-116

Update May 6th, 2025

After reviewing the feedback on the comment changes, specifically regarding the plan to disable the "No longer needed" (NLN) flag during this experiment, we discussed internally and decided to follow the suggestion from TylerH & Starship. Instead of removing the flag, we will rename it to "Not needed" for the duration of the experiment. We may also slightly revise the descriptive subtext to better align with both existing use cases and the experiment's goals.

Additionally, we are exploring adding an optional text field when this flag is used during the experiment. This would allow users to provide brief context for their flag, which could offer valuable feedback for the experiment and offer more context for moderators on those types of comment flags. We will provide further details as the experiment launch approaches.


TL;DR: Soon we will be launching an experiment on Stack Overflow to enhance the quality and usefulness of comments. Our goal is to make them more suitable for follow-up questions and conversations surrounding specific content, and shifting away from the notion that they are strictly temporary or suggesting post improvements. Based on learnings and feedback, we will test features such as threaded replies and code formatting, and introduce an updated UI. This is a temporary test to gather data and feedback before deciding on permanent changes. Should this experiment be successful, these changes will be released network-wide.

We're launching a new set of experiments on Stack Exchange, beginning with Stack Overflow, focused on evolving the way comments function. This experiment represents a significant shift from the long-standing philosophy that comments are ephemeral and intended solely for post improvement or clarification. We recognize that this change may be unpopular to some and welcomed by others. To be very clear, the primary purpose of this experiment is to discover ways to increase meaningful engagement across the entire network.

We aim to develop an experience that welcomes comments that are on- topic, professional, and beneficial to users. Since commenting on the internet is a well-understood feature, we’re also exploring how improvements to our commenting UI, similar to how commenting is commonly experienced on the internet, will affect engagement on Stack Overflow.

Why This Change? Learnings from Past Experiments

These experiments stem directly from what we learned in our recent experiments focused on comments and the 'Discussions' feature.

Need for Follow-Up Questions

A key takeaway from our previous experiment was that users want to ask specific, technical follow-up questions related to existing questions and answers. Our analysis showed common needs such as asking direct follow-ups, clarifying answers, sharing variations, or explaining why an answer didn't work for them (and why). These attempts often included code blocks or images, which current comments do not support.

Visibility Challenges

While we experimented with directing users to Discussions for follow-ups, we found it wasn't the ideal solution for these specific, contextual questions. This was partly due to lower visibility compared to the original Q&A page. Users also expressed confusion about the distinct purposes of comments, Discussions, and Q&A.

Current Limitations

The existing comment experience with its character limits and formatting restrictions creates friction for users who need help. It either requires users to create multiple comments or to demonstrate their own code in awkward spaces. Rather than trying to find a way to fit Discussions into this need, we are shifting our primary focus to better facilitate follow-up questions directly within the context of the Q&A posts as comments, where many users already try to engage in follow-up questions.

The Proposed Experiment on Stack Overflow

We plan to run a series of rapid tests that build on each other on Stack Overflow to explore ways to make comments more suitable for these types of exchanges. Much of this will be UI updates, such as:

  • A general update to the look and feel of comments, with an updated user card
  • Comment threading for easier to follow conversations
  • A full editor for commenting that supports images and code blocks
  • Some improved commenting moderator tools. We expect that we will begin work on this towards the end of our experiments, once we have a better idea of what requirements the experiments have surfaced.

To inform our strategy, we conducted interviews with developers who are regular users of Stack Overflow. These interviews affirmed the findings in our previous commenting experiment that users are encountering situations where they need to ask follow-up questions or need to clarify what is in an existing answer. Those individuals felt that our approach to enhancing the commenting experience could provide better context and encourage the back and forth needed to fully solve problems.

Please be aware that the screenshot below is a preliminary view and the final version may differ. The comments are expanded here just for demonstration purposes. Default would be to have them collapsed.

We intend to act fast, potentially testing variations on a weekly basis, while gathering your feedback and iterating solutions. We will provide updates to this post to communicate test start and end dates as soon as possible.

Screenshot of a web page layout of a new commenting experience on Stack Overflow. The main content area displays an answer block, followed by a section with five comments, some of which are replies nested under others. Below the comments is a rich text editor labeled 'Your Answer'. Placeholder 'Lorem ipsum' text is used throughout the content areas. Sidebars show navigation and related links.

Measuring Success

Our primary metric will be an increase in constructive engagement within comments. We will also closely monitor comment flags and deletion rates to understand the impact on content quality and moderation load.

What are the new acceptable commenting rules?

To better support an environment where discussions can take place on Q&A pairs, we are experimenting with these updated commenting guidelines on Stack Overflow for the duration of the experiment:

  • Asking specific follow-up questions about the post.
  • Seeking clarification on how an answer works or why it might not work for you.
  • Sharing variations or related experiences pertinent to the Q&A.
  • Engaging in constructive, technical discussion sparked by a question or answer, even if it explores associated concepts.

These guidelines will allow for more interaction than the traditional methodology of only using comments to improve a post or request a clarification of its intent. We want to build a space for technical exchanges. Comments should still adhere to our Code of Conduct, remain polite and professional, and stay generally focused on the topic at hand. Comments should not be a substitute for answers, social chit-chat, or expressions of thanks.

Additionally, we will modify the flagging modal to disable the “no longer needed” flag type for the duration of the experiment. If users attempt to use it, they will be directed to the help center article on commenting, which will have an updated copy for the duration of the experiment.

Changing the rep to comment

After rolling out and evaluating the initial feature changes, we plan to experiment with lowering the reputation requirements for commenting, potentially down to two reputation points. Reputation limits on most features have shown that having rep of two or more prevents most spam. The fifty-rep limit is an arbitrary threshold that was not firmly based on beliefs or backed by data. We are aware that some spam will still get through; however, we need to start stress-testing this part of the system to identify where the reputation threshold breaks down. To be transparent, we would like to explore making it so any user can comment regardless of reputation, but we would first like to test out what we have outlined above before embarking on that experiment. Lowering rep further will require considerably more anti-spam work on our end to make that a viable possibility.

Looking Ahead

These updates are strictly an experiment. Once the testing phases are complete, we will evaluate the results and provided feedback to decide whether to keep any changes, disable them for further work, or revert them entirely.

We recognize that fostering longer discussions in comments might lead to threads that eventually warrant being turned into their own canonical Q&A pair. We see this as a potential follow-up initiative to explore if this core commenting experiment proves successful.

Should these changes prove favorable and effective, we will begin to undertake additional work to enhance visibility into comment moderation and implement general improvements to moderation tools in that area. Thanks to a post by Mithical, which has some good suggestions, we have some strong ideas that we are exploring, and we plan to have a more active discussion with moderators to better understand the needs here.

Ultimately, the goal is to make Stack Overflow and the entire Stack Exchange network more helpful, particularly for those seeking clarification or have related questions on existing answers. We believe that enabling these direct interactions on the relevant post is a crucial step in growing user engagement across the network.

We Need Your Feedback

These experiments are a significant change, and we know there will be strongly held opinions. We would like to hear from you on any of the following:

  • The commenting UI we have previewed
  • Any of the changes related to threading or the expanded editor
  • Changing the commenting policy
  • Lowering the reputation limit

We'll be monitoring this post for feedback until May 12th, 2025.

29
  • 95
    This experiment seems to suffer from the same issue as a lot of other failed experiments in the past several years. Namely: it tries to do too much at once. You should be changing the layout, or the functionality, or the policy on what comments are allowed. Not two or all three at once. Commented Apr 28 at 17:08
  • 28
    Related Make it clearer that comments are temporary Commented Apr 28 at 18:07
  • 15
    Am I confused or something? The first three bullet points in your expanded purpose of comments are part of the existing scope of comments. Commented Apr 28 at 20:45
  • 3
    @Catija We were using the help center article and expanding on the "When should I comment" section when we laid out those points. Which I do feel are quite a bit different. Commented Apr 28 at 20:51
  • 4
    Request clarification covers the first two points. Relevant but minor information is the third point. If you want to have more detailed descriptions, fine... but they're already (generally) allowed. Commented Apr 28 at 21:08
  • 5
    "Evolving" comments? Sounds more like devolution to me. Maybe change the SO logo to the Devo hat while you're at it. Commented Apr 29 at 11:57
  • 2
    @NoDataDumpNoContribution Nope. We are still working on Discussions and are in the process of trying another experiment there. Commented Apr 29 at 13:19
  • 34
    Personally, I think the concept is sound, just not.... not there. Why not take inspiration from the likes of Wikipedia, and just completely separate "Discussions" from the core Q&A, to maintain that cleanliness of "ask questions, get answers, no distractions"? Commented Apr 29 at 14:30
  • 3
    @Robotnik "Why not take inspiration from the likes of Wikipedia, and just completely separate "Discussions" from the core Q&A..." Maybe it's the mentioned visibility challenges ("we experimented with directing users to Discussions for follow-ups, we found it wasn't the ideal solution for these specific, contextual questions. This was partly due to lower visibility compared to the original Q&A page"). Commented Apr 29 at 15:52
  • 20
    @Robotnik I assume you are referring to the Talk pages on a Wiki article? We have discussed that, and I wouldn't rule that out as a potential experiment in the future. Commented Apr 29 at 17:16
  • 4
    @Hoid re: your latest comment and Robotnik's idea, I suggested a very similar thing a long time ago. You can see a demo of it on this deleted Sandbox answer (it was previously a JSFiddle but since the migration away from Imgur for images, the hotlinking on JSFiddle broke). That's a clear example of a 'middle ground' between easily-viewable comments and a separate page/tab for them. Replace traditional comments on the other tab with this more in-depth 'threaded' view, et. viola! Commented Apr 30 at 18:50
  • 3
    Will the length of comments be increased as part of this experiment? Or will code blocks count differently re: the character length? Commented Apr 30 at 19:33
  • 4
    The experiment seems to be live now on SO. Those that think that SO got preferential treatment all those years, just think about how it has to suffer from getting experimented on too. ;) Commented May 13 at 5:14
  • 3
    Alright, it's been a while. Can you now please revert these layout changes? Having comments be displayed at the same font size as answers only makes them distracting. It's harder to spot the next answer, and that is everything this change is doing. StackOverflow already is a very sophisticated platform. There's little room for improvement, so I'd suggest not wasting time on questionable experiments. In particular, I don't think obscuring the UI in this way will change the way people interact with this platform. This does not just take some "getting used to", the new layout is simply worse. Commented Jun 19 at 11:29
  • 4
    The last update was beginning of May, more than two months ago and the experiment was still running then. Is there a word for the pattern when you repeatedly start things in public but then cease to communicate about them, maybe even crease to work on them? Commented Jul 13 at 23:00

47 Answers 47

1
2
11

Despite how the current chat system is implemented, this should be handled as a feature of the "third space", ref. Do Trilogy Sites Need a "Third Place"? instead of only working on the "comments space"

The "comments space" is intended to improve posts. It might be convenient to have a specific post from the staff bout this space name and purpose.

Anything else beyond improving a question or answer, or asking for clarification, should not invade the "comments space". However, there should be a way to "move" or refer a misplaced post from one space to another.

Reference

3
  • 2
    Yeah, I've been banging on about this for years. However, Discussions is a sort of second third space, so it wouldn't be so bad imo for Discussions functionality to be integrated into the Q&A. Commented Apr 28 at 17:25
  • 1
    I don't really see a reason this space can't or shouldn't serve both purposes (follow up vs clarification/improvement) as long as there's a way to clean it up when necessary. Both are things someone finding the answer in the future may find value in seeing Commented Apr 28 at 17:26
  • 4
    @wizzwizz4 I think discussions should have their own full site, i.e., discussions.stackoverflow.com, instead of being integrated as they are nowadays. Commented Apr 28 at 17:27
9

Has any thought gone into how this new design will affect the handling and perception of duplicates?

I would generally expect dupe handling to not change due to this, but it’s certainly possible that people will use “it’s not an exact dupe but a follow up Q&A on this other question answers your question” as a means of closing questions. There’s also the possibility of follow up questions themselves being duplicates, which may need their own process of being handled.

2
  • 2
    Or people could argue that closing as duplicate should not happen because there are interesting follow ups. Commented Apr 29 at 18:59
  • 1
    @NoDataDumpNoContribution that also kinda leads into how it may affect other closure reasons. "This isn't off topic! it lead to a useful followup that is clearly programming related! We can't delete useful content!" Commented Apr 29 at 19:07
8

How does this integrate with chat?

Right now, comments and chat are sort-of one thing. Comment enough, and the system suggests moving to chat. However, chat's threading works by having messages begin with special :nnnn codes. On the other hand, chat messages are quite limited when it comes to content (though not as limited as current comments). Are you intending to integrate this properly?

Per How might Chat evolve? Help us identify problems and opportunities, you want to increase the visibility of chat. However, comments-moved-to-chat is one of the few things that expose Chat to the common user. Removing this integration would be a loss for Chat.

2
  • 3
    To be fair, comments and chat are not one thing at all – it suggests moving extended conversations to chat, and copies the comments over into chat if you do do that (or if the mods move a comment thread to chat), but they're technically separate entities. That said, I am curious how these threaded comments work with/interact with chat, particularly with the two different mechanisms for "moving" comments to chat. Commented Apr 29 at 19:36
  • 2
    I would assume, given the goals of this experiment, that moving a convo to chat would be a removed feature... but yea it's not even mentioned. Commented Apr 29 at 19:43
7

Collapsing comments (and answers?)

I think all comments should be collapsed by default irrespective of this experiment. Comments make it more difficult to get from question to first answer, and from first answer to second answer. If this is the only outcome of the experiment, I would be happy. Come to think of it, maybe in cases where there are more than 5 answers, answers should be collapsed as well. For this post, there was no efficient way to scan all the answers.

Visibility of posts and comments

Right now, every time a new answer is posted, the question moves to the top of the queue. If someone asks a follow-up question to an answer in a comment according to the new model, only the author of the answer (and the OP?) sees it.

I would encourage users to ask a new question, referencing the old answer, to get access to the entire community expertise. If they want to, they can add a comment to the old answer to specifically alert the answerer. However, the person who answered does not have a special responsibility to address follow-up questions for the next 10 years.

If you were contemplating to have new comments move questions to the top of the queue, that would probably bury new questions, so it would be a bad idea.

As an aside, sometimes questions move up the queue because someone posts a spam answer. After the answer gets deleted by moderators, the question remains high in the queue. It shouldn't, though, because there is nothing new to see.

Turning comments into new questions

How about a mechanism that detects follow-up questions. It would open a dialog "Would you like to ask a new question". It could auto-generate a link to the answer, and auto-generate a comment below the answer linking to the new question. This would be similar to the moderator turning a short non-answer answer into a comment, but in this case, the user would make their own determination and save some typing time.

5
  • 3
    I think most of these follow up questions written in a comment wouldn't have sufficient details if the user just writes it as a comment. Commented May 1 at 13:30
  • 3
    "Turning comments into new questions" -- exactly. Such a feature would guide new users into how to effectively ask follow-up questions. Commented May 1 at 13:58
  • If one could also guide a user how to repeat the pertinent parts of the question and answer(s) in their follow-up question, that would also be great. Perhaps the UI could guide a user into selecting the answer(s) to the question and then shows the existing question and answers on-screen while they ask their follow-up. That would truly make asking follow-up question easier. Commented May 1 at 14:01
  • Similarly, I'd love a feature that allows the OP or other trusted users to convert a comment into an answer, giving the comment-writer credit as the author of that answer. Commented May 1 at 14:02
  • 1
    meta.stackexchange.com/q/224044/997587 Commented May 1 at 16:06
7

Our primary metric will be an increase in constructive engagement within comments. We will also closely monitor comment flags and deletion rates to understand the impact on content quality and moderation load.

I think it's pretty obvious that this metric will improve with the new system, so it seems that by using it you are not really constructing a fair experiment. Will the number of answers decrease as the number of comments go up, with answers being replaced by comments which are harder to moderate?

A metric that will probably increase (but is not easy to measure) is the 'average time spent reading before I get a useful answer'.

1
  • Or another useful metric: do I come back reading another time. Commented May 13 at 5:12
7

As this feature got rolled out to me just now, the comments are by default expanded not collapsed - in contrast to what the post states (bold is my emphasis):

Please be aware that the screenshot below is a preliminary view and the final version may differ. The comments are expanded here just for demonstration purposes. Default would be to have them collapsed.

So they are not just expanded for demonstration but also for testing purposes?

enter image description here

My gut reaction to the feature is disgust, mostly because of the amount of space it takes. There might be good things hidden, I can't tell because of the overwhelming negative emotion.

2
  • Is this not the same behavior as before? Comments don't get collapsed until a certain number have been posted, and that number is smaller on posts with a ton of answers Commented May 19 at 14:27
  • @KevinB good point, I hadn't thought about ambiguity in what they meant was uncollapsed. I think they meant the replies to comments being expanded, not the comment part as a whole. You see that v next to "7 comments" - that's what I thought they would collapse by default. I guess I misunderstood. Commented May 19 at 15:42
6

Currently, there is an "Add a comment" link for the comment feature, intended to ask for clarification and help improve a question and answer.

For this experiment, instead of replacing the "Add a comment" link and mixing it with other types of content that will be allowed during the experiment, consider adding a new link next to it. Make it clear that it is part of an experiment and will follow the experiment content policies, along with details about what those policies are.

The regular comments should keep the "No longer need" flag.

If the above is not possible, please don't allow replies to old comments.

6

A specific usage scenario comes to mind that doesn't seem to have a good solution. Take the following example of a response to a question about turning on a compiler's optimizations:

Illustration of problematic interaction

The answer contained an error, which was pointed out in a comment. The error was corrected in a reply to the comment, but not edited into the answer itself (common practice on discussion forums).

This leaves both the question asker and other readers in a confusing position about what to do next:

  • Upvote the question/mark as answer - seems wrong because the answer itself is demonstrably incorrect and doing so would be misleading or harmful to future readers.
  • Downvote or ignore the answer - also seems wrong because there is indeed a correct answer here, if you dig far enough down to find it.
  • Upvote the comment but not the question - makes it harder to find accurate information. Should an answer with 30 upvotes rank higher than a 35-upvote comment on a 1-upvote answer? This gets complicated very quickly.
  • Post a new answer that contains the corrected information - would essentially be stealing someone else's content and is liable to get flagged/deleted.
  • Edit the correction into the answer, flag and delete the now-obsolete comments, and then upvote the question - lots of extra work for editors/moderators.

The last option is the only one where the question and answer end up in the desired state. It's also the least likely to be done since it's a multi-step process that takes time to complete. There's nothing preventing a user from doing any of the other options, and some of those can leave information landmines for future readers.

Comments already have this problem somewhat, but this change in its current form will make it much worse. Discussion forums typically frown upon editing existing posts (to avoid distorting the discussion) and encourage updates to be posted as conversational replies. People coming from those worlds will find it much easier to use this site incorrectly unless additional guardrails are put into place.

Also consider how this will impact SEO. Currently, when I do a general web search for a programming-related question and a StackOverflow page is in the top 3 results, DuckDuckGo will show a sidebar with a preview of the top-rated answer. This is by far my most common entry point to SO, and thanks to that helpful preview I'm significantly more likely to click that SO link than I am some other site where I have no idea if the page contains the answer I need. When meaningful content gets buried in discussion threads, these sorts of previews become difficult. Discussion requires context to understand. Machines aren't good at handling that, and neither are humans who are using a language they're not fluent in.

I don't know what the best way to reconcile this would be. At a minimum, we'd want a new way to flag a question as a "landmine" (meaning there's something buried in the comments that could be dangerous if a reader doesn't know about it). Setting this flag would need to do a couple of things:

  • Send a special type of notification to the answer author politely asking them to edit the information into the answer. You'd need to have some way to specify a comment so that the answer author knows where to look.
  • Provide a streamlined workflow for editing the question and cleaning up the obsolete comments all at once and without moderator involvement
  • Add some sort of obvious visual styling change to the answer, alerting readers that this answer is not what it seems
  • Add the answer to a separate "landmine" review queue so that high-rep users can easily find and fix them

That only helps fix these problems after they happen, though. It would be far better if we could prevent them from happening in the first place.

3
  • 2
    There is no confusion. There is nothing to reconcile. "there is indeed a correct answer here, if you dig far enough down to find it" There is not a correct answer post. So act appropriately--comment that the poster should edit, edit, post and/or downvote. When there is a correct answer, act appropriately--delete & flag obsolete comments, upvote and/or vote to reopen. Commented May 10 at 1:19
  • 6
    @philipxy We shouldn't have to go through all those extra steps though. The system should be designed to naturally guide users towards the correct way of doing things. The proposed design seems to be doing the opposite of that. Commented May 10 at 2:43
  • False dichotomy, just edit the answer, ignore the comments. While flag/delete is nice, it's not necessary. Commented May 19 at 20:51
6

What I don't understand here is... Why are we changing the UX here at the same time that we are experimenting with allowing longer-form comments?

UX changes are always going to get blowback and negative feedback because either they break someone's workflow, make information someone thinks is more important harder to access, or people just don't like the new look and think it's crowded. But the whole point of what you're doing here doesn't even need a UX change, you could have experimented with expanding the content allowed in comments without moving to this new UX at all.

These several initiatives did not need to be combined and likely would have been received/responded to better if they weren't. You could have simply replaced the commenting editor with the stacks editor, increased the allowed length, then when displaying comments implement a "show more" button that becomes available when a comment takes up more than a set height. The curation buttons didn't need to move, we didn't need a new vote button, we didn't need more whitespace, we didn't strictly need threaded functionality (but that too could be tested without the UX update,) it just feels like you're trying to do too much.

5

Soon we will be launching an experiment on Stack Overflow to enhance the quality and usefulness of comments.

That's good... I guess? I recall that the previous recent comment experiment went really poorly. Given the feedback for example here, the only sensible thing is to delete that whole previous experiment, since every single thing about it was poorly designed by people who don't even understand how the site is used.

So that doesn't exactly build confidence in a larger experiment. Please make sure that this is an opt-out experiment.

threaded replies and code formatting

That's great and long overdue!

This experiment represents a significant shift from the long-standing philosophy that comments are ephemeral and intended solely for post improvement or clarification.

That's probably ok but maybe we could at least design in some mechanism against what common misuses of comments: detailed follow-up questions or "I'm having this problem too". Perhaps a feature can be made so that you can click on some comment option "ask a follow-up question" or "I'm having this problem too" and then it takes you to ask a new question. With a link back to from where the question originated.

Please be aware that the screenshot below is a preliminary view

That looks pretty good! But I'm missing a down vote button or another way to deal with blatantly incorrect, harmful or misleading comments without flagging for moderator attention. Moderators have better things to do and often lack the domain knowledge to deal with such flags anyway.

Overall I feel like a comment system overhaul could create more moderator work, so maybe there needs to be a mechanism for trusted users to deal with bad/incorrect comments, like a 3 user consensus delete vote (15k or 20k rep required, something like that?). Or maybe related to tag badges.

This would mainly focus on incorrect comments. Things like CoC abuse should obviously be dealt with by moderators still.

Asking specific follow-up questions about the post.

Again, I don't think this should take place in the comment field, but through a link to a new question, with some manner of relation between the "parent question" and the "child follow-up question".

9
  • "some manner of relation between" We have links which also puts them in the related section and timestamps. On the other hand, questions should also be standalone, so cannot rely for anything essential on other questions (or can they)? Commented Apr 29 at 15:56
  • "Please make sure that this is an opt-out experiment." Opt-in, you mean Commented Apr 29 at 16:01
  • 8
    "maybe there needs to be a mechanism for trusted users to deal with bad/incorrect comments, like a 3 user consensus delete vote" Fun fact: 3 or 4 "No Longer Needed" flags from any user on Stack Overflow (I think it's higher on other sites) will already automatically delete a comment. Less fun fact: this experiment inexplicably intends to remove the No Longer Needed flag option. Commented Apr 29 at 16:03
  • @TylerH It was always a rather vague flag anyway, we could do with more detailed flags like break this one up into: comment too chatty/socializing, comment off-topic, comment outdated because of post edits, comment factually incorrect etc. Then maybe outsource flag moderation of the last one to trusted users with the relevant bronze badge or something like that. Commented Apr 30 at 6:53
  • 4
    @Lundin that sounds needlessly complicated. We don't need 37 different specific reasons which all boil down to "this should never have been posted" or "this is now obsolete". Commented Apr 30 at 7:58
  • 1
    @l4mpi We might have to if we are allowing user moderation of comments. If you aren't forced to motivate the flag, people are going to abuse it and pick "no longer needed" when they simply don't like what someone else is saying or to dictate what topics that are allowed to be discussed etc. We had this problem with vague flags/close vote reasons getting treated as "I personally don't like this" many times in the past. Commented Apr 30 at 8:13
  • @Lundin abuse of flags or close votes to remove content that should be kept is a thing. But so is leaving lots of trash on the site due to friction for flagging/closing it, or because no flag/close reason matches exactly, or simply general opposition to closing (I've seen variations of "but OP did their best" to justify why trash Qs should stay open way too often). IMO the site has leaned too far towards preservation since "too localized" was removed and getting worse from there, so IMO a correction towards more deletion would be great. Granted that's more required for Qs than comments though. Commented Apr 30 at 8:51
  • @l4mpi I totally agree with that and I've actually been fighting the "crap huggers" on SO meta for a decade or so - particularly annoying is when someone gets the idea of "you picked the wrong label for the crap so we are keeping the crap until you pick the right label". But that's a separate issue. There's a tendency that people pick bogus close vote reasons when they don't like or understand the question - see for example electronics.meta.stackexchange.com/questions/9696/…. Commented Apr 30 at 9:07
  • 4
    @Lundin User moderation of comments has worked OK for the last 16 years. Any changes re: that should make it easier/more effective. Requiring more time and legwork on the flagging users, and/or turning users into moderators having to discern different types of flags, does the opposite. Commented Apr 30 at 18:45
5

Overall, I really like the premise of this idea. Long overdue, in my opinion.

  • Asking specific follow-up questions about the post.

We have a line to walk here, because I believe the phrase "follow-up question" can be ambiguous. A follow-up question could be asking for clarification about post (the original intent of comments). A follow-up question could be interpreted by people as the equivalent of clicking the "Ask Question" button on the SO home page. We saw LOTS of this in the Discussions experiment where discussions were started that should have been legit Q&A posts on SO.

I still think "ask a follow-up question" is a missing workflow linking a new question to an old question or answer. I think a lot of noise will be introduced with this aspect of comments.

  • Seeking clarification on how an answer works or why it might not work for you.

Yes! Needed for a long time! Good stuff to have. Again, the challenge is in the presentation to the user. How do we, as people seeking the answer to a question, cut right to the chase — how do we go past the comments to get to the answer? This is a UX issue, in my opinion, and not a fundamental flaw with the premise of this experiment. UX issues can be sussed out with future revisions to this feature.

  • Sharing variations or related experiences pertinent to the Q&A.

This is an unheralded path to getting an answer. Countless times I've stumbled across a Q&A where the comments actually answered my question, because someone posted saying the answer almost worked for a similar problem.

  • Engaging in constructive, technical discussion sparked by a question or answer, even if it explores associated concepts.

This is the "community aspect" of comments. I like this idea. Again, though, we need to present the Q&A first. The UX treatment for comment threads should be easily accessible, but downplayed. I still want the Q&A, but I also want a quicker and easier segway into these deeper discussions which are interesting and can be insightful in unpredictable ways.

4
  • 'I still think "ask a follow-up question" is a missing workflow linking a new question to an old question or answer. I think a lot of noise will be introduced with this aspect of comments.' -- "power users" can easily link a new question back to an existing question. My most-upvoted questions does this. You can even link to a comment. But I've learned to NOT do this because I've learned that comments should not be considered enduring on Stack Overflow. Commented May 1 at 13:51
  • "Again, the challenge is in the presentation to the user." -- Again, I feel the existing technology already provide a good technique, which I've used many times before: Link back to the existing question or answer from either a new question or answer. And link forward from the existing question or answer to the new one. I think a new UI to help new users perform these sorts of tasks could be useful. One might even allow new users to create a comment that ONLY links to a new question that the tool guides them to crafting which both links back to the existing question and provides ... Commented May 1 at 13:54
  • ... the proper context for the new question. But I don't think branched comments will help SO in the long run. Rather, they will bring much harm to the site. Commented May 1 at 13:55
  • @JosiahYoder - that might be changing; comments might become enduring. Commented May 1 at 14:22
4

What will happen to the content when the question or answer is deleted?

The normal behavior is:

  • Deleting a question will also delete its answers and comments.
  • An answer creator can delete the answer unless it is the accepted answer.
  • This might be an edge case, but it's plausible that the question creator deletes the question even if it has answers, if the answers don't get upvotes.

References

7
  • 3
    "All those moments will be lost in time, like tears in rain". Commented Apr 30 at 14:47
  • 1
    Why would you expect comments to behave differently here? The experiment is solely about comments, so deleting the parent post will still presumably still delete any child comments. Commented Apr 30 at 18:46
  • 2
    @TylerH if the company wants comments to be more permanent, then the question is quite pertinent. A very straight forward example would be: 1. somebody answers 2. a technical discussion that explores associated concepts forms up in the comments (what the company allows and seems to want) 3. the answer is deleted. Now that technical discussion is "lost". Commented Apr 30 at 18:50
  • @VLAZ I mean, sure, but that's making a massive assumption about something that isn't really mentioned in this announcement. It would also probably triple or quadruple the amount of work for this experiment if that was something they wanted to address. Commented Apr 30 at 18:52
  • 1
    @TylerH and this is what is asked here, I think - what the company plans there. Or doesn't plan. Commented Apr 30 at 18:57
  • @VLAZ It's just so unrelated that it sticks out. You have to bend over backward to see how it might be something relevant. Commented Apr 30 at 19:26
  • 2
    @TylerH: Eh, I think it's very relevant. It's just also beyond the scope of this specific experiment, looking into the broader idea behind the experiment. Commented Apr 30 at 22:11
4

Markdown

With this fancier formatting, please add a "view source" button, or something like that, to view a comment's Markdown. That will enable converting comments to answers more easily. Currently, with the simpler formatting, it's easy enough to do it by hand, but with fancy formatting it would be more difficult.

You could even add a "copy this comment to an answer" button, or something like that, which could automatically copy the Markdown, put it in a blockquote, and add a link to the comment and the commenter's name (in order to fulfill the attribution requirements). In practice that could look like:

Comment:

Comment with MarkdownUsername May 4 at 17:03

Copied to answer:

Comment with Markdown

comment by Username, May 4 at 17:03

This is similar to "promote comment to answer" like Mad Scientist suggested.

Background

I do this pretty often on Stack Overflow when I see someone who should know better posting an answer as a comment. I'll copy the text verbatim, add the formatting, put it in a blockquote, and add the attribution below, then post it as a community wiki answer. If it needs any edits, I'll wait 5 minutes to let the grace period expire then edit it; sometimes if the edits are different enough, I'll remove the blockquote and move the attribution into the edit summary.

Sometimes I use Paste to Markdown to automate copying the formatting, but it's not perfect.

4

Another suggestion: reply-to commenting, basically automating copying the user's ID into an @ reference at the start of a comment. Yes auto complete helps but is still a more manual task than is really necessary. Would require adding a "reply" link to the comment's attribution, which some may consider excessive additional bulk.

(Comment threading might also be good, but would require more changes to the system.)

4

As pointed out the design has the issue of taking up quite some space vertically. As scrolling should be minimized maybe comments could be moved to the right side?

The hastily put together mockup below shows how this might look like:

Alternate design with comments to the style

In this design I respected the content-boundary set for general content. Perhaps the whitespace to the right couldn't be used - in the desktop configuration at least? Nested comment threads might become unreadable otherwise.

1
  • still wasteful. you could have put it on the left, where it would use space the user card "created" Commented May 14 at 16:10
2

Code block formatting is long overdue, especially for Python (which uses significant whitespace). Until now, we've had to use creative workarounds, like, say this is the code I want to post in a comment:

from sys import version_info as v
if v.major > 2:
    print(v.minor)
print(v.micro)

This can't be put in one line, so the only option is to break it up in a comment, which is ugly, e.g.

from sys import version_info as v then if v.major > 2: print(v.minor) then print(v.micro)

And there's at least one syntax difference between comments and posts (namely how to write a single code-formatted backslash: \), so hopefully this does away with it.

5
  • P.S. LMK if there's an existing feature request for code blocks in comments, so I can link it here. Commented May 4 at 20:36
  • P.P.S. If anyone could jog my memory on the syntax differences, please Commented May 4 at 20:37
  • P.P.P.S. (this is getting ridiculous) If anyone wants me to explain why this code can't be put on one line, I'd be happy to. I'm leaving it out of the answer for now since it doesn't seem super relevant and I'd need to look up the details. Commented May 4 at 20:38
  • Re: tP.P.S. - comments allow for a small subset of markdown: see How can I format and link in comments? for more detailed information on comments, but the markdown supported is bold, italics, inline code, links. And MathJax on sites where it's enabled. Maybe the only other thing I know which might be different is Backslash escaping in code regions in comments. Commented May 7 at 13:38
  • @VLAZ Oh, I meant it's different in the subset. I found what I was thinking of and edited it in. Thanks for jogging my memory :) Commented May 7 at 20:49
-4

Not quite sure if this is the place to report issues with the new comments but I have attached a screenshot of a formatting issue with code in a comment. For any unable to see the image, new lines seem to be showing as tabs, and the code block itself extends beyond the borders of the comment and into the "Hot Questions" list on the right side of the page. Honestly it makes the code harder to read than just having it as plaintext in my opinion.

Bugged code display in a comment

5
  • 7
    Erm, those are the regular, old style comments, aren’t they? Commented May 12 at 16:20
  • 1
    The comment is from 2019, so hardly part of the experiment. That will be starting tomorrow anyway. Commented May 12 at 19:40
  • @MisterMiyagi are they? I dont recall having seen anything like this before, could be mistaken Commented May 13 at 14:46
  • 1
    This is not the right place. You should make a new Meta post tagged "bug," with a link to the post where the broken comment is. Commented May 13 at 14:57
  • 2
    @murrag This issues regularly turns up. To give a few examples: meta.stackoverflow.com/q/269890, meta.stackoverflow.com/q/398119, meta.stackoverflow.com/q/398449 Commented May 13 at 15:03
1
2

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.