-29

This quarter, the Community Enablement team is focused on researching the asking experience. This research has led the team to consider the types of questions allowed on Stack Overflow and how to expand the process to accommodate more types of questions. Over the next few months, one of our key focuses will be centered on finding a home for a wider range of technical questions on Stack Overflow. We've observed that many valuable questions are closed as 'opinion-based' because they don't fit our traditional format for objective Q&A. While this strict format is the foundation of our site's quality, it also means our library of knowledge has lost out on crucial technical discussions. Our goal, as a company, is to find a way to welcome more extensive questions while maintaining the high quality our community expects. Back in July, we talked about this a little bit on Meta Stack Exchange and the Stack Overflow Blog:

Simplified posting experience: We're exploring simplifying the process of asking questions on Stack Overflow, so that any technical question has a place and can find an answer. Our goal is to make it easier for all technologists to share their knowledge gaps, no matter how specific or broad the problem, while maintaining the quality of traditional Q&A. This means welcoming a wider range of questions and perspectives, ensuring Stack Overflow continues to be a comprehensive and supportive resource as technology evolves. We believe this will diversify the knowledge available on the platform and make it even more valuable for everyone.

In our upcoming experiments, we will explore allowing well-reasoned, opinion-based questions to sit alongside our traditional Q&A. This effort is to simplify how more users want to ask and answer questions that have not typically been allowed. This isn’t an attempt to diminish the valuable and high-quality questions and answers we have, but to expand the knowledge base for more types of content to exist alongside it. To be explicitly clear, here are some things we, as a company, are trying to do:

  • 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.
  • Users will not be forced to see this content; community members will have the option to filter their feeds to their liking. Or opt out of opinion-based content entirely.

In addition, here are some commitments we’re making prior to moving forward with this initiative:

  • We will work with the Charcoal team to make sure that all of these questions are exposed to their tools to help prevent spam. We’ve learned from our experiences with the Discussions project.
  • We will work with moderators to ensure that the appropriate moderation tools are available to moderate new question types if this becomes a permanent change.
  • We are focused on attracting any quality technical questions; we do not intend to attract generic homework questions, or frivolous conversations
  • We do not want to become a generic homework help desk.

I'd like to leave you with a few discussion prompts to consider, thinking about how you would like to see something like this come to life.

  • What kind of tools would you like to see that allow you to curate your question feeds to include only the content types you are interested in?
  • Opinion-based content allows for a considerable amount of room in defining what makes a good question. What guidelines would you like community members to see when asking these types of questions?
  • We don’t believe that reputation as structured today is the correct incentive for addressing these types of questions. How should quality discussion-oriented content be rewarded or incentivized?

We'll be back very soon with more details on how we might alpha test this. We will be monitoring this post for questions till October 16th, 2025

18
  • 4
    To be fair, if you are going to allow opinion based questions then they should be part of main Q/A. the easiest way to remove reputation is to make such questions and their answers as community wiki. I would not give options to remove such questions outside general public view as this makes moderation harder. Even if we would allow opinion based questions that does not mean that any kind of questions should be allowed in that category. Or just bring back Discussions for such content. There is no need to reinvent hot water. Commented 12 hours ago
  • 21
    This, admittedly, sounds a lot of Discussions. I know that Discussions failed, but some of the reasons for that was because it lacked the features that you state you would provide with such a release; working with Charcoal and giving mods appropriate tools were two significant reasons why Discussions failed with the community. I think you have a good foundation (admittedly, you may need to purge the existing content to a degree), but if you actually add those tools to it, I think it could be more of a success. Commented 12 hours ago
  • 2
    @DalijaPrasnikar, for testing and being able to iterate along the way, rep just seems like added baggage for this purpose. I think it especially makes the value of rep today murky. Not everyone views reputation the same way, but it's mostly aligned with being trusted in an objective manner; rewarding it on opinion-based content seems not ideal for what reputation tries to signal. So we are just not combining the two. Commented 12 hours ago
  • 2
    @DalijaPrasnikar in terms of what we are thinking of testing this gonna look somewhat like discussions and be directly added to the question feed with signals, probably something looking like a tag to indicate it falls under that its opinion based. We have spent a considerable amount of time looking at old closed opinion questions and think the value is pretty obvious, but a lot of those are the really good ones, so there is definitely gonna need to be thinking on what constitutes a good one and how we deal with ones that are hard to engage with in a meaningful way. Commented 12 hours ago
  • 1
    @ThomA Discussions and the follow up experiments gave us a lot to think about on what we would do differently to improve it, for example, we know that a decent number of questions are closed every day for being opinion based so we think the smartest point to capture those is on the ask page, where users think they should be asking them and then have their content be clearly labeled for what it is. I agree that Charcoal is a big one for this. Moderation tooling is another we are being considerate about, it's why this product team will start out helping with this content's moderation. Commented 12 hours ago
  • 6
    " We've observed that many valuable questions are closed as 'opinion-based'" Can you maybe quantify this a bit more. I counted around 30k such closed questions, which isn't a lot and how many of them are valuable I wouldn't know. Commented 12 hours ago
  • 8
    "our library of knowledge has lost out on crucial technical discussions" -- not buying it for a minute. We curate a library, as you say, of knowledge. Discussions are not knowledge. The fact that we (mostly) do not host discussions is among the primary discriminators of StackOverflow, and large among the reasons for its perceived quality. Commented 10 hours ago
  • 3
    One of the problems with having separate spaces is shuffling content and (users) back and forth. Having a single place for all kinds of posts makes that easier. If you want high quality opinion based questions then they can be part of main Q/A, or we need ability to effortlessly move content. In that light, the question is what are we going to do with existing opinion based questions which are closed or even deleted? All that needs to be taken into account before starting experiments. We need to have all those fine details worked out. Otherwise this will just be a huge waste of time. Commented 10 hours ago
  • 7
    As a user of Stack Overflow for nearly 2 decades, and as programmer with 2 decades of professional programming experience and an additional decade of programming as a hobbyist, I will never find a question seeking my opinion as helpful. Additionally, " has lost out on crucial technical discussions", my immediate response is that is a GOOD THING since Stack Overflow is not the place nor should it ever be the place to have technical discussions. Of course I fully expect feedback on this topic to be ignored. Feedback on similar changes in the past have been ignored by staff. Commented 10 hours ago
  • 1
    One could look at those closed questions and try to deduce which of them might be interesting if they were allowed to be answered. On the other hand I think it's more likely they could become normal ontopic questions, by removing or reformulating the opinion based part. That might be possible. If I have time I will try to demonstrate this in an answer. Commented 8 hours ago
  • 10
    "it also means our library of knowledge has lost out on... technical discussions." Yes. This is a good thing. They are not "crucial" at all. If you keep trying to take Stack Overflow away from Stack Overflow users, they will continue to take themselves away from you, too. Commented 7 hours ago
  • 1
    Your Question wouldn't have passed the 'Staging Ground' because of its "elliptic" Title: "Exploring new types of questions on Stack Overflow" => From the 'SG': "Please edit the title of your question to be descriptive, unambiguous, and specific to what you are asking. For more guidance, see How do I write a good title?." => Care to abide by the 'SG' Rules...? => ... And give your Thread a clear a descriptive Title about its Contents/Intent...? // I haven't voted yet, because not read the Post, ... that I find so "vague" because of its Title. Commented 4 hours ago
  • 2
    If you remove the "opinion based" close reason, I will just vote to close them for a different reason. "lacks focus" would be appropriate. Commented 4 hours ago
  • 1
    @chivracq The oxymoron "well-reasoned, opinion-based questions" would make the experiment's paradox painfully obvious given a descriptive, unambiguous, and specific title. Commented 1 hour ago
  • 1
    @user (with a Nb - apparently the 1st one on this Thread, you are still oxymoronly-lucky...! (You fully understand what "oxymoron" means, I hope...)) I like your Comment, even if I still haven't read the Post (Question) due to its "1-Rep User Level Title"... (Which I usually refuse to read and directly DV... (either directly or within 24h, => applies to this one (24h), even if it's already at -29, so I guess an extra DV won't (really) make a difference, ah-ah...) Commented 1 hour ago

9 Answers 9

Reset to default
23

Look at the history of the network, specifically the first few years. Purely opinion-based questions were allowed then, and they were quite popular. But they caused a lot of issues with the system, though many of these issues were related to reputation. Many of those questions were non-technical, so not quite what you're proposing here. But the problems are mostly similar.

There's a lot to learn about the potential pitfalls from the discussions of that time. If you exclude reputation gain, there are still big issues around voting. Those posts tended to attract a lot of votes, much more than any objective, technical questions. So if you merge them into regular questions, they will drown out any list sorted by score.

They also attract a lot of answers. More than you can effectively moderate. Not everyone knows the answer to a specific question, but everybody has an opinion. Nobody will read the late answers, which is problematic in terms of moderation and also not a nice experience as a user.

Established site rules don't work for them. So you need to change the rules for these, which will make the rules even harder to understand.

Duplicate rules are even more of an issue than for normal questions. Adding new answers to old questions is futile, as they attract too many answers in the first place. The opinion-based questions also get out of date more easily. So you might again need special rules for this content type.

I think you're underestimating again just how difficult it would be to make this work well.

0
21

Let me quote something from before:

As many of you know, Stack Overflow has mainly focused on objective content: questions that can clearly and definitely be answered. [..........] aims to expand upon this with subjective content.

[..........] threads may not have a specific conclusion with an answer that’s right or wrong. Instead, participants can discuss, debate, and explore varying opinions, perspectives, and implementation strategies in order to make more informed technical decisions.

We expect some [..........] threads to only be relevant for a finite duration in time, while others will become long-lasting points of reference. Our goal is for these conversations can be easily accessible and discoverable in both the short and long term. a

or

[..........] provides a way for developers to learn or share a new perspective outside of the traditional question-and-answer space, with a lower barrier for engagement and covering a broader range of topics. b

[..........] = Discussions

These are from the posts concerning launch and expansion of Discussions;

a. Discussions experiment launching on NLP Collective

b. Discussions: Learnings and experiments

Do you see the similarities? Basically, you are bringing Discussions to the main Q&A. They'll be a great source of cluttering the main. We have already pointed out the issues with Discussions, things that need to improve, and why and how they are different from Q&A.

Take this part of Tyler's answer about Discussions:

Discussions are not Q&A. They are a new product; users need not expect or rely on similar interaction options as Q&A (like voting). For open-ended, opinion-based discussions, voting never really made sense in the first place. Especially if it's not tied to any user reputation (which was the entire point of adding voting to Stack Overflow Q&A in the first place--to give the post authors reputation to signify their knowledge and effort on the site, and then also to unlock site privileges).

Now let me go through some of your commitments:

We will work with the Charcoal team to make sure that all of these questions are exposed to their tools to help prevent spam. We’ve learned from our experiences with the Discussions project.

Same promise was made about Discussions at some point. It never happened.

We will work with moderators to ensure that the appropriate moderation tools are available to moderate new question types if this becomes a permanent change.

Same as above.

So, I believe it when I see it.

21

Opinion-based questions by their nature do not lead to answers, they lead to discussions. The Q&A format, even with the new comment system, is not a good format to support discussions, especially when we have spent literally decades trying to train people to not use Stack Overflow like it is a forum.

Why is this not a separate product from Q&A with systems designed specifically to support discussion rather than trying to shoe-horn it into Q&A?

14

Opinion-based content allows for a considerable amount of room in defining what makes a good question. What guidelines would you like community members to see when asking these types of questions?

Are they actually going to be used? because we had guidelines for people asking questions, but they were replaced with AI slop a while back.

The question we need to ask ourselves is... what kind of opinion based questions are we not going to allow. For example, are we going to allow questions comparing different SaaS solutions, either by price and/or by feature? Are we going to allow "What is the best practice way to do X?", even when it has a properly worded "How do i do X" dupe that has a good answer?

What is an example of a good useful opinion based question that the team wants to exist here? It's clear (at least to me, given we've never pushed to change the guidelines to allow opinion based content) the meta community doesn't generally want opinion based questions here, so getting us to decide what should be allowed seems kinda... out of place to me.

7
  • 5
    Yeah, I think determining "what makes a good/high-quality opinion-based question" is sort of the key here. The SO blog post Good Subjective, Bad Subjective seems like a decent starting point. (That post also points out that what is now the Software Engineering Stack Exchange site originally arose specifically out of needing a place for more subjective programming/development-related questions...) Commented 10 hours ago
  • @V2Blast, I was just reading that blog post last week, balpha pointed me to it. Perhaps we could discuss it further in the near future to refine or adopt it as guidelines. Commented 9 hours ago
  • Presenting good guidelines is something we want to be able to offer, but presenting good guidelines and adherence to them are, of course, two different things, and you can only do the latter with the former. So we want to start with thinking about guidelines first. Commented 9 hours ago
  • 1
    I would personally not allow price comparisons; that's a personal and budget choice for an individual anyway. I can imagine that some kind of feature-oriented discussions, within the larger framework of accomplishing something, may have a some sort of reusable value, but that seems like the kind of thing that you know when you see it. So, it's not as easy to lay out, I have been spending quite a bit of time looking at discussion posts to try and get a feel for better guidelines or signals Commented 8 hours ago
  • 1
    @Hoid I feel a lot like that post, and the context behind it should have come out a lot earlier. Also, the two recommendations sites. The third factor is - how prepared the community folks are to step in and help moderate if things go bad, and winning over the community we have now. On a smaller scale, and with much more personal trust - I've tried (and failed!) to do that sort of thing on MSE when I was a mod. As much as the KPI seems on "new user numbers"- I think its worth looking at what it takes to win over the more active userbase to these ideas. Commented 6 hours ago
  • 1
    What's the unique selling point and benefits to the people who're on this site now and what does it bring the community we have, rather than the community we might? Commented 6 hours ago
  • it also doesn't really... solve anything does it? the gripe people on the outside have with SO is how strict content curation is. I don't think this will even begin to address that issue. Commented 6 hours ago
9

One quality standard should be providing details in the question. "I want to program, so what language should I use?" is a terrible question, regardless of objectivity/subjectivity, because most programming languages can be very useful for programming. And "in 2025" would not make it any better. People looking for opinions should share what truly distinguishes their needs, goals, or context.

7

What kind of tools would you like to see that allow you to curate your question feeds to include only the content types you are interested in?

I'd like to be able to use watched/ignored tags to curate my question feeds... but currently, watching tags (and enabling SG posts to appear) makes the question lists worse by randomly injecting tagged questions at the top, out of order from the rest of the content rather than simply highlighting this content.

But as far as this new content type is concerned.. we simply need a labeling system that supports both this new content type and SG posts, with a filter that lets you pick and choose which types are shown.

1
  • 3
    This is a requirement for this feature. We likely will start with using the experiment opt-out feature to keep people out of it if they don't want to see it. But the end state is to be able to let users have some sort of labeling system so they can keep all opinion-based content out of their experience, or maybe a specific set of filters where it's all blocked except for a particular tag. Commented 12 hours ago
6

Given this quote:

Simplified posting experience: We're exploring simplifying the process of asking questions on Stack Overflow, so that any technical question has a place and can find an answer.

In what way will this simplify the posting experience? Are you going to have some kind of AI inspect the question and determine whether or not it is opinion based and have it automatically route the question to being an discussion opinion based question? or... I'm just not sure how giving people two different kinds of posts they could make and making them decide would be simpler.

The simplest solution would be for there to only be one route, asking a question, and having our guidelines simply expand to meet whatever new criteria the company wants to force the community to accept. As long as this asking process involves getting the asker to choose what kind of question they are asking, it's going to have the same problems discussions had where people were routinely asking regular questions as discussions.

2
  • 1
    Maybe the community could reclassify a question if the OP chose wrongly? Commented 7 hours ago
  • that would be ideal, but doesn't work if these are stored as a different content type that better supports discussion Commented 7 hours ago
4

I'm super-wary about the concept. It seems at odd with the aspects of Stack Overflow that drew me here. But I'll run with your premise and offer a sketch of a possible design that might address some of my biggest concerns with this.

Focus on aggregating expert opinion. Focus on where you can add special value: questions that need answers from experts, and where Stack Overflow can collect and aggregate information and opinions from experts. Don't try to be everything for everyone. Don't try to be another Reddit/Quora/Yahoo Answers clone where anyone can post anything and anyone can post any answer they like.

Set clear expectations for subjective or open-ended questions. I would suggest that such questions must (1) ask about software development, (2) be practical and answerable, (3) be directed towards answers that can be supported by evidence or experience, and (4) ask a question that requires answers from experts.

Restrict who can answer. Require 250+ reputation to answer. This is designed to signal that the goal is to gather expert opinions, and the goal is to focus on questions that require aggregating expert views. Also, it helps reduce answers from people who just want to engage or chime in, from people who have opinions but no expertise, and from people who want to share their opinions but have no buy-in to the requirement for providing evidence or experience. Only rep from standard questions and answers should count towards this 250+ threshold; rep from these new subjective questions or answers on them should not count. If people want to engage with subjective questions, they should first contribute to the knowledge base and establish their credentials as knowledgeable in this domain.

Restrict who can vote. Require 250+ reputation to vote (same as for answering).

Require that answers be supported by evidence or experience. Set a clear rule that all answers must be supported by evidence or experience. Document this rule prominently. Delete all answers that don't follow this. Create a new flag type on answers, 'primarily opinion not supported by evidence or experience', and either (a) get buy-in to act on it from SO moderators or (b) enable the community enforce this, e.g., if there are 2 such flags, the answer is deleted. The tradeoff for lowering requirements for questions and making it harder to close or delete questions is higher requirements for answers and making it easier to delete answers.

Mark subjective and open-ended questions separately. Put a post notice above such questions that indicates it is a subjective or open-ended question, and different rules apply, and link to the rules.

Change the closure rules. I suggest disabling the "opinion-based" and "duplicate" close reasons on such questions, but allowing other close reasons to remain.

Change how answers are sorted and vote totals are displayed. Use Bayesian ranking to sort answers, to take into account aspects of the current system that don't work well for opinion polls. These issues aren't a big deal for most existing questions, because most existing questions don't receive a lot of answers or a lot of votes; but opinion questions often receive many answers and many votes, which creates new problems.

See https://stats.stackexchange.com/q/26833/2921, https://planspace.org/2014/08/17/how-to-sort-by-average-rating/ (also: https://stackoverflow.blog/2009/10/23/alternate-sorting-orders/, https://web.archive.org/web/20111003060234/http://blog.reddit.com/2009/10/reddits-new-comment-sorting-system.html, https://julesjacobs.com/2015/08/17/bayesian-scoring-of-ratings.html, https://www.evanmiller.org/bayesian-average-ratings.html, http://www.evanmiller.org/how-not-to-sort-by-average-rating.html, https://web.archive.org/web/20150214083151/https://camdp.com/blogs/how-sort-comments-intelligently-reddit-and-hacker-).

Currently, answers posted earlier and answers at the top of the sort order tend to receive more votes. So someone who posts an early answer gets more votes, which raises it to the top of the sort order, whence it receives even more votes. Late answers languish at the bottom, get little visibility, and don't receive many votes. So the current sort order is influenced by "which was posted earliest" as much as it is influenced by quality. Bayesian ranking reduces these issues.

Do I expect to like such a system? I don't know, it feels pretty different from Stack Overflow's existing mission, so probably not. There could be some value in opening up in a limited way. But, do I trust the company to implement this in a healthy way? Not at all, the circumstances right now are so dire for the financial future of the company that I don't trust the company's ability to resist the desire to maximize engagement regardless of quality. But I thought I'd do my best to sketch a possible way that is as consistent as I can make it with the goals articulated in the question.

2

With no rep for answers, will closing apply to opinion based questions the same as for regular questions? Is the community going to be expected to police comment answers, which closing does not affect? Or are comment answers or threads the goal here, instead of self-contained answers?

Also, you say stopping spam is a goal. However, the line between an opinion post and spam can be much thinner than for an objective post. When we see what could be marketing copy for a recommendation, do (or when do) you want us to try to remove it as spam? And perhaps more importantly, are mods going to be ready and supported in handling these flags?

1
  • 2
    Regarding your second paragraph: I suspect that generally, the kind of spam they're talking about here is blatant, off-topic spam posted by bots/bot farms – not simply users subtly (or not-so-subtly) promoting their own product. The latter generally happens at a scale that can be handled by community flagging and mods; the former tends to be overwhelming in scale and difficult to handle without staff assistance, if proper safeguards are not in place. Commented 10 hours ago

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.