TL;DR: The opinion experiment revealed that users aren't looking for subjective debates; rather, they most often seek detailed explanations. To capture this value, we propose a radical shift: stop closing questions and introduce a new curation model—The Workshop for collaboration and The Archive for curation. Sit down, grab a drink, and start reading.
Years ago, Shog9 wrote that managing a community is like managing water in permaculture. He described laying “contour strips” to catch the rain, managing the torrents that threatened to wash the farm away.
Shog9 wrote from the perspective of managing abundance. I grew up in the southwestern deserts of the United States and know what it means to run out of water.
I know the feeling of scrambling up the walls of a slot canyon to escape flood waters brought by a coming storm, watching precious resources flash by and strip the land away. It is not lost on me how slight miscalculations of water management can be the difference between a landscape that survives and one that withers away.
For a long time, the water of this platform–user engagement, and contributions–was plentiful on Stack Overflow. We built dams to control that flow. This made sense in the past and was the correct call. Stack Overflow, once a plentiful paradise, is now experiencing a drought. Somehow, we find ourselves somewhere in the desert with a dam. Our fields are parched, yet our sluice gates are almost entirely closed.
The Data: Opinion Question Experiment (The Rain)
After running the opinion-based question experiment for just about two months, we know it's still raining upstream, and we have water sitting in a reservoir just behind the dam.
Are closures for “opinion-based” reasons going down and are we seeing more “opinion” questions?
In short, yes. Though this should not surprise anyone, given that they can’t be closed and there is no real, defined set of rules. Anyone poking around on SEDE can see that question closures are continuing to decline.
Are we seeing ongoing activity on these questions?
Yes. Opinion-based questions are getting about three times more replies than regular questions (answers + comments). As of this writing, there are 1600+ opinion-based questions, 78% of which have at least one reply. In terms of engagement, this is a clear signal of demand. Our month-over-month numbers show an increase of 16% in engagement (questions, answers, comments, replies), which is unusual given the typical slowdown in participation we see around this time of year. Some caveats: novelty is a potential contributing factor, but only time will tell us by how much. By our count about 85 replies were about telling the asker that their questions should have been asked as traditional Q&A (which is less than 2% of overall replies). This however isn't an easy thing to track, so there may be a few unaccounted for.
Is any of this good?
Our specific goal was to increase engagement, and to that end, this is a measurable success. Given the 16% lift in activity we observed, these results suggest this is a viable strategy to stabilize participation and open new avenues for growth. Crucially, that 3x increase in replies came with a lower flag rate for these new question types (2.9% vs 3.26%). This indicates we can expand participation without sacrificing quality. While technical bugs caused some content to leak into the wrong queues or question lists, these are solvable engineering issues, not existential threats.
Even more interestingly, this wasn't just new users asking lazy questions. The median reputation for users asking these opinion questions was 10x higher than those asking troubleshooting questions (113 vs 11), and their accounts were significantly older (median age of ~ 7 years vs ~ 2.5 years).
This tells us that experienced developers are hungry for these conversations. And the community valued them: nearly 30% of non-deleted opinion questions received at least one upvote, proving that when we allow space for nuance the communities find value in it.
We won’t ignore the fact that this is far from a finished product, and that there is work to be done to get there, but if you have found yourself concerned about the drought of participation, this experiment offers a clear signal: the rains have not left us entirely; we just need to adjust how we capture them.
Our Learnings: “Opinion” is a misnomer
For years, we have treated opinion-based as synonymous with subjective. They both have been defined loosely as content that dilutes the factual purity of the content on the platform. A light analysis of what users are asking in this experiment revealed four distinct intents that staff categorized into four groups:
- Explanations: Focusing on conceptual understanding, learning, or finding why rather than just how.
- Implementation: Focusing on achieving an operational result, like getting code to run.
- Improvement: Improving a working solution.
- Other: Self-promotion, or any content that didn’t fit the other categories.
Ironically, we found that the “Troubleshooting” type we introduced as an option during the experiment was almost entirely categorizable under our new implementation category: users trying to get bugs fixed.
The revelation came when we analyzed the question users submitted using the “Opinion-based” option provided during the experiment. These were not just random solicitations for comments or opinions, which was a concern for both the company and the community, but a majority appeared to be questions asking for explanations. Users were asking for help making architectural decisions, understanding why a framework behaves the way it does, and weighing considerations before writing code.
In the current curation system, curators close questions as opinion-based because they don’t fit our strict “one right answer” format. In reality, a lot of these are questions of conceptual understanding, based in facts, reasoned through collaboratively, and grounded in truth. By damming these up for years, we have incidentally been blocking meaningful attempts by people trying to gain deeper understandings and have “why” conversations that help them grow. “Opinion-based” often just meant “nuance required.”
The historical struggle
Despite the reality of declining participation, we seem to have shut the sluice gate even tighter. Our current model of curation has resulted in a state where nearly 40-50% of all incoming questions are ending up closed every month. This high rejection rate isn't a failure of the people doing the curating; it is the inevitable result of a system designed for flood control operating in a time of scarcity. The dam doesn’t help anymore. We are aggressively filtering for a volume that no longer exists.
The question of closure has been a topic of conversation for a long time. Let's have another! That linked post is still worth a read; functionally, nothing that was laid out there has changed all that much. Closures are more user-friendly than they were 12 years ago and usually take far fewer votes to do, but essentially every issue raised in that question and most of the linked questions remain today. The vast majority of casual visitors’ and new askers’ questions do not naturally align with our content standards. Various attempts over 16 years to change that reality, from both the company's side and the community's side, have not succeeded. This has cost us.
Our curators built a resource that every developer on Earth has relied on at some point in the last 16 years. That was done using a system that worked for an era of abundance, when we could pick and choose only those questions that mattered most. But it is becoming clearer that the system we currently have is considerably out of alignment with the needs of developers today.
We have had this structural problem for a while. Renaming "close" to "on hold" didn't fix it. The problem isn't that the dam is rude, it's that it works too well. We don't need a new dam—we need to let more water through.
Consider this principle: “If a question is closed and goes unanswered forever, we'll never get a chance to see who it could have helped.” The answer is what gives value to the question – both to the asker and to future visitors. Maybe it turns out an answer is only valuable to the asker, but an unanswered question is valuable to no one.
The proposal: The Workshop & The Archive
This is a concept; no development is underway. While this introduces new edge cases, we believe it addresses our most pressing problem: our current curation model is designed for an era of abundant questions that no longer exists.
The Workshop: The contour strips
In Shog9’s answer, he spoke of contour farming: a way to catch water and allow it to be absorbed into the fields rather than running off.
The Workshop would be our new contour strips. Just like contour strips are intended to be the place to catch all the water, The Workshop would be intended to catch almost every appropriately on-topic, answerable question on Stack Overflow.
Think of this as the public evolution of the Staging Ground. We want to take the tools that worked in that sandbox (nudges, templates, structured feedback) and make them the default experience for everyone. Spam and abuse would still be deleted, but the strict rules against conversation would be lifted here.
The Workshop would be the single entry point for all questions and utilize a threaded reply structure with improved notifications rather than the strict Q&A format. This would allow for messier collaboration. Incomplete or unclear questions could be discussed back and forth without penalty. We want to keep the conversation open while the water soaks in. The purpose is to see what value these posts might hold for the future.
With proper irrigation in place, we’d want to see organic growth:
- No GenAI: We want human contributions, not synthetic filler. For the purpose of this feature the policy will remain in effect.
- No link-only answers: We want the irrigation to happen on-site. Dropping a bare link doesn't water our fields; it just pipes the water somewhere else. Content should still be able to stand on its own.
Beyond that? Minimal friction. If you wanted to help by saying, “Have you tried checking your variable names?”, that would be a valid form of irrigation. We wouldn’t be checking for formatting perfection: in The Workshop, the first and foremost goal would be to help the asker.
The Archive: The harvest
The Archive would be the “harvest”—the curated knowledge we offer the developer community at large. You are likely thinking, “But how do we curate if we cannot close?”
Introducing Expert Endorsements
Expert Endorsements (a working title), a unique vote reserved for some community members, would be a validation of the question’s future value from the established curators on Stack Overflow. It would be the opposite of a close flag. Getting one means you have created content that is the cream of the crop; it means your question can be found easily and that it should be surfaced to future searchers and learners, because it's good.
Instead of voting to close, you’d vote to elevate. The system wouldn’t be built to hide or remove; it would be optimized to highlight what's worth looking at. Curators would not need to classify and categorize every single incoming question, only pull out the most useful posts and cultivate them for the harvest.
Today, a small, dedicated core of you are doing the majority of the work of maintaining the knowledge base. This is meaningful work, but it is not always popular. Few do it, and those who do earn criticism in the wider developer community for just trying to preserve and improve the knowledge for others to use. Further, this system is structurally predisposed to cause burnout. Despite sincere intentions, the current tools channel you toward burying questions rather than improving them, and ask you to apply those tools to each one of the tens of millions of questions on the platform.
We want to build tools that allow you to truly curate— selecting the best content for presentation, distribution, and refinement. We would need new tools, like proper mapping and support for duplicate content, but for now we only want to get the high-level idea out there. We want to move past the era of “curation as cleanup duty.” Instead, we would like to position the most committed members of our communities as what they are: a cast of unique individuals who should be celebrated for their commitment and expertise. These are people who leverage their extensive knowledge to make the best knowledge on the Web shine.
Our request
This idea comes from rank-and-file staff after confronting the results of the opinion experiment and the reality of our platform activity data. It is not anywhere near a finalized roadmap, since we wanted to get it on Meta as soon as possible for discussion, but we are actively exploring these notions of "irrigation" and "harvesting."
However, to make The Workshop work, we might need to make some changes to how the core Q&A pages work for that space. We would probably want to move towards a threaded discussion format to support The Workshop. This presents a critical challenge: how do we visually protect The Archive when content is promoted from The Workshop? If a discussion page in The Workshop looks like a standard forum thread, we risk losing the prestige and discoverability of the best answer. We cannot let the harvest be buried by the chaff.
We need you to help us design the harvesting process:
The mechanics: How does a discussion in The Workshop become an artifact in The Archive? Is this a physical migration (moving the Question and its best Answer to a new page) or a visual transformation (signaling high quality within the existing thread)? What are the mechanics of promoting a conversational chain in The Workshop into a standalone, Endorsed Artifact in The Archive? For example, if a “messy” Workshop question receives a brilliant answer, does the question need to be pruned before it enters The Archive, or can the answer stand on its own?
The curator: Who should have the power to ‘Endorse’ a question or answer? Should it be reserved for Gold Badge holders, elected moderators, or a new tier of subject matter experts?
Visual distinction: On a discussion page in The Workshop, when a curator gives an Expert Endorsement, how should it look? If the rest of the page utilizes a threaded discussion format, does the Endorsed Answer get "pinned" to the top? Does it get a gold frame? Should the classic answer styling be reserved exclusively for Endorsed Content to signal its status as an artifact on The Archive? And crucially, how do we allow users to toggle between the “messy” collaborative view and the clean “curated” view?
To summarize: we’re re-imagining Stack Overflow as a platform where curators can focus on elevating high quality content in order to maximize our yield of worthy content, rather than needing to be focused on burying the poor content because we’re holding on to a system that is designed to prevent high volumes of participation which are no longer our reality. We want to hear your feedback on this vision, either in the form of answers to this post, or via email if you don’t feel particularly inclined to share your opinions publicly. For email, send your feedback to: [email protected] with the subject header including: Attention: Hoid
If you don’t think that change is possible or that question closures don’t need a serious rework, that is feedback we want to hear: please articulate why the status quo remains viable in light of the current state of the platform’s ecosystem, if you hold that view. If, like us, you agree that systemic change is needed to meet the current reality of declining participation, your feedback on the above questions is particularly important to ensure we uphold the principles that ensure Stack Overflow remains a source of reliable, high-quality content, for generations to come.
We want to acknowledge that we’re posting this right before the holidays, and like many of you, we will also be taking time off. Rest assured, no immediate work on this concept is planned, and we are at a highly conceptual point in the process. We also take an end-of-year hiatus, and won’t be checking back in until the first week of January at the earliest. Pace yourself accordingly, as there will be multiple opportunities to hammer out various details as we address what has been brought up before we make any decisions. Happy holidays!