideas for incentivizing curation of old questions,” and my poor brain translates it as “how to improve the appearance of the baseboard in the kitchen while the ceiling is falling on my head”, really interesting.

We can continue to “look the other way” we can continue to shift responsibility (“it's AI's fault”**), or we can do something truly reckless: accept that we have a “very big” problem with how the site works and try to reverse the situation... Yes, it's not easy, but it's not impossible, it's not even a Herculean task. .

I don't expect the company to do anything about it, history shows us that it has always followed its “roadmap,” turning a “deaf ear” to the community's complaints and, in this particular case, to its own convenience.

I don't expect the curator community to do anything, we know how it is (before I get into the description, I apologize for the generalization; many of the curators do not deserve to be described in the following lines, so please, feel excluded):

Our community:

• It systematically resists almost any attempt to “rethink” the model, and when I say “model,” I don't mean the one proposed by the company, but rather the “own” model, which can either follow the company's model (which already has enormous shortcomings) and going completely in the opposite direction.

• It has embraced the protest vote as a method of "personal satisfaction"? (if you say this is not the case, try to explain why most negative votes, even to correct answers, have no comments attached, or why a poorly worded question... or not, can receive more than ten negative votes, some after being closed, and a good answer rarely reaches five).

And if I don't expect anything, why do I publish?.

In principle, I am human and therefore imperfect, so I force myself to accept that I may be wrong and that perhaps what I consider immutable is not so. On the other hand, remaining silent would make me an accomplice.

** I know that many will use statistics to reaffirm that “it's AI's fault,” so let's take a closer look at them.

• “Coincidence” is not the same as “causality.” The fact that two events coincide in time and space does not necessarily imply that one is a consequence of the other.

• In this particular case, I believe that there is a correlation between the two events (proving this is very complex, if not impossible), but we must be very careful in our analysis, as it could refer to two very different scenarios. Let's look at the following two scenarios:

• “Our site was working very well, but then AI appeared, offering a much superior service, and displaced us.”

• “Our site had very serious problems, concealed by the lack of competition, and when AI appeared, offering a service of inferior technical quality but greater human quality, and given the pronounced dissatisfaction of ‘visitors’ with how the site worked and how they were treated on it, it ended up displacing us.”

They are very similar, but the implications are vastly different. If the first scenario had occurred, there would have been an initial loss of visitors, but it would have been restored over time due to the serious limitations of LLMs.

Note: The description of the “flaws” of our community of curators is incomplete and does not differentiate between behaviors determined by the help center and those that are unique to it.

They could use the site for what it is actually intended for: technical Q&A. That's the core purpose of this site.

Curation is just a side-effect of that. If someone's only interest in this side is curation and not technical Q&A, then maybe them leaving the site is not a bad thing.

Once one is so out of touch with the core purpose of the site that one only does curation, one also gets out of touch with the site culture. Enter: sub culture, where people start thinking the purpose of the site is curation and to turn everything into the perfect post, rather than filling it with relevant technical content.