-230

Update December 16th, 2025

Threading was released this morning. You can reply to different replies to the question and then respond to those replies, and so on. We don't have a cap on levels of nested replies, but we'll monitor it to determine whether one is needed. It's a simple addition on our end if the need arises.

For updates on where the experiment could be headed next see this new post about curation

Update November 4th, 2025

As mentioned in the initial announcement, we held a chat with the product manager leading this experiment to answer questions. The chat room transcript can be found here.

Update October 27th, 2025

We have officially launched this experiment to ten percent of all users on Stack Overflow who have not opted out of experiments. Couple of things to note:

  • We have moved the staff chat mentioned at the bottom of the post to November 4th at 9 AM EST. Again, please check back before that date and time for the link. We wanted to give the community at least a week from when the experiment launched before having the chat.
  • For any examples of opinion-based questions that you might come across and believe to be good or bad candidates for the site, please leave those examples and your rationale on this separate meta post.
  • All other feedback, bug reports, and long-form suggestions on the feature's design and concept can be left on this post. Please use the bug or feature-request tag in your answer for these specific purposes.

Today, we are following up on our announcement regarding the topic of including opinion-based questions on Stack Overflow. The Community Enablement team would like to share some details about our design perspective, what we hope to measure, and the specifics of our upcoming experiment, which will soon be launching on Stack Overflow to a small number of users in the coming days.

This project aims to create a designated space for technical discussions, best practices, tooling recommendations, and architectural questions or advice that are vital to a developer’s workflow but often are closed under our existing, highly structured Q&A model. Of all the questions asked since the beginning of the year, 24% got closed as opinion-based or off-topic. These questions aren't necessarily bad; they simply don't fit the definitive-answer format of traditional Stack Overflow. By allowing these questions, we aim to unlock valuable, previously unresolved questions.

Designing for the New User: A Softer Entry Point

The goal of this new approach is to add more quality information to the knowledge base. While also providing a softer on ramp for community members who can’t confidently engage in traditional Q&A.

We recognize that the strict, objective standards of traditional Stack Overflow can feel like a high barrier, and those barriers are there for good reason. New users often struggle to phrase their questions in a way that meets the minimal reproducible example or troubleshooting standards, especially when they are still in the planning or decision-making phase of their current project.

Visibility and Opt-Out

The new question types will be integrated into the question feeds, clearly marked with their category (Advice, Tooling, Best Practices). We are bringing this subjective content directly into the Stack Overflow experience because better visibility is one of the key lessons we took from Discussions.

Users who prefer the traditional experience will be able to opt out of this experiment to avoid seeing the new content entirely. If this experiment is successful, the next set of features to be prioritized will focus on user content filtering, allowing users to filter out specific content or selectively choose which types they want to see.

Screenshot of a logged-in user's personalized Stack Overflow home page, showing a welcome message, reputation, badge progress, watched tags, and a feed of questions including opinion based questions signaled by the following labels:Advice, Tooling, Best Practices.

Screenshot of the Stack Overflow operating-system tag page, showing a list of recent questions related to operating system development with some holding the new labels to signal they are opinion based.

Asking Experience

As you can see in the mockups, the key change is a simple ”type” selector on the Ask Question form. We ran some research with users to confirm if they could successfully and consistently label these questions based on the question labels we offered them. Based on our research questions, questions were labeled correctly about 90% of the time. We landed on these question types based on suggestions made during Discussion experiments.

  • Default: The experience remains Troubleshooting / Debugging for classic Q&A.
  • New Options: Users can now select categories such as Tooling Recommendations, Best Practices, or General Advice/Other.

When a user selects one of these new types, the guidelines on the right adjust to provide more specific guidance for consideration while writing their question. Such as:

  • Questions that invite more in-depth explanations
  • Questions that invite community members to share relevant personal insights, direction, or solutions that have worked for them in a similar situation.

Differences from traditional questions:

  • The UI for these new question types is intentionally simpler (see mockups below)
  • We're replacing the voting model with thumbs up and down
  • We're removing reputation
  • We're removing the ability to accept one of the answers

This shift is designed to encourage nuanced, conversational answers by signaling that there is no single “correct” response, allowing multiple solutions to coexist and be valued.

Screenshot of the Stack Overflow "Ask a question" form, with the question Type set to "Best practices" and a Guidelines sidebar explaining that this question type is for open-ended discussions about topics like best practices, recommended tools, or architecture.

Opinion-based question UI

Once posted, these questions look a little different. We have replaced the vote buttons with thumbs-up/down buttons at the bottom, and moved the user avatar and tags to the top of the question header. This is because the research group responded most positively to thumbs for the “score” behavior, instead of the other options that were presented to them. For this experiment, we will only ever show the thumbs-up count, both on the question post and in the question feed. Eventually, we intend to use this for filtering purposes in the feed, should the feature continue to demonstrate success. Given the opinion-based nature of these question types, we determined that showing a thumbs-down score may not be beneficial to the user experience and how welcome they can feel as a result. After initial testing, we may further improve the thumbs down action by incorporating a feedback mechanism that nudges the asker to improve their question based on feedback collected from users who give the question a thumbs-down.

We have also replaced the answers header with a “replies” header, and functionally made the replies look more like comments; ideally these will eventually support threading, and continue the design style of using thumbs up/down. Again, we will only show the up counter, and in the future, we will consider adding logic here to highlight better replies.

Options for flagging will be Spam, Abusive, and Other. For spam and abuse, four votes will result in the deletion of the content, while content that doesn’t meet that threshold will be directed to the moderator queue. For items flagged as “other,” the Community Enablement team will be monitoring them, as well as taking moderation actions on them when necessary, to gain a deeper understanding of the tools needed to support this content and to understand the moderator experience. Current moderators are welcome to participate if they like, but there is no expectation that they have to help. The Community Enablement team will stay in touch with Stack Overflow moderators through the process to take advantage of their expertise and collect whatever moderator specific feedback that comes up during this experiment.

Screenshot of an example Stack Overflow opinion based question titled "Good patterns or strategies for long term maintenance of mid/big sized apps," which is an open-ended question tagged as Best practices and includes a few example replies. The user avatar, and tags have been moved to the top of the quesiton instead of the bottom

Question Closure Options

While we won’t rule out some form of question closure in the future, we want to focus on improving the current process here to something more constructive that encourages the asker to refine their question in a different way, rather than the current closing process. Once we have determined whether this initial experiment has been successful, we will reassess the closing of these types of questions and how that is communicated to the asker.

Moderation, Community Guidance, and Staff Proactivity

Opening the door to subjective questions requires a commitment to quality control. We will not be leaving this content unmoderated. It took Stack Overflow a few years to establish its current standard of content quality; we don’t expect to reach that today, and should this team find success with these experiments, we plan to refine those standards over time. This team only asks that you keep an open mind while we work through the premise that opinion-based questions can be high-quality.

In that vein, we will be approaching this in a few different ways:

  • Proactive Staff Response: Rather than simply closing vague or low-quality questions, staff will proactively engage with users. Asking for more details and attempting to engage with those users to encourage them to improve their post and bring it to a better state.

  • Active Feedback: We will open a chat room and maintain a designated MSO post, where community members can provide examples of opinion-based questions they believe are of suitable or poor quality and explain their reasoning. The chat room will be monitored by members of the Community Enablement team and we will have scheduled time for staff to be present there to answer questions. More details at the bottom of the post in the, “We want your feedback” section.

  • Experiment Exposure: We will be releasing the experiment to ten percent of users to start. That ten percent will be able to use the new ask form, and everyone else who opted into the experiment will be able to view, reply, etc. The Community Enablement team will be monitoring questions that are asked, spam and moderation flags. We will continue to increase the exposure of the experiment for the next few weeks as we monitor activity.

The Alpha Test: Measurement and Success

We want to be crystal clear: what we are releasing here is an alpha test to validate the core concept, not a beta or general release. This initial implementation is bare-bones and lacks many features (like comprehensive tooling for comparing different variations of the UI to determine what the best experience is) that we would build out later. We are purely looking at this in terms of the concept's survivability:

  • Plainly, are we seeing opinion-based question closures going down, and a consistent stream of the opinion-based questions being asked?
  • Are we seeing responses? We want to see these new question types getting at least one reply, ideally within seven days.
  • Rate of flagged questions - We know that spam was an issue with Discussions. We’ll be monitoring this rate and prioritizing additional spam mitigation tools should we see related concerns.

Spam Prevention

We know that spam will show up in some form or another, so we have implemented the following to help mitigate that as much as possible:

We Want Your Feedback

As an alpha experiment, we want to be open about the fact that if this experiment continues to progress, it could change a little bit or quite a bit from what you see today. So if you have feedback on anything we have presented or changes on how you would like to see this presented on Stack Overflow, please let us know. We have set up two channels to capture your thoughts:

  • Designated Meta Post for content quality feedback: Please use this different meta post for feedback on opinion-based questions you see that you believe are good or not good candidates for the site. All other feedback, bug reports, and long-form suggestions on the feature's design and concept can be left on this post. Please use the or tag in your answer for these specific purposes.
  • Chat Room: We will be opening a dedicated chat room, once the experiment has been launched, where staff will be there at scheduled intervals to talk through the experiment. The first will be held on October 29th November 4th, 2025 at 9 AM EST. Please check this post to confirm the time, as it might change. Chat room can be found here It will be changed from gallery mode about 30 minutes prior to 9 AM EST.

We will be monitoring this post til November 5th, 2025 for feedback.

117
  • 54
    Calling something "best practices" is problematic in itself, as it surely will lead to "which programming language is best for x" type of questions, without anyone bothering to define "best". And it will quickly become very tiresome to point that out, over and over and over. Probably one of the main reasons why such questions aren't allowed on SO these days. Commented Oct 22, 2025 at 15:08
  • 72
    "Proactive Staff Response" We were promised that for the n most recent new features too and in practice it meant "some staff will check it out for 5 minutes upon the day of release then forget all about it and then leave it to rot". Notable examples being Discussions and the chat overhaul. Commented Oct 22, 2025 at 15:12
  • 64
    You don't need to wait to see. Discussions were full of awful "best language" type of questions, we already know that this will happen based on previous experience. Not only from Discussions but also from early days SO. Commented Oct 22, 2025 at 15:16
  • 25
    Surely the default should be to opt OUT. Let those who really want the pain opt IN. Commented Oct 22, 2025 at 15:20
  • 14
    @Paulie_D making it opt-out is far more effective at "proving" it's a feature that should exist. Remember the sticky header option that got removed because "almost no-one used it"? If the default is on, we're relying on positive feedback to determine success but have no means for negative feedback, and marketing it to the masses as some "anything goes" kind of feature, it's going to get (ab)used and there's no real reason to call this an experiment because the outcomes are known. Commented Oct 22, 2025 at 15:55
  • 17
    So the purpose is to divert opinion-based questions & your measure of a successful diversion is it gets a response? What about the quality of the response? Shouldn't you test the reason this experiment is being done, user engagement? I don't mean that to sound hostile. I appreciate that y'all have put some thought into how to get the content some visibility without impacting Q&A. I think English Language Learners could benefit from something similar, because there is a divide in the community over where "opinion-based" starts and some of those closed questions are useful to learners. Commented Oct 22, 2025 at 16:36
  • 13
    @Hoid Seems you have a severe communications problem internally if you can't even interview those who participated in Discussions. I mean what's even the point of launching experimental features then.... Commented Oct 22, 2025 at 16:51
  • 51
    @Hoid The short story is that staff support was promised, a few brave diamond mods volunteered, only the diamond mods did actual moderation - they were essentially thrown under the bus. In the new chat lobbies staff support was also promised but staff didn't do jack, so those are also handled by a few diamond mods from underneath the bus. To launch a new system without even having a clue how it is going to be moderated is to set everything up to fail. Which we also learned from previous experience. Commented Oct 22, 2025 at 16:51
  • 17
    "We recognize that the strict, objective standards of traditional Stack Overflow can feel like a high barrier, and those barriers are there for good reason" Can't wait for you the company to break down those barriers anyway when this "experiment" is graduated after overwhelming success. Commented Oct 22, 2025 at 16:56
  • 33
    @Spevacus Umm... the lobbies were launched in May. You were hired late September. Concerns such as Is Stack Exchange pretty much finished with the two Lobbies? were completely ignored. As were the immediate concerns about consistent moderation outside US office hours here: meta.stackexchange.com/a/409038/170024 Apart from the initial 24 hours, what exactly did the staff do in these chat rooms from May to September. "Jack" seems quite fitting afaik but please prove me wrong. Commented Oct 22, 2025 at 17:28
  • 20
    Euh...!? Misconception...: "Default: The experience remains Troubleshooting / Debugging for classic Q&A." This is not correct, 'Troubleshooting' and 'Debugging' are more or less the same thing, "classic Q&A" means How to? + Why? Questions. (And 'Debugging' = Why? Questions...) // Opinion-based Questions would then be Which? Questions... Commented Oct 22, 2025 at 19:19
  • 23
    I also wonder what the purpose is of having a button (thumb-down) that has no effect. How will people react once they find out? Will they feel cheated on or just shrug it off? I know I wouldn't really like it and I think that only buttons with functions should be present and everything else is just bad UX. Or am I completely wrong? If you decide you need the button later on, you can still add it or not? Commented Oct 22, 2025 at 19:45
  • 48
    Can we please curtail this experiment NOW? It is actively doing harm to the network by confusing many people, even experienced longtime users, into directing questions that should go to traditional Q&A into the opinion-based streams instead, with no mechanism, as far as I know, to correct that after the fact. Commented Nov 7, 2025 at 13:32
  • 12
    @Hoid I think you are misrepresenting higher engagement. First of all, some of those questions are misplaced, and there is no closure so more engaged users are leaving comments about that too. Also regular Q/A expects improving and editing existing posts instead of adding new replies. Combined with the fact that this is new feature and it is only natural to get more attention. This means nothing in terms of viability of the new experiment in the long run. Commented Nov 11, 2025 at 20:28
  • 15
    I mean... these "questions" also have a different look to others, in the question list, that will naturally cause more clicks. They also tend to survive the "this is obviously a duplicate" step and thus receive duplicate help that otherwise wouldn't have been provided, i don't see how any of this can be reasonably compared. Commented Nov 11, 2025 at 22:06

78 Answers 78

Reset to default
1
2
3
16

The UI for these new question types is intentionally simpler

There's an "edit" link on the question & each reply yet there is no button/link for the corresponding timeline or revisions. Normally these are reached for a post via its clock icon button and "edited ..." link.

It happens that the corresponding URLs can be reached by clicking on "edit", which for those posts goes to an URL of the form

https://stackoverflow.com/posts/postid/edit

and changing the last word, edit, to timeline or revisions.

13
  • 6
    This actually has legal implications, since the CC license requires that modifications be disclosed and attributed. I've contacted SE Inc about this via the "Contact us" form. BTW it also affects Staging Ground pre-publish edits. Commented Nov 4, 2025 at 15:19
  • 3
    We have a fix in progress on this one, think it will be ready as early as tomorrow morning or laster this week, but its a high priority. Commented Nov 4, 2025 at 20:44
  • 1
    This has now been fixed, the creation / last modified date on opinion-based posts now link to the corresponding timeline page. Commented Nov 7, 2025 at 11:39
  • 3
    Like I mentioned on your answer you've used the wrong URL for the "last modified" text, @Darce . That should be the revisions page. Commented Nov 7, 2025 at 12:59
  • @Darce The timeline & revisions pages, currently reached via 'clock icon button and "edited ..." link', are not appropriately reached through 'creation / last modified date' labels & I don't know why you would think they should be. I can imagine an obscure rationale that associates creation with timeline & last modification with revisions (and another for vice-versa), but it's not a reasonable labelling of what the pages are--timeline & edits/revisions. Commented Nov 7, 2025 at 13:07
  • 1
    @Darce "This has now been fixed" I don't know why you think you fixed access to both my issues, ie also the revisions page. Do you understand that my answer here is about both timeline & revisions & ThomA's question is only about revisions & your fix is only about timeline? Commented Nov 7, 2025 at 13:31
  • @Darce Please give a link to a Question & Response page that exhibits your (eventual) solution(s) accessing timeline and/or edits/revisions, ie one that has an edited response. Commented Nov 7, 2025 at 13:37
  • @Darce It's bad design currently that "edited ..." & a date get to revisions instead of "revisions" or "revised ..." getting to revisions .... And that it's a "revisions" page not a "versions" page, so that there's never a revision page that contains just one version even though every revisions page has the first version .... Commented Nov 7, 2025 at 13:42
  • 1
    @V2Blast This is NOT status-completed, did you read my comments here in reply to Darce saying it was? Did you read my question & confirm? Post time links to timeline, that is NOT all this is about. Although a timeline page does offer some of what a version page contains. Commented Nov 9, 2025 at 4:30
  • 2
    Maybe it's time for a [bug] post, @philipxy , but with it saying the link is incorrect, and you can cite your answer here. Commented Nov 26, 2025 at 16:06
  • @V2Blast Trying one more time ... please see my last comment. RSVP Commented Nov 26, 2025 at 21:54
  • 3
    @philipxy: I'd agree with Thom A's suggestion; make a new bug report or feature request clearly laying out the issue, and I/the mods can [status-review] it to get it on the relevant team's board. (I believe they're already aware of the issue, but this makes it easier for them to prioritize and address it.) Commented Nov 26, 2025 at 22:31
  • 1
    @V2Blast Could you please [status-review] Display Revisions link on Opinionated Questions? Commented Jan 18 at 19:17
16

Looking at the questions that have ended up in the opinion-based question types since the experiment has started, it becomes clear that the overwhelming majority of questions there aren't opinion-based. I would estimate something around 90%.

Also, there are lots of people being confused by the types. Something is wrong and I think the main culprit currently is the question type selection. It doesn't work.

You said you had done research and people can do that, but it seems that they can't. It doesn't make sense to even talk about opinion-based questions if there are none and as long as the feature is actively harming the people seeking help with non-opinion based questions, who never wanted to ask an opinion-based question and only got the type wrong.

I think that the experiment should be halted quickly, the question type selection should be reworked or replaced by something else and only then the experiment can continue.

As you can see in the mockups, the key change is a simple ”type” selector on the Ask Question form. We ran some research with users to confirm if they could successfully and consistently label these questions based on the question labels we offered them. Based on our research questions, questions were labeled correctly about 90% of the time.

For the experiment it looks opposite. There must be a reason for it.

P.S. Unfortunately the experiment had not been interrupted and clearer solutions sought for the question type sorting problem. The company did not want that. The alternative would have been to try to correct for this on the fly, although it seems to be a rather difficult problem to me. All in all it kind of speaks of a certain inflexibility in reacting to unforeseen effects regarding the experimental approach. That's why hundreds of people asking normal Q&A but selecting the wrong question type had to suffer.

2
  • 1
    Just my guess. They got the 90% data by having a "correct answer" for the type of some test question, and just see how many testers can pick the type that matches the answer. That picking different types leads to totally different question modes and the strong intention of the user favoring one question mode over another is never part of the test. I assume there would be much less complaint if the types are just soft categorization without impacting question modes. Also see this answer. Commented Nov 8, 2025 at 5:32
  • 1
    correctly... according to stack, who apparently didn't know how to questions were on topic. Commented Nov 14, 2025 at 16:19
15

New types of questions are not visible on all questions list. They are visible only on tagged questions lists which makes them less discoverable. Since there are no other options to filter them, being visible on all questions list is critical.

We cannot monitor or give feedback on the new kinds of questions if we cannot see them. Please fix this.

9
  • 1
    Could you share which question you couldn't see and which list it wasn't showing up on? I am not seeing any issues on my end. Commented Oct 27, 2025 at 23:49
  • 2
    @Hoid I've yet to encounter any of the new questions on stackoverflow.com/questions?tab=newest (I've also tried other sorts), I've scrolled to where the questions are 6h old, so to my understanding I should see stackoverflow.com/questions/79802131 (5h at time of writing) amongst them, but it's not there. Same for the home feed at plain stackoverflow.com Commented Oct 28, 2025 at 0:05
  • If it matters I'm being served rev 2025.10.27.35959 for that page. Commented Oct 28, 2025 at 0:12
  • 1
    @cafce25 there are only two of them as of my initial comment. One at the bottom of the 1st page when sorted by newest and the second on the second page about 5 hours ago. Commented Oct 28, 2025 at 0:13
  • 1
    @Hoid I don't see either of them on that page. Commented Oct 28, 2025 at 0:14
  • @Hoid I was specifically looking for question mentioned in other thread, since that is the only one I know of: stackoverflow.com/q/79802131 I can see that one when I click on any tag and scroll down, but not when I go looking for it on list of all questions looking in the timeframe the question was asked. And I did that very carefully not to miss it by chance. Commented Oct 28, 2025 at 7:33
  • 3
    This is a very good point, apologies for the oversight, we're currently rolled out to 10% of all users but mods should have been included, this has been corrected Commented Oct 28, 2025 at 10:09
  • 3
    @cart I have tried both as mod and with completely anonymous account. In both cases I am seeing the question on tagged questions list, but not on full list. So if you are doing gradual rollout, you still have potential bugs because there is discrepancy between what can be seen on full question list and tagged questions list with the same account. Commented Oct 28, 2025 at 10:30
  • 1
    So 100% see them in filtered questions lists but only 10% see them on unfiltered ones? Still quite unclear when they should or shouldn't. Commented Oct 28, 2025 at 12:26
15

Unless you abuse the comment bug on ad-hoc loaded answers there is no way to notify another user about anything making the new questions the least communicative variant of posts yet, from the original post it seems to be the opposite of the intended way they should work.

People participating in "conversational answers" and open ended discussions should be able to converse and it should be possible to notify them about things.

2
  • 2
    Yeah, we are seeing that. The lack of commenting makes it a bit messy, with what would have been comments mixed in with the answer(replies). We might just put comments back on the question only. Or maybe consider something different, like a back-of-the-question meta. Very brainstormy right now. Will approach the community if we come up with anything that deviates from what the community currently expects. Commented Oct 29, 2025 at 15:48
  • 3
    @Hoid Isn't this what you built the entire "threaded comments" feature for? I don't like it in the staging ground, and would really hate it on regular Q&A, but it would really shine on discussions, uh, opinion questions Commented Oct 31, 2025 at 8:23
14

How does this interact with the Staging Ground?

You created a new sort of question flow but you didn't mention how interactions with the Staging Ground would work.
Can these questions get in the Staging Ground?
If so, what happens?
Are the reviewers informed about the "type" of the question?
Would it affect off-topic reasons/comment templates?
Can they be approved normally?
Can the question type be changed while a question is in the Staging Ground?
Is it possible to change a normal question to an opinionated question in the Staging Ground?
Does the Staging Ground workflow change in any way?

I can understand the Staging Ground not being considered to be important for this alpha test and these questions being excluded from the Staging Ground for now (as long as there are no internal server errors or unreviewable questions) but if it goes well according to your definitions/metrics, you'll have to make a decision at some point.

3
  • 2
    Staging Ground is just full of spam and unanswerable questions that cannot be moderated by the community. I don’t have high hopes for this new question flow. I suspect it will just be full of AI generated opinions and questions that don’t need to be answered Commented Oct 22, 2025 at 23:26
  • 5
    Do you have any evidence for the Staging Ground being full of spam? While there is some spam there (just like on the main site), it is handled by Charcoal and others. And it containing unanswerable questions is intentional - questions are supposed to be improved (or not published) and not answered in the SG. But to be honest, I'd say it isn't 'full' of anything. I also don't get why you think the SG couldn't be moderated by the Community. Could you expand on that? Commented Oct 23, 2025 at 5:14
  • 3
    To start, these won't be exposed to Staging Ground. Not saying its a forever no, but probably something we would add much later once we have a stable feature and a lot of the finer details of it are resolved. But their good points, and I don't disagree, but like you mentioned in some of your questions. SG was designed for objective Q&A, so it would need some tweaks to support these question types. Commented Oct 24, 2025 at 15:08
14

I have mixed feelings about this. On the one hand, I see the value in providing a venue for questions seeking the opinions of human experts who actually understand what they're talking about, in contrast to GenAI bots that merely appear to understand anything.

However, I'm concerned that it's intrinsically difficult to apply useful quality control (QC) to such content. I don't believe that your QC plan involving special mod flags and staff oversight will scale well. Also, it's not just the questions that need QC. These questions will invariably attract low-quality replies, and they also need QC. But how do you fairly evaluate the quality of subjective, opinion-based replies?

One of the core reasons that we don't currently permit opinion questions is that it's hard to rate the answers they attract, and we want to avoid the ensuing "my opinion vs your opinion" battles. What mechanism do you have to prevent such battles?

Also, the answers to "best practice" questions can have a short useful lifetime, especially in fast-moving languages like JavaScript, and in new languages and libraries that are still in the early phase of rapid evolution. We already have a huge problem with outdated answers in main Q&A (which the recent change to comment timestamps has made even worse). Explicitly permitting "best practice" questions will just add to this pile of obsolete information, IMHO.

6
  • 1
    for the "short term/lifetime" stuff, I wonder if chat could be a good avenue for it. not many people are in active in programming-language/technology-specific chatrooms though, and most don't have the privilege to talk in them. Commented Oct 22, 2025 at 23:03
  • 1
    @starball Yes, chat is good for that sort of discussion, IMHO. We certainly allow best practices and resource recommendation questions in the Python chat room. That room is fairly quiet these days compared to a few years ago, but there are still several regulars lurking who check the room frequently, although you may not get quick replies to your questions. Commented Oct 22, 2025 at 23:11
  • 2
    You're right, it won't scale well long term. That is not the long-term plan though. Instead, it's intended to bring the team closer to the workflow to think about what is needed to make it scalable. You are also right that it's difficult, but difficult doesn't mean impossible. While I know this is a bit of a hot take, I am not sure we should be aiming for 100% QC coverage in opinion questions; we should aim for around 80% and signal that mileage can vary and that consumers of that content should be understanding that whether it works for them will depend on their unique circumstances. Commented Oct 24, 2025 at 15:20
  • 2
    I think opinion-based questions are less likely to yield an absolute answer and instead offer helpful guidance to help some progress through their specific circumstances. That's a big departure from where we have been the last 16 years, but I think it can be done well, and if enough people are interested in figuring it out, this is the right group to make it happen in a way that works for today's community and the tech landscape we are all dealing with. Things are a bit different than when Stack Overflow started. I think we would be to approach some of these questions with a new perspective. Commented Oct 24, 2025 at 15:24
  • 2
    To your last point. Absolutely, wish I could go back in time to the early days and beg for versioning to be added to every Q&A. Commented Oct 24, 2025 at 15:25
  • Probably they should implement a system for POB discussions like this so that they just automatically get deleted after a year or two. Opinions or facts on the ground will probably shift completely on the topic after a couple of years anyway. Commented Oct 24, 2025 at 21:24
14

or possibly

Please make reply placement consistent

(I previously posted this as a separate question before finding out feedback should go here. It acquired 16 points.)

For "classic" Q&A, the place to add an answer is at the bottom of a question page. Screenshot for reference:

screenshot of box "Your Answer" and button "Post Your Answer" below existing answers but above the footer

However, for open-ended questions, the button to add an answer (AKA reply) is at the top of the questions list, just after the question.

screenshot of the button "Join the conversation..." at the top of the reply list

However, clicking the button jumps to a reply box at the bottom of the page.

screenshot of box "Your Reply" and button "Post Your Reply" below existing replies but above the footer

Please make this consistent. I would prefer the button at the bottom to encourage people to read existing replies before posting their own, to avoid duplicate info.

As well, if I want to reply to a discussion, I have to scroll all the way up.

13

Users who prefer the traditional experience will be able to opt out of this experiment to avoid seeing the new content entirely.

I have opted out of experiments for other reasons. I would also want to see what the new questions look like and how they are utilised.

Is there really no way to discover these new questions at all? Should they not be searchable or have any other way to reach them?

Even if I were to change my experiment setting, it seems like the only way to find the new questions is trying to locate one in the middle of a random question lists. Which is not a good overview at all.

Discussions at least was collected all in one place, so not only could they be easily found, you could also clearly see what topics are there and which topics were in the past. Much easier to get a feel for how the feature is used.

6
  • 8
    Maybe search would be extended to filter by question type. Commented Oct 24, 2025 at 9:11
  • "Is there really no way to see these new questions at all" I'm sure you can see a new question if you have the URL, if that's what you mean. Or do you mean from questions lists? Commented Oct 24, 2025 at 21:27
  • 6
    @TylerH I mean without having the URL. I want them discoverable. Some way of being able to have an overview of what is being asked. Commented Oct 25, 2025 at 5:04
  • 4
    Right now, we don't have any other way to find them other than being opted in and seeing them in the question feed. But robust user filtering is something we are working on, and eventually, the question labels seen in the mockups will work similarly to tag pages, where you can see all of the questions with that label on them. Commented Oct 27, 2025 at 18:40
  • @Hoid The "staging ground" label should be clickable and lead to a page showing only staging ground questions, much like a tag; and these new opinion-based questions should work the same way. Commented Oct 27, 2025 at 23:38
  • 1
    @KevinKrumwiede, is that just to improve visibility into the experiment, or for some other purpose? We are likely going to add the question labels as a kind of tag page that you can see all questions with that label, but I struggle to see the value in a sidebar option for it. But do please leave an answer on this post with the feature request tag to make your case, and we can give it proper consideration. Commented Oct 27, 2025 at 23:52
13

The Advice, Best practices or Tooling 'tags' disappear after (attempting) to change the regular tags:

enter image description here

The same appears when editing a 'question'.

1
  • 4
    Thank you, I discovered this myself as well. We have it logged. Commented Oct 29, 2025 at 14:03
13

Ugh. Just as they did with Discussions, SO is sabotaging opinion-based Q&A with a deficient user interface. We can't even comment on responses.

Advice Q&As should use exactly the same format as any other Q&A. Advice could simply be a tag that people can ignore if they want. Or Advice could be its own SE, as Discussions should have been.

7
  • 2
    I just noticed that there are those ugly "advices" everywhere. Why are they trying to build into Q&A some forums? Stupid. Commented Oct 30, 2025 at 10:54
  • 6
    Yes, this is my thought too. The UX genius of the QA-Format forces the gist of the page to be directly responding to the question, instead of becoming an endless, meandering discussion with drifting topics. Why would you not want the same for opinion-based questions? SE sites which by their nature often don't have a single correct answer (cooking, parenting, workplace, christianity...) all use this QA UX because it makes sense. Commented Nov 3, 2025 at 11:09
  • 1
    Separate site yes, tag - no. That just mixes "advice"/"opinion-based" with the general questions, diluting the latter unnecessarily. Commented Nov 7, 2025 at 12:48
  • 1
    @einpoklum If you don't want to advice questions, just ignore the tag. This leverages the existing mechanisms for correcting a question's tags. Commented Nov 12, 2025 at 15:13
  • 1
    @KevinKrumwiede I think what you want is meta tags, they existed a long time ago. subjective was one of them, technologically they were just normal tags. Here is an explanation of why they got removed: stackoverflow.blog/2010/08/07/the-death-of-meta-tags Commented Nov 16, 2025 at 1:37
  • and no, I personally don't see why meta-tags were harmful. Maybe the information they contain should not be in a tag but in some other system. But I feel like SO could have avoided a lot of the loss of activity if posters could use beginner and subjective and so on to (1) apply different moderation rules to their question and (2) allow other people to ignore them without kicking them off the site. But what do I know, I wasn't there back then. Commented Nov 16, 2025 at 1:40
  • @julaine "I personally don't see why meta-tags were harmful." The linked post makes that pretty clear (it's not exactly a matter of opinion). "apply different moderation rules to their question" Sure, if everyone gets a veto on their own questions, what could possibly go wrong? The experiment already gets used mainly for regular non-opinion based questions, because that bypasses the rules. Then stricter rules for others, wouldn't that be discrimination, just formalised? Also, nobody was ever kicked off the site for being a [beginner] or posting [subjective] questions. Commented Nov 16, 2025 at 2:42
13

Duplicate detection has a bug – opinion-based questions (e.g. "General advice / Other") should not be detected and blocked as a duplicate of a regular question ("Troubleshooting / Debugging"), or vice versa.

I posted a question --- let's call it "Question Q in Post P1" to avoid ambiguity --- as General advice / Other without understanding how this (new, poorly documented) feature works. (https://stackoverflow.com/questions/79808219/never-mind-this-accidentally-ended-up-in-the-advice-bucket)

I deleted the original Post P1, and went to post Question Q as a new post, as Troubleshooting/Debugging. SO blocked me from doing this, claiming the new post was a duplicate.

I then tried editing Post P1, so that it was no longer contained the text of Question Q, but SO still blocked me from creating a new post of Question Q, presumably because Question Q was still in its history.

To get around this, I finally had to create Post P2 with half of Question Q, as Troubleshooting/Debugging, and then when it was created, immediately update the post with the rest of Question Q.

Please remove this irritating barrier to posting a question.

6
  • 11
    another way of looking at this is that it would help if there was an easy-enough way to change the type of the question post. Commented Nov 4, 2025 at 21:13
  • @starball Especially for the owner of the question and until answers or replies have been created. Afterwards it becomes messy. Commented Nov 4, 2025 at 21:52
  • 4
    I'd be much happier if the type was changed to "Classic Q & A" rather than "Troubleshooting / Debugging" and all the other choices had some kind of asterisk with a note saying something brief like "Discussion-based" and a link to the implications of choosing those types Commented Nov 5, 2025 at 22:40
  • Or better, if after selecting from the current "type" dropdown, the asker got a "for Opinion questions, we recommend our new open-ended discussion format" and then a choice between the two formats (with the recommendation pre-selected). That way the asker can use the question format they choose. At the very least, the question format should not come as a surprise (this could be fixed by showing "Replies to your question will use our new open-ended discussion format" right under the "type" selection, allowing the user to change their selection if they don't want that). Commented Nov 7, 2025 at 19:16
  • I had the same problem, except a moderator deleted the mistaken question for me, so I cannot even undelete that myself. Commented Jan 8 at 9:48
  • The workaround does appear to be creating a dummy question, then editing it to be the actual question you want. Commented Feb 4 at 17:52
12

Removing downvotes on opinion-based questions (which is a signifcant portion of what people try to ask, according to your metrics), just protects precious new users who don't bother to read the rules and jump straight to SO without using Google. If people only see positive feedback, they'll just keep posting junk and it will reduce the overall quality on the site. Neutral, detailed discussion about experience-based and opinion-based questions is great, but people should know if their posts aren't being received well, or we risk new users who don't follow or know community standards.

(This might seem hypocritical since I am listed as a new user here, which is because I don't post very much - I have read answers on here for a long time and am more active on other sites, though, and have really low rep mostly because of bounties.)

2
  • 3
    remove downvote on opinion-based ??? huh? why? Commented Nov 14, 2025 at 1:21
  • 4
    @LãNgọcHải wdym? they removed downvotes from the new discussion style opinion based question interface. SO removed it "to create a more welcoming environment" Commented Nov 14, 2025 at 16:16
11

Rebrand "opinion-based" as "experience-based" to better express why this kind of Q&A is so incredibly valuable.

13
  • 9
    I like this idea! The whole discussion is pretty ironic because the majority of technical answers here is already opinion/experience-based as many things can be implemented in multiple ways and most of the time there is no single definite answer so every one of them shares their experience about how to solve the question. I wonder why this even needs to be debated and SO does not just admit it's how it always has worked. Commented Oct 23, 2025 at 17:07
  • 5
    This is honestly a good point, IMO – if we want to still retain high-quality knowledge in this new format, we should emphasize the kind of responses we'd like to see: "good subjective" ones, supported by experience or other evidence. (Again, this is just my personal opinion; I'm not working on this project at the moment.) Commented Oct 24, 2025 at 19:32
  • 6
    @t3chb0t You are incorrect. The majority of technical answers here are objectively testable as correct or working, because that's the kind of question we allow here. Commented Oct 24, 2025 at 21:31
  • 3
    @TylerH so why allow other answers to even be visible after OP has accepted one? According to your non-discussion theory of SO, other answers should be considered irrelevant and hidden, yet they remain visible and are often higher voted for than the one with the green tick. Can it be that it's about opinions after all? Commented Oct 24, 2025 at 21:37
  • 6
    @t3chb0t Because there can be multiple technically correct answers to a question. I recommend you read the tour page if you aren't familiar with how Stack Exchange sites work. "According to your non-discussion theory of SO, other answers should be considered irrelevant and hidden" No, this is also incorrect. Commented Oct 24, 2025 at 21:40
  • @TylerH each technically correct answer is an opinion. Commented Oct 24, 2025 at 21:46
  • 7
    @t3chb0t I guess you don't understand what the word 'opinion' means. Commented Oct 24, 2025 at 21:50
  • @TylerH "Correct" and "working" are trivial and not interesting. "Best" is difficult and interesting. Commented Oct 25, 2025 at 14:17
  • 5
    @KevinKrumwiede Cool, there are lots of other sites where you can ask those questions. Stack Overflow has never been that place. What exactly are you trying to respond to, there? Also, if you think "best" is difficult or interesting, you and I have very different view of what quality content looks like. To me, "best" is code for "I have been lazy and put zero effort into understanding the problem before me, and also put zero effort into describing it for you to answer". I have never seen a question here of any worth that simply asks for the "best" X. Commented Oct 27, 2025 at 14:21
  • 2
    @TylerH SO was that place and it was better then. Your (collective your) obsession with objectivity has made SO a joke among real developers who get things done better, faster, and cheaper. Commented Oct 27, 2025 at 23:34
  • 1
    Strongly agree. SO was that way and was better. If current standards were applied to old questions, many good questions would have to be deleted. My opinion is there isn't a problem with the questions or the rules, only the way they are now being applied. When that is suggested, the response is that SO has changed over time; but I don't find that change positive. Sometimes we don't know/understand enough to ask just right, and the experience-backed answer clarifies the real problem and gets the discussion to "this is what you should be asking/considering," as seen in many older questions. Commented Nov 13, 2025 at 2:49
  • @Gary "and gets the discussion to "this is what you should be asking/considering,"" Yes discussion (or helpcenter on how to ask), not a question. The requirement for minimal understanding was already dropped because such discussions are either XY-problems, too broad, opinion based, or lack details/clarity. The outcome only applies to individual cases but that's not what Stack Overflow is for (and never was). "SO was that way and was better." A massive activity/moderation mismatch in the past may have contributed to this helpdesk misunderstanding, but it's just catching up with the backlog. Commented Nov 16, 2025 at 3:39
  • @user4157124 You're retconning what SO actually was in its early days. It was what we're asking for, and it was better. Commented Nov 16, 2025 at 16:01
11

For this experiment, we will only ever show the thumbs-up count

Somehow this entry has negative thumbs up. As I understood it should always be >= 0.
Always reminds me of the joke when the police officer teaches his son of the negative numbers:

There are 10 people on the bus. At the stop 14 of them get off. How many should get back on the bus so it is empty?

Advice of negative thumbs up

11

Is this a good idea?

First of all, I'm all for opening up the scope of SO for other types of programming questions, as was the intention here. The "primarily opinion-based" close reason is often abused and outright smelly, used when certain users want to dictate what type of questions that are allowed, rather than to prevent questions that are inherently bad.

"This question might attract bad answers". Yeah, so? That's a problem created by the person who posted the bad answer, not by the question. Suppose it can also attract great answers. I'd much rather have a good question with a good answer and x deleted answers, than a closed question with zero answers.

Some users will disagree with that and that's a cultural problem of this site. The purpose of this site is Q&A for programmers. The purpose of the site is not to moderate questions; that's a side effect. We have a bunch of users who do 100% moderation and 0% Q&A, and while they mostly do a good job they also create bad site culture, as they have forgotten what the site is about.

So yes, opening up the scope to different types of questions is a good idea.

What is the meaning of "opinion-based" and what is not opinion-based?

Everyone will repeatedly tell you that opinion-based questions are a poor fit for Q&A and that's because they are correct. An opinion-based question goes like "What's your favorite programming language?". That's just a trash question and not welcome here.

Whereas "How do I re-design this class in Java to make it properly object-oriented? [code follows]" is not an opinion-based question. Object-oriented design is a mature concept, it is not particularly subjective and it is certainly not opinion-based. Furthermore there will be encouraged practices specific to the Java language. All of this can be backed up with references and tons of books have been written about it.

So we should conclude that "opinion-based" is a plain bad term and the term itself is likely a significant part of the reasons why this whole experiment too is met with great skepticism from the community. I would much rather use terms like "big picture issues", "best practices" or "de facto standards".

We have to recognize that there are big picture type of questions that are, contrary to SO site culture, not opinion-based, but industry de facto standards for how to do it proper do exist and can be used for references.

Typical examples of this is program design, coding style and version control. How to do these things correctly is not a matter of opinion. Oh sure you can just go la-la-la-lets-improvise and create some abomination based on personal subjective whims. But if you do that in a professional context you will lose your job.

Specifically, this is the very difference between software engineering and hacking away aimlessly. In engineering, nothing can be left to chance, personal opinion or aesthetics. Programming is not art, it is a craft. And as with any craft there is established practices learnt from experience, or in some cases even based on scientific research.

In software engineering, you will have to design your program in an established way. You will have requirements gathering. You will have methods to pick the tools most suitable for the project. You will have coding standards and coding style standards. You will hopefully have version control and testing practices.

Ok so maybe you don't work systematically like this, but that doesn't mean that others don't, nor does it mean it's all subjective and a matter of opinion. I think a lot of users have built the SO site culture based on horrible work environments that didn't practice software engineering - if you don't even realize that there is a widely recognized correct way to design a program, then naturally you will think program design is all subjective whims.

How did SO screw up the implementation of a good idea this time?

The biggest problem with the implementation of this whole idea is probably the poorly considered question categories. These need to be broken down into something that actually makes sense for programming Q&A. Namely, the kind of big picture questions that would actually be a good fit for a new SO with a somewhat expanded scope.

  • Advice. Seriously? Did you ever meet a person asking a question who aren't at the same time asking for advice? It is a tautology. This just needs to be immediately removed - it is a pure nonsense category. Which we can already see - most questions posted under "Advice" should actually be plain classic Q&A questions.

  • Best Practices. Those who have experience from opening up a Q&A site to these type of questions will instantly spot the problem here: define "best". This would have to be broken down into actual programming terms in order to make sense. "Best" could be one of: fastest, least memory consuming, most readable, most portable, most maintainable, and so on. So use categories like that instead.

  • Tooling Recommendation. I'm not sure this is a great category either. It could be, in terms of "how do I design and handle this little detail of my repo in version control". Or it could already be a good fit for Q&A: "How do I make a commit in version control x?". Or it could be awful, opinion-based crap like "Which IDE is best?".

    In addition, a whole SE site https://softwarerecs.stackexchange.com/ was dedicated to these kind of questions, so why not simply re-direct these questions there? That site does pin-point what makes a good tool recommendation question (define "best"):

    Good software recommendation requests have two components:

    • A purpose — a task to accomplish, a user story
    • Some objective requirements — a minimum set of features

How to fix - what categories would actually make sense?

I would break down "best practices" into Program Design, Optimization, Readability & maintenance, Portability. Where Optimization and Portability are already covered by the main Q&A so we actually don't need to invent new categories for those.

  • Program design. This is perhaps the most important thing you ever encounter as a programmer. It is not subjective at all. However, pretending that program design does not exist or that it is somehow a matter of opinion, on a site dedicated to programming, is plain delusional.

  • Coding style. Matters of formatting, indention, white space, snake vs camel case etc. As well as language-specific concerns.

  • Readability & maintenance. How to write source code so that it is readable and properly maintained. Again, not really subjective at all, well-established practices exists in all manner of contexts.

  • Development process. Any technical questions related to the program development process.

  • Software life cycle management. Big picture questions about build configurations, version control, deployment etc that don't fit classic Q&A.

I'm sure there are other similar categories that ought to be on-topic too. We don't necessarily need to perfectly nail all of this from the start, maybe more categories will emerge over time.

3
  • 5
    This is what the Software Engineering/Programmers site was supposed to cover, but they too have the same cultural problems as SO, rejecting software engineering questions and thereby managed to make their own site superfluous. If SO manages to implement this proper then maybe we can finally close down that site. Commented Nov 7, 2025 at 13:28
  • 3
    I always thought that people would instinctively always choose opinion-based as question type if it were allowed just to be on the safe side where nobody can attack them. If the asker isn't sure or if the community isn't united about the type of a question, wouldn't it be better to simply not categorize these questions? We could simply remove the opinion-based question close reason and that's it. Or can the categorization be done with high accuracy? Commented Nov 7, 2025 at 17:14
  • 1
    I agree with this, except--Being in a category you suggest doesn't make a post on-topic or good. So the question is begged, what are reasonable effective generally & easily understood & identifiable properties of questions by which to triage & close vote & score vote. We need to measure/judge answerablility & searchability/findability, for purposes of maintaining a library & of just plain helping people (presumably those 2 being the site's ultimate goals). Commented Nov 7, 2025 at 20:42
11

With regards to closing such questions, here's an interesting case:

Python for finance

I have a question regarding core finance using logical reasoning in it:

A stock rises by 200% in the first year. Next year, it falls to 1/3rd of its value.

Is the overall percentage change over the two years +100%, 0%, or –50%?

This is so not programming related, but also not obviously spam. Extrapolating this trend, people will be talking about car repair here next. How do you propose this should be dealt with?

1
  • 5
    The answer uses python code, therefore it's obviously programming xD Commented Nov 13, 2025 at 13:05
11

I cannot figure out how to Opt-out of this feature. I can't filter them out. I can't find a setting to turn them off. I can't downvote them. I can't ignore people who post them.

They are acting as a kind of spam forced down my throat. I guess I should flag every one I see as spam, as it is unwanted commercial content by stack overflow being forced into my feed? That seems reasonable, right?

How do I turn this off? I could write a filter on stack overflow to hide the html? I could log out from stack overflow and block it in my browser?

Every single time I click on one of these questions and don't notice the advice tag, it feels like I'm being tricked to eat rotten garbage. Before I realize this is "Advice", I'll scroll down a bit and see a wasteland of incoherent comments that should be flagged for deletion and inconclusive discussion. The comments have no way to figure out if they are nonsense or not or to fix errors or any reason to think any errors will be fixed by someone else, so I skip over them once I realize they aren't terse single issue things.

So then I'll try to find the actual good answers and discover they don't exist and I've been tricked into drinking human puke instead of finding a stack overflow question.

This isn't a pleasant experience, and I want it to stop happening.

It says we "will" be able to opt-out. How do I do it now?

7
  • Opting out is currently broken, it will be fixed in 6-8 I guess. Commented Nov 18, 2025 at 14:45
  • 1
    Where did you find one of the opinion questions? Was it via search or something else? Commented Nov 18, 2025 at 14:47
  • 4
    "a wasteland of incoherent comments"? Are you referring to the 3X increased engagement, and how much users are being helped by this new format compared to regular Q&A? Ref meta.stackoverflow.com/questions/435293/… Commented Nov 18, 2025 at 14:48
  • 2
    @Hoid One doesn't have to search for them. The primary page I use on SO is a tag based search stackoverflow.com/questions/tagged/c%2b%2b and these questions pop-up there. Commented Nov 18, 2025 at 14:49
  • @cigien Sure, great for them? I'm not asking to delete the feature, I'm asking to hide it so I don't have to look at it. Lots of people enjoy Twitter, but I wouldn't want it to be added to the SO feed either. Commented Nov 18, 2025 at 16:43
  • sigh I was being sarcastic, which I thought would be obvious, but clearly it wasn't :( Not very surprising in text form, of course. To be clear, I agree that the feature is problematic, and the comments on the opinion posts range from occasionally useful, to complete crap/irrelevance, with no way of distinguishing between them. There should absolutely be a way to opt out of this, but it doesn't appear to be on the cards. Commented Nov 18, 2025 at 17:59
  • 2
    @Hoid I don't want to see them anywhere, and I haven't recorded why I'm seeing them. When I realize I've accidentally clicked on one and wasted my time, I close the tab and usually don't want to be on SO anymore after the bad taste in my mouth it leaves. How do I remove them everywhere? Commented Nov 18, 2025 at 18:24
11

You told us:

As you can see in the mockups, the key change is a simple ”type” selector on the Ask Question form. We ran some research with users to confirm if they could successfully and consistently label these questions based on the question labels we offered them. Based on our research questions, questions were labeled correctly about 90% of the time

At this point, I think it's safe to say that your research methodology was flawed because, in reality, a vast majority of users are selecting the wrong type.

Others have pointed out the harm this is actively doing.

My questions:

What steps are being taken to review and analyze what led to using this flawed methodology, and will the lessons learned from that review change how experiments are researched and evaluated for success in the future?

Also, can you share why this particular research method was chosen and what caused it to be so inaccurate?

30
  • 2
    I would quibble with it being flawed; we are just seeing a different outcome than we expected. Our main takeaway is that the type selector UI isn't the right move. Getting askers consistent with SO classification in this way clearly isn't it. That being said, we are reaching out to mods and active curators to meet with our PM and Designer to get their thoughts on what we think the next step is gonna be. Once we have enough feedback confirming the next step we want to take, we will share it publicly. Commented Nov 21, 2025 at 22:08
  • 9
    @Hoid Whether you agree with the term "flawed" or not, you made decisions based on research that determined users would be very good at categorizing their questions. The opposite is actually happening. Why is there such a large gap between what you expected per your research vs what is actually happening? And what steps are being taken to ensure the research you're basing your decisions on is reflective of what the actual results will be? Commented Nov 21, 2025 at 22:22
  • 1
    @devlincarnate could just be that the people they tested with before experimental public release are different than the people using the experimental public release. this is an alpha experiment, and I think it's commendable that they did internal testing / dogfooding before the alpha release (analogous to the kind of thing we've requested at times such as in feedback to An experiment on pairing askers with experts in the community. what I think matters now is how the company responds to present observations. Commented Nov 21, 2025 at 22:56
  • 3
    @Hoid "Once we have enough feedback confirming the next step we want to take, we will share it publicly." Which sounds good to me. The only thing I don't understand is why you cannot turn the feature off during the time you need to figure out what to do. To me it looks like you accept the damage being done in the meantime. I wouldn't do that. Commented Nov 22, 2025 at 8:05
  • 2
    @starball "could just be that ..." Yes could be although one always should try to assemble a test group as similar to the target group as possible. Or the test group was simply too small. Or maybe they gave additional guidance to the testers that isn't given now. This answer here aims at finding that out, not the least by asking for sharing methodology. The difference in test accuracy and actual accuracy is unfortunately quite large. If it would be smaller in the future that would greatly help any experimenting. Commented Nov 22, 2025 at 8:14
  • 1
    @Hoid Which if the case draws the entire concept of closed research for low tier/new users into question, since under this hypothesis, those most likely to cause issues, are the least likely to be part of the research group. Commented Nov 22, 2025 at 11:10
  • 1
    @NoDataDumpNoContribution I could imagine that there are people who fail, but that the traits that make it easier are likely to correlate with how likely people are to participate in the companies research. Commented Nov 22, 2025 at 13:33
  • 1
    @NoDataDumpNoContribution I think "damage" is a very murky term to throw around at this point. Is it consistent with how SO has operated for most of its existence? No, but I would say a lot of askers probably enjoy this if not plenty of people who have been to nervous to write answers. It's not an ideal end state, but I think we will land on something that works for everyone. At this point, we have more of a UI problem to solve rather than an idea one. Commented Nov 24, 2025 at 19:27
  • 2
    @user1937198 The test group was decided on a random selection of users who are signed up for research emails and responded to a request to participate in one. I didn't recognize any of the participants when I had a look at their profiles after the fact, but they seemed a healthy mix between established and newer users. Commented Nov 24, 2025 at 19:28
  • 2
    'Signed up for research' and 'responded to emails' are both factors highly likely to bias against users with low platform engagement. So the random sampling doesn't alleviate that bias. Commented Nov 24, 2025 at 19:50
  • 2
    @Hoid did the test group include any users currently under question bans? Or what percentage had never posted on the platform or had only posted once? Commented Nov 24, 2025 at 19:53
  • 2
    My hypothesis is that stackoverflow has a major learning cliff, and any user who is more than 25 rep would be very likely over that cliff, regardless of if they have 30 or 30k. Commented Nov 24, 2025 at 19:58
  • 2
    @hoid did your sample group include anyone who had never posted a positively scored post? I would be be curious what the results of the same experiment, conducted without any trappings of stackoverflow, on a user group sampled from reddit tech communities would be. Commented Nov 24, 2025 at 20:03
  • 5
    @starball as evidenced by the feedback on this meta post from the get go… simply asking meta before making it live would have been a good start. Commented Nov 24, 2025 at 21:15
  • 3
    @starball if this is the result we can expect from their current ways of obtaining feedback… they are failing. I would have been more than happy to provide feedback on this feature before it released, but I refuse to participate in zoom calls after the fact. It feels more like a hostage negotiation. Meta is always here, all they had to do was post this very feedback post a week or more early, or even just bring us in on the conversation of what categories to put in the dropdown. Anything. Commented Nov 24, 2025 at 21:47
11

I don't like that my "comments" there, show up as answers in my profile.

4
  • 4
    Well, they're not comments. See Weijun Zhou's point about the lack of comments, as well as my point about the lack of comment shorthands. Commented Feb 1 at 15:51
  • 9
    @wjandrea They're certainly not answers, either. Commented Feb 1 at 20:12
  • 2
    Yeah, that is an improvement we have planned, as is it makes the answers list on your profile to hard to read. Commented Feb 2 at 14:43
  • @Hold Are you going to do, like, top-level ones show up as answers but nested ones show up as comments? That would work for me personally. Commented Feb 21 at 20:01
10

Will questions in this new opinion-based category award reputation to users from 'thumbs up' reactions?

You don't explicitly mention this, just "removing features like the traditional reputation/voting model" which is vague; it doesn't say how it will work, only that it will be different. So... please say no. Or at least, not 10 points like questions get now. The reason we give reputation to questions, especially now that it's equal to rep for answers, is because asking good questions takes effort. Opinion-based or discussion questions do not really take the same level of effort; anyone can ask "what's the best way to do X" or "what are some good tools to do Y".

Will questions in this new opinion-based category qualify for the same badges that current, 'normal' questions get?

IMO there should be separate badges for these questions because of the much lower level of quality they will be.

While we won’t rule out some form of question closure in the future, we want to focus on improving the current process here to something more constructive that encourages the asker to refine their question in a different way, rather than the current closing process. Once we have determined whether this initial experiment has been successful, we will reassess the closing of these types of questions and how that is communicated to the asker.

So, does this mean you won't be able to vote to close or flag to close these types of questions at all? I'm thinking from the perspective of "what do I do if I see one of these questions in the Close Votes Queue? Obviously they can still be flagged as spam or rude/abusive, etc., but I want to clarify what the expectation is since you didn't say how it would work, only that you don't like applying the current closure system for these questions.

7
  • 3
    I believe Hoid has since clarified elsewhere in the thread that no, no reputation will be granted by this feature as it currently stands. Commented Oct 26, 2025 at 16:13
  • 1
    V2Blast is correct that this won't impact rep. It was mentioned explicitly in the first post talking about this. Opinion questions will reward badges. Though I believe it will only do so for non-rep-related badges. I will have to confirm. For now, these won't show up in review queues. Commented Oct 27, 2025 at 19:22
  • @Hoid Thanks. I had to go back and look again for the explicit mention about rep, and all I can find is "The UI for these new question types is intentionally simpler, removing features like the traditional reputation/voting model and the concept of an accepted answer." which is not particularly clear or noticeable. It mentions the UI won't include those things, but reputation is not just a UI feature. If you mean that there will be no reputation awarded at all for the new type of question, I recommend rephrasing that (and adding some header formatting to the section for better call-out). Commented Oct 27, 2025 at 19:32
  • 1
    First bullet point after the second paragraph I have written: "We will not tie these new question types to reputation or privilege-earning opportunities. We are open to considering a new incentive system for them." Commented Oct 27, 2025 at 19:35
  • @Hoid Sorry, I'm not seeing that phrase anywhere on this page except in your comment just now: i.sstatic.net/Dd3nPQz4.png Did you maybe say that in some other post? Commented Oct 27, 2025 at 19:38
  • 1
    Sorry for the confusion, this was mentioned in the very first post about the experiment Commented Oct 27, 2025 at 19:41
  • @Hoid Thanks, it's been a month so basically a lifetime since that, I had forgotten about that detail from the original idea post. Commented Oct 27, 2025 at 20:34
10

When an answer comes in while a question is opened in a tab it let's me load those answers, but the UI on that answer isn't quite right.

  1. It's not correctly sorted, it appears on top instead of at the bottom of the answers
  2. It allows me to add a comment to the answer which also shows on that loaded-after-the fact answer, but doesn't show on a fresh load of the page.

answers are sorted incorrectly and comments are available on a new style Q/A

10

The design is plain annoying: the people that care make sure to select the section, they think applies best to their question. But actually, those people are the ones who do ask well-received, on-topic questions (or can be easily taught to do so).

The people who simply dump their codes etc. go with the default and end up on QandA.

Do something about opinionated questions, if you insist, but this experiment isn't a useful solution. It causes more confusion than help.

10

Can questions that should absolutely have been asked in the normal Q&A style be converted from open-ended after the fact? By moderators (might be too much work for them), by the voting system for higher rep users, whatever?

Stuff like How to determine a lock-free integer type with a given minimum size? is not opinion-based, and has concrete answers that really should be subject to normal voting, yet because it was opened as an open-ended question, finding the useful information is much harder. I'm seeing more and more "it's not a discussion, just a straight-up useful technical question that's opting (intentionally or not) out of useful ranking of answers", and it's frustrating to have to scan some arbitrary number of wildly varying quality answers start to finish to figure out if the question has a reasonable answer yet.

8
  • 3
    We are currently deliberating the best way to change question types. A few things to figure out, like do existing answers carry over, are they turned into comments, if they are turned into answers, do we give them rep, etc. Who can do it, etc. Commented Nov 12, 2025 at 15:38
  • 7
    this will create frustration, when people's low quality posts are converted to regular Q&A then closed/downvoted/deleted. Commented Nov 12, 2025 at 18:46
  • @KevinB: I'm not arguing crap should be converted. But I've run into several perfectly decent Q&A questions inexplicably posted as open-ended questions, that would be up-vote-worthy and potentially useful if they weren't the unordered chatroom-style question. Commented Nov 12, 2025 at 20:48
  • I would do the conversion as quickly as possible, optimally before any replies take place. Otherwise I think the working and the expectations of the two formats are too different to enable a successful conversation. Even the question might simply not be up to standard. The problems already start with "should have asked". Who knows if the OP actually agrees. A forced conversion against the will of the OP or those that replied might spell disaster. Commented Nov 13, 2025 at 7:36
  • 4
    Better in my eyes would be simply closing as not opinion based and giving the OP a chance to try again in the other section. We only need closing for that. Commented Nov 13, 2025 at 7:38
  • Agreed with NoDataDumpNoContribution. That's what is working best for migrations currently. Directly migrating has proven to be difficult to get right time and time again. Maybe a fast path if several people agree to convert, just close by default. Commented Nov 13, 2025 at 10:08
  • So note though that for them to ask it again, they would need to ask it with a new title. Not the end of the world, but it’s definitely an inconvenience Commented Nov 13, 2025 at 14:18
  • ^ At the least, I think the duplicate checker (or one of them, anyway) does need to be adjusted so it's not detecting one type of post as a duplicate of the other, and thus preventing a user from legitimately reposting a Troubleshooting / Debugging Q&A as an opinion-based post or vice versa. (We've gotten a few support tickets from users encountering this issue.) Commented Nov 22, 2025 at 19:36
10

I'd like the ability to click labels on the "questions" page and see only questions with that label, and/or search syntax for labels similar to the syntax for tags. Basically, sometimes I'd like to see all advice questions and only advice questions.

5
  • 3
    Basically the same as this answer by VLAZ Commented Oct 28, 2025 at 6:12
  • 2
    @WeijunZhou I was specifically asked to make this an answer by a mod in the comments on that very answer. Commented Oct 28, 2025 at 14:59
  • Sorry I've missed it, but it's not properly tagged as requested by the staff. I will edit it in. Commented Oct 28, 2025 at 15:57
  • 2
    This planned functionality is to be added, and I believe it's already being worked on, but I will double-check to see where it's currently at. Commented Oct 29, 2025 at 15:36
  • 2
    @Hoid I hope you don't make a decision about the success or failure of opinion-based questions before this feature is added. Commented Oct 29, 2025 at 18:32
10

Infinite vote glitch

Steps to reproduce (on Chrome):

  1. Vote for an answer.
  2. Click a link on the page.
  3. Use the browser's Back ⬅️ button.
  4. The vote count remains in the state before you voted (e.g. 0 in the example below).
  5. Return to step 1.

When you finally reload the question page, the vote count will have increased (or presumably decreased if you down vote) by one for each time that you voted.

I discovered this by accident yesterday when I misclicked and then I did it again to check it was a real bug. Please don't ban me for voting irregularities.

infinite vote glitch

2
  • 2
    It persists after a refresh?? I appreciate there's no rep being awarded for these, but still, some validation on the backend seems like a good idea. Commented Nov 11, 2025 at 16:03
  • 3
    Thank you for letting us know! Rest assured, no ban to worry about! We have just released a fix so it'll now just allow 1 vote per post. Commented Nov 14, 2025 at 15:03
9

Are these posts going to be integrated into question lists the way you've presented above, where they organically get included if they exist and are recent enough to fit into the existing list, or will it be more like SG posts and the watched tags feature where the posts get inorganically injected into the list despite not being more recent/relevant than the things they are displacing?

3
  • 2
    They will be included organically, just like regular Q&A. Under the hood, these are basically identical to regular Q&A, except for the vote type, which is different and unique to opinion-based content(OBC). That being said, for feeds that monitor the score, which I don't have a list of off the top of my head, the score will factor in. Commented Oct 22, 2025 at 16:13
  • 1
    @hoid Will it be discoverable in Google search? Commented Oct 22, 2025 at 19:47
  • 2
    @Dharman Yes. Since its still regular Q&A for the most part, it will be crawled the same way Q&A. Should the experiment progress, we will look at this more intently to determine if we need to make some changes so that it can be signaled differently in Google search results. Commented Oct 22, 2025 at 21:03
9

I am glad to see the thought that has gone into this so far and how you've presented it here. Thank you. I share a number of the concerns already presented by others, which I won't repeat here. A couple questions:

  1. Will the standards for when to cast Spam and Rude/Abusive flags be the same as for regular Q&A posts? Or what might be different?

  2. Same question, but for AI generated content:

    • same policy? (prohibited)
    • same mechanism? (Other flags)
    • same consequences? (post removal, contacting user, account suspension if necessary)

    Or what might be different?

2
  • 1
    Speaking solely for myself (I'm not currently working on this project): For spam and rude/abusive flags, I don't see why the standards for flagging would need to be different from regular Q&A. And I don't foresee AI-generated content being allowed there either. Commented Oct 23, 2025 at 16:15
  • 4
    To confirm V2Blast. Standards for spam and Rude/Abusive are identical; there won't be any changes there. The same goes for your second point. Commented Oct 23, 2025 at 19:59
9

Voting on a deleted 'answer' gives the following text which contains an escaped apostrophe ('):

This post has been deleted; deleted posts can't be voted on

enter image description here

If you click it 2 more times, the text changes to

example image showing the text 'vote was unsuccesful'

0
9

I thumbs-upped this response a few days ago, but its thumbs-up count is zero:

Screenshot of the buttons under a response. The thumbs up is blue because it has been clicked, and the thumbs down is not because it hasn't. The count of thumbs up is "0".

3
  • That could also be caused by the vote total still being negative or zero. The displayed number is clamped to 0 at the lowest, but hopefully downvotes aren't discarded if the total was already zero. Commented Nov 8, 2025 at 16:31
  • 1
    @PeterCordes I was under the impression (from the OP here, the visual design, and other people's discussion) that downvotes do not affect the upvote total. (If that's not the case, then that's what I'd like clarified.) I have heard and seen, however, that spam or rude/abusive flags do reduce the upvote total, which technically could be the reason for what I saw here, but I thought pretty unlikely—it would be a very inappropriate use for those specific flags, in my opinion. Commented Nov 8, 2025 at 16:52
  • 3
    @PeterCordes I tested and downvotes indeed do not reduce the upvote total. (Apologies to the person I tested this on; I will try to remember to reverse it if/when that functionality is added.) Commented Nov 13, 2025 at 20:24
9

or

There's no way to use comment formatting, including shorthands like [edit] and @-tags like @wjandrea.

This hinders communication, e.g. giving advice to new users:

screenshot showing "Please don't post pictures of text. Instead, copy the text itself, [edit] it into your post, and use the formatting tools like code formatting", where "edit" is unlinked.

For reference: Markdown source: [Please don't post pictures of text](https://meta.stackoverflow.com/q/285551/4518341). Instead, copy the text itself, [edit] it into your post, and use the formatting tools like [code formatting](/editing-help#code). Link to post.

5
  • 1
    Not a bug, as these are not comments. They are, for the most part, regular answers that have been reskinned for this experiment. So anything you would expect from commenting wouldn't be there, but should otherwise carry all answer-related features. Commented Nov 6, 2025 at 14:37
  • 9
    @Hold Then the bug is that comments don't exist (which I think someone already posted). How are we supposed to give advice to new users otherwise? Commented Nov 6, 2025 at 14:57
  • 1
    Found it: Weijun Zhou already brought it up here Commented Nov 6, 2025 at 15:04
  • 2
    For now, in the replies. Commented Nov 6, 2025 at 15:25
  • @wjandrea: you ask a new endless discussion tagged [Advice?] and put a link to that discussion for the user who needs advice. Commented Nov 7, 2025 at 18:29
1
2
3

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.