17

This Q&A is an AMA* with some members of the team supporting you. Post your questions as an answer; responses will be comments.

As the Stack Overflow community gets to work on building the future (Looking ahead, starting conversations, and seizing the moment), these are the staff members who will be working with you and supporting you.

Community Managers / Product Owners

  • Philippe - He made the case that you all were the right ones to steer this process, and we share his faith in the ability of this community to identify the best path forward.

  • Berthold - You may not hear from me often, but when you do it’s usually a lot at once. I’m often talking to you about unpopular things.

  • Spevacus - He saw behind the curtain and hasn’t run screaming yet. He’s got a lot of ideas powered by his time as a community member.

  • Frog - He’s the one who can take you on a deep dive into the data. A true amphibian who can live both above and below the surface of that digital pond.

  • JNat - JNat's been here long enough to know where the unicorns are buried. They bring history, context, and analysis that's sorely needed to a changing landscape.

The rest of the community team is here to help too, but you’ll hear from those four the most regarding Stack Overflow. Some CMs are working on projects with other sites on the network (Partnering with Communities to Modernize Policies & Norms). Others do quite a lot to keep the ship running. I want to give a serious shout-out to Bella_Blue. In addition to overseeing Trust & Safety, she is working at the bleeding edge of her field as part of the team building Stack Overflow for Agents. A lot of you probably think that project is weird. It totally is. But it’s also got one foot in the future, so give it some space to figure itself out.

Developers and Designers

  • Connell - He’s worked on many of the core contribution initiatives recently and loves taking a sledgehammer to the monolith

  • Cart - An upfront guy building out the front end

  • KyleMit - He’s been around for a long time and is deeply committed to making sure the site works for you

  • Lauren Ipsum - Lauren guides our work with excellent designs and users flows and has one of my favorite usernames on the site

  • marrados - He keeps us on our toes by asking the hard questions and automating our world to save everyone time

  • AI (in the spirit of transparency) - In addition to day-to-day human development work, we’ve been experimenting with using AI to tackle smaller and easily-defined changes. If you’re a software developer, you are probably nodding your head knowingly. If you’re not, this might seem scary. Don’t worry, our bots and agents are not sentient. They are monitored and their work is carefully reviewed. But we do enjoy having them be kinda snarky.

This is not an exhaustive list. There are some team members who don’t want a spotlight on them. There are others who contribute ideas and expertise but whose development and design work is focused elsewhere.

*Ask Me Anything doesn’t really mean “anything”. Any questions focused on things beyond our jobs are answered at the recipients’ discretion, and the recipients may be constrained by company policy or common sense from speaking to particular topics or issues. Please keep questions respectful and non-leading. Levity is welcome, let’s maybe aim for a 70/30 split of serious/fun.

9
  • 3
    Weirdly enough, I think this sort of AMA concept would have worked a lot better with the open ended question format. Or at least the comment threading from the main site. (Not that I'm a fan of either of these features, but I can recognize that for a question that's not really a question in the Q&A sense, those features would fit better.) Commented Jul 9 at 21:28
  • 1
    (it's a meta question, so I'm posting it as a comment) - would you be able to featured this? Honestly if the sidebar is capped at six items, not sure what should go (maybe the survey could get unfeatured early? or one of the blog posts?), but it might be worth it Commented Jul 9 at 22:19
  • 12
    "If you’re a software developer, you are probably nodding your head knowingly." -- I am a software developer. In fact I am a developer who is very familiar with the Stack Overflow codebase, and who, until a couple of months ago, found and fixed obviously LLM-generated bugs and vulnerabilities that made it into said codebase. Changes that looked confident and plausible on the surface. Changes that encouraged a "seems fine" kind of code review from engineers who were under pressure by management not to spend time on gaining an understanding. Commented Jul 11 at 14:17
  • 7
    "Nodding my head" is not my initial reaction to the fact that the company thought "this is good, let's make this the official policy." Commented Jul 11 at 14:17
  • @RoddyoftheFrozenPeas: Except that you don't get notified of replies to your "answers" in the open-ended Q&A format on the main site, making it terrible for discussions or conversations or feedback on other answers. If that was fixed, then yeah the threaded format would actually be good for something like this. (Exactly like a reddit AMA with threaded replies, unsurprisingly.) Commented Jul 11 at 15:46
  • 2
    @balpha thanks for giving an account of the actual happenings re LLMs at SE instead of the rosy management perspective. That mirrors my experience in a different domain (enterprise java telco provisioning) - LLMs still spit out lots of garbage and the few devs which use them cause the overwhelming majority of bugs that require hotfixes. So "nodding my head knowingly" is only applicable because it's been clear for a while that SE is all in on the LLM garbage train (see SO.AI and SOfA). That the "carefully reviewed" part in the OP is apparently not accurate is also expected, but concerning. Commented Jul 13 at 11:33
  • 3
    @l4mpi I didn't make a statement about the accuracy of that, because I can't (I haven't been there in a while after all). Commented Jul 13 at 11:56
  • 1
    @balpha well we can glean from your comment that it was not accurate when you were working there. Given that SE is now even more resource constrained, there is no reason to believe it would be different now. Especially because, as you say, the garbage produced by LLMs looks plausible - so a dev would need to inspect it rather thoroughly, which IME is a hard ask for a code review in general and twice so if the reviewer has lots of other tasks waiting for them. Commented Jul 13 at 12:07
  • 4
    This really should be featured in the sidebar. Commented Jul 14 at 1:19

12 Answers 12

Reset to default
27

This is specifically for Connell:

Why haven't you reverted the dates back to their original format in the comment "experiment" (graduated now)? You are still sticking to "Over a year ago" and did not provide any explanation or justification for that choice.

You did ask for the "ideal" format in a couple of comments. First of all, I am not sure we have an ideal format for dates underneath comments; ideal in what sense?! Putting that aside, amongst the terrible, the good, and an unknown ideal format, I'd stick to the good option for now, till we figure out that ideal one.

There have been many many posts and comments regarding this; I will list some of them here:

  1. Comment times are hidden/inaccessible when over a year old
  2. Can we not do "Over a year ago" please?
  3. What is the rationale behind "Over a year ago" for when a comment was made?
  4. Comments are dated "Over a year ago" instead of with the date posted

what if it's not just a backlog...

If this is not just a backlog and Stack Overflow is going to continue using that format, then someone, from the CM team probably, needs to provide justification by sharing clear methodology, solid data, and robust reasoning as an answer to one the aforementioned posts; moreover, you should justify how those metrics trump the issues laid out by the community. And questions should be tagged .

The previous comments regarding this provided nothing more than some conceptual ideas that you had when running the experiment which even back then were challenged. Now, after all this time, a short answer like "cuz new users like it and engage more" is not enough. We first need to see the data AND understand how that justifies all the issues caused by this.

18
  • 6
    I still think it's right to question the choice. But there was justification - it's to trick users. I've catalogued the reasoning in the third link. So, I think "why" is the wrong question because it invites the same garbage answer as always. "We saw a statistically significant increase in comments". And no verifiable data to follow. Even with sound methodology and even if the result really was an increase- these justifications are just another way to ignore feedback and sweep criticism under the rug, since stats don't tackle how useful the new comments are or what other problems are caused. Commented Jul 10 at 17:09
  • 6
    Let's put it another way - SE had PLENTY of opportunity to explain their reasoning or try to convince us that the change is good. They didn't. And still don't. The latest on the topic that I know of is this comment saying "Fair enough. Perhaps we can have that discussion another time." but I'm aware of no discussion and no other time that were had. Commented Jul 10 at 17:15
  • 2
    right... it's for the same reason the blog widget currently artificially shrinks every title even when it is entirely unnecessary. for the same reason the home button for not logged users walks you through a pointless process for "customizing your experience" just to throw a signup prompt at you and not actually customize anything. engagement above all else. Commented Jul 10 at 17:17
  • 2
    @M-- I'm saying that "why" doesn't lead to such responses. It's just an easy out when they answer the "why" with a padded out "cuz" and then never return to the topic. Commented Jul 10 at 17:36
  • 6
    As a former (as of seven months ago) developer at SO, let me say this: I don't believe Connell is the right addressee for this question. Sure, like many others, he's certainly technically capable of reverting that change (although depending on implementation details, a second developer may have to confirm the change). But the fact that he's engaging in discussions on meta does not mean he has any authority to make such a decision. Commented Jul 11 at 12:44
  • 9
    As a good engineer, he's gathering context, and he may be using that context to argue internally for or against certain things (successfully or not). But ultimately, such a prominent product decision is made well outside the realm of individual engineers. Commented Jul 11 at 12:44
  • 2
    @balpha thank you. I am a little shocked that you categorize a date label underneath comments as a prominent product decision, but probably I have been spoiled at my role(s) with so much freedom and authority. Putting that aside, if this is out of devs' hands and if it is not just a backlog, then the posts should be tagged status-declined with an answer (by CMs probably) explaining why. Commented Jul 11 at 14:52
  • 1
    Also related: Experimenting with New Comment UI Features on Stack Overflow Commented Jul 11 at 22:30
  • 1
    Fair point, @balpha. But it's not obvious to many community members that the staff who talk about the site mechanics don't have the authority, or at least influence, to affect those mechanics. We're focussing on Connell because he's been the main person justifying these vague comment dates. I'm grateful for the communication, and I don't want to shoot the messenger. But he's kinda made himself a target... Commented Jul 11 at 22:38
  • 5
    @balpha is correct - I don't make the final decisions on which experiments graduate, and overall I am probably not the best person to answer these kinds of questions. But, I did work on this experiment, I gathered our data, I engaged here on meta to gather feedback, and I was in the meeting when the decision was made. So, I'll try my best to give a response: Commented Jul 13 at 13:09
  • 5
    This was 1 of 3 simultaneous A/B tests: small vs large font, capped date vs not, and whether we display the user avatar. We measured which tweaks led to more comments, more upvotes, more users leaving comments, longer comments, users returning to comment again, the average score of those comments after 2 weeks, flags, edits. We sliced that data by the user's rep, account age and site visits to see if any changes were biased to new or old users. Then ultimately based on all that data, + user interviews + feedback here on meta + our own intuition, someone has to make a decision of which to keep. Commented Jul 13 at 13:09
  • 4
    The driver for this change was about getting users to engage with older content, which I am all for - let's keep the library fresh. The hypothesis was that users saw very old comment dates, assumed the content was stale, and moved on. So we wanted to test if users were more likely to engage if it wasn't so obvious how old the comments were. I didn't come up with the idea, I can't say I personally like it, and I suspected it wouldn't make a difference, but we wanted to see what happens, even if it was just gathering data and trying to understand our users better. Commented Jul 13 at 13:09
  • 3
    The capped date won on most measures we were optimising for: more comments, more upvotes, more users commenting, more users returning to comment again, longer comments, it even led to higher post scores (a metric I added because someone suggested it here). It wasn't all good: the comments left by users who saw a capped date tended to score lower, and, perhaps unexpectedly, higher rep users were more likely to upvote the capped date comments than new users were. Then based off all data + feedback, we had to make a decision, and the decision was to keep it. We can always change our minds later. Commented Jul 13 at 13:09
  • 2
    @M-- yes we did look at non-engagement metrics - two of my comments above listed them (post scores, comment scores, flags, number of edits). I also specifically mentioned how the data did not show a bias towards new users. Commented Jul 13 at 14:54
  • 3
    @Connell you stated (a.k.a mentioned) it did not show that. We didn't see the data, that's my point. And in my book, post score (as in voting), comment score, and edits all are considered engagement. Again, this is not the venue to resolve this. One of those post should get a detailed answer, and all of them should get either closed as dupe or tagged status-declined. Commented Jul 13 at 15:14
21

Philippe, almost exactly 2 years ago you promised that you'll see that folks would get a backup of all data dumps upon request:

@JourneymanGeek, If you request the backup for that purpose, I'll see that we get it to you. — Philippe, 2024-07-12 15:54:41 +00:00

It did not happen and you did not provide the backup upon request. Journeyman Geek posted a question to raise the issue publicly. It was and is a well received post, with 7 bounties started on it in order to draw (company's) attention to it, and it has a on it since February of this year.

I also left a (somewhat sassy) comment to notify you about that post:

@Philippe I hope you are already aware of this post; but in case you've missed it, I wanted to bring it to your attention. You promised over 14 months ago that you will see that we get the full data dump upon request. You also "comforted" VLAZ saying that you will not forget about this as you are committed to work on these. (Pardon my snarky tone, I couldn't help myself). But C'mon... I am sure you see the reason behind our lack of poetic faith in SO. It's hard to forget the years of history when each year is just more of the same. — M--, 2025-10-30 21:17:21 +00:00

Even if you cannot fulfill that promise for whatever reason, simply ignoring it and hoping it would go away is objectively the worst way of handling this.

At this point, addressing that request is not going to achieve anything as Journeyman stated in a comment:

I'd also note - my goals here were to show we could work together to get archives and that working through the company's processes meant we could get mutually beneficial results. In theory I could get the datadumps through unofficial channels - and will have to since I'd be missing ~ 4 quarters of dumps cause this took so long. This wasn't the only situation for me where these processes broke down. It isn't much use for us to provide opportunities to build trust, if the company makes commitments, then doesn't keep them. — Journeyman Geek, 2025-10-31 08:27:29 +00:00

Can you make yet another commitment that you'd keep your promises or would communicate clearly why you failed to keep them going forward? And would you promise if anything like this happened again, you'd give us a nickel so we'd be rich?

(Again, sorry for my sarcasm, but surely you understand.)

20

Question: why did Jody promise a conversation back when the beta mess happened, and then left after writing four comments?

4
  • 8
    Well, Jody isn't a part of this conversation (note the names at the top), but I'll take a swing at this one. While I can't speak to someone else's motivations, I'll say this: The product of the post that Jody made, informed by all of your responses, was to walk back the site redesign we had planned. Jody acknowledges that he should've been far more involved in the post, but as more and more feedback came in it became clear that we needed to course correct, so we did. His attention was pulled to managing that rollback. So not ideal on one front but I think the end result was good. Commented Jul 9 at 21:41
  • 11
    @Philippe The end result was back to square one and the communication in between was mostly absent. All posts were negatively received. And while the company was busy introducing the feature and then walking it back, it couldn't work on the things the community wanted. This counts as good? Commented Jul 10 at 6:28
  • 3
    @Philippe the result (beta gone) may be good, the path there wasn't. Also, Jody was "managing that rollback" and SE still needed a month and a critical meta post to remove the button and access to the beta site (which should take minutes, not weeks)? You also wrote there would be "one heck of a retro" regarding the beta, but if that happened, nothing was communicated - no post-mortem, no admission of any problematic behaviour by SE, no "lessons learned" etc; same for the other recent desaters. Yet yall come here acting like nothing's wrong and asking the community to fix your latest mess. Commented Jul 10 at 9:28
  • 7
    @Philippe He really ought to give an explanation why there are so many ideas, many of them actually decent, that fail completely during the implementation. It's a recurring pattern for pretty much any experimental feature attempted these past 2-3 years. Failure after failure - if the CTO is not responsible for that, then what is he responsible for? Commented Jul 10 at 12:56
20

One for the devs, from an outside perspective it appears as though you are hemorrhaging technical debt - with many company proof of concept ideas started without any additional time being given to complete their development (i.e open ended questions, free votes, coding challenges, staging ground and so on).

How do you maintain morale as a team?

4
  • 8
    Honestly, good question. I'm not sure how much morale there is left internally. Commented Jul 12 at 22:49
  • 1
    It's a good and fair question - I'll take an honest stab at it. Some days it's hard; other days it's fine. I'm not trying to shill here - but I very much like working on our Public Platform. It was one of the reasons I was excited to join Stack and it still holds true to this day. I've been at many dev shops before that time is always under a crunch, features need to get shipped sooner, and technical debt takes a back seat. So I wouldn't entirely classify this as a novel or Stack-only problem, but it's certainly something that we wrestle with. Commented Jul 13 at 12:39
  • 1
    On the other side, we've also had many tech debt / backend wins that aren't always celebrated (or sometimes even seen) within the community at large. Jadedly, sometimes it's like, "okay, good for you - but where are those features/bug fixes we've been asking for." I want to be clear - sometimes those are also good challenges too - we build the public platform for YOU ALL. But just to point out that not all tech debt work has the same level of visibility as other initiatives. Commented Jul 13 at 12:44
  • 1
    Also, I'd say morale and tech debt are two entirely different buckets here (although very occasionally related). Some of the challenges are also not new. We want the community to grow (that's good), but we also want to do so without abandoning what made the platform successful - strict quality adherence (that's also good). Solving those two things is often at odds and it often leave stakeholders on either side of the isle disappointed. I'd say that tension is a common, healthy, but frustrating source of morale (but again, not all that novel in 2026) Commented Jul 13 at 13:21
15

Why did it take 8 months to add "How to / Troubleshooting" to the dropdown? My understanding is mods gave feedback on this before it was even released, so surely it wasn't a dev time issue? Was there some other reasoning to keep the original over including "how to" in the funnel for traditional Q&A?

5
  • 7
    There were relatively recent changes to who makes decisions about such things. Commented Jul 9 at 19:58
  • 1
    @Berthold That sounds… good I guess? Are you allowed to go into detail about this? Commented Jul 9 at 20:29
  • 20
    I will. We have transitioned decision making and leadership for the network. Rather than being owned by the product team, decisions around the network are now owned by the community team. Obviously, I lead that team, and Berthold is stepping up to help with that piece of it. In addition, we've reconfigured some things and moved a few engineers over to support efforts on the network. Commented Jul 9 at 21:54
  • 4
    @Philippe: A long needed change. Thank you. Commented Jul 12 at 22:52
  • 1
    @Philippe I asked meta.stackexchange.com/questions/418132/… a while ago on MSE. It feels like your comment here (expanded a bit, perhaps) would answer it quite nicely, if you have a moment and inclination? :) Commented yesterday
15

I looked through the network profiles of the 9 people that you mention in your post. Only 2 of them have any meaningful amount of reputation on any non-meta site in the network (0 from the CM group). Are important decisions really being made by people that have such little experience with the site itself? Does the company not require employees to regularly use the product that they're responsible for?

9
  • 16
    (fyi) some staff have separate personal accounts. Commented Jul 10 at 5:15
  • 10
    Perhaps, but the overall community impression is that leadership is out of touch and doesn't understand how things work or why. Being seen only through a low-rep "work" account only reinforces those optics. Nobody necessarily wants to know their main account's username, but they would help their case tremendously if they were more open about how much site rep/experience they have and how often they actually use the site. Commented Jul 10 at 5:43
  • 16
    Spevacus was a regular community member before a CM. He was involved with things before the hire: in Tavern on the Meta, a Charcoal member, one of the coordinators of the 2023 strike, eventually a MSE mod. I don't know what a meaningful rep number is but he has 3k over at Arqade. More importantly, he participated and experienced the system beyond just rep: flagging, reviewing, editing, etc. I've seen users with far more reputation who show far less understanding of how the sites and community operate. Also, with fewer contributions outside answering. Commented Jul 10 at 7:13
  • 5
    Even his MSE rep is well earned pre-hire with posts like I feel helpless. Is there anything we can do to help guide SE? (and look - he stepped up as is doing more!), insights into the user-facing system, as well the technical system aspects. Post-hire, Spevacus is still one of the very few staff members who regularly keeps in touch with the community. And looking at site activity, he still occasionally does reviews on SO. Commented Jul 10 at 7:13
  • 12
    You're bringing a tear to my eye, @VLAZ - Thank you for the mention. As for the others, while some of them have also used non-staff private accounts, they've been around Stack for a long time now, and where needed they will always bounce technical/cultural questions off of those who have been involved more deeply in the sites as users. Berthold might not have a lot of network rep or contributions, but I can assure you he understands the cultural landscape of our little corner of the internet with intimate familiarity. Frog, too; just ask the SO mods - he's a regular in their mod room. Commented Jul 10 at 14:23
  • 8
    Philippe, despite having low network rep aside from Meta SE, has been here for over five years now. From working with him directly, I can say this with absolute confidence: he truly "gets" it. I recognize that, as his employee, my saying this may be taken with some skepticism (and hey, it should), but that trust is genuine and was an important part of my decision to join the company. I seriously would not be here if I didn't feel confident in the abilities and understanding that's present on the CM team. Commented Jul 10 at 14:32
  • 8
    For the developers, KyleMit goes without saying (check out his bounties tab if you want an eyebrow-raiser). Connell's been here for 3 years and has a track record that predates that on Stack Overflow. The other three, despite having smaller network activity histories, have all been here long enough to understand how this ancient code base works, and will always ask when a decision would benefit from user-level experience. Commented Jul 10 at 14:40
  • 11
    Also adding on that I do have a personal account, and I do post a bit on it :) Posting Q&A from my staff account feels like a no-go. I've done it, but the few times that I did, I felt I had an unfair advantage over other askers, getting treated with softer hands. But what I wanted was to understand the "true" asker experience. I'm not gonna get that with a staff badge next to my name. Commented Jul 10 at 16:12
  • 3
    +1 @Spevacus, and I would add that all of our engineers have a lot of experience using SO and have been regularly using the product for most of their careers. The vast majority of our users don't contribute directly to the content, and our engineering team needs to represent those users too. Commented Jul 13 at 13:27
13

What have the developers been working on for the past few months? Has there been any work done on the public site? The "new" feature is a max 5-minute job, so clearly they must be busy with something huge.

5
  • 1
    probably chasing the LLM hype, as SOfA seems to have gotten some updates over the last weeks... Commented Jul 9 at 16:37
  • 12
    I'll answer this honestly, but happy to talk to you more about it in greater depth. SOFA is a big part of how we want to stay relevant in an increasingly agent driven world. It doesn't mean humans using Stack should have any less pie, but it does try to capture the traffic that's increasingly being routed through Agents. Commented Jul 9 at 17:48
  • 5
    Totally fair that our "new" feature isn't all that new. We're trying to 80/20 our way into bringing in moderation tooling that has existed on traditional Q&A while opening up the platform so more types of conversations can happen - even ones that didn't fit well here 2/5/10 years ago. It took longer than 5 minutes, but we're also trying to get things out early and iterate. There are more QOL features we plan to add on top of Open Ended Questions, but want to get the ball rolling rather than waiting and shipping waterfall style. Commented Jul 9 at 17:51
  • 1
    @KyleMit So not much or almost no work went into the public site part (ex SOFA) in the last months? Commented Jul 10 at 6:31
  • 7
    @nodata you can see for yourself Commented Jul 10 at 6:35
12

What is the contingency plan when the AI hype eventually crashes?

By including all manner of AI-this-and-that in a product that originally didn't have jack to do with AI, the company is setting itself up for getting hit by the backlash when the AI hype crashes. Then what?

In fact SO has already down-sized several times while at the same time jumping on the AI hype train. The whole reason to jump on that train is to make money I thought. If that isn't even happening then why persist?

If the company had instead continued to focus on Q&A then it wouldn't be an "AI company" but a regular IT company. Those too will get hit hard when the AI hype crashes, but they might at least dodge bankruptcy.

For example Nvidia has a long tradition of making GPUs, so they can always resort to that if their stock value gets erased. Similarly, SO could have fallen back to Q&A, but it's too late for that since all users have left and it is no longer a viable may to make enough money.

4
  • 2
    I don't think the nvidia comparison is apples to apples. compute infrastructure is (as far as I understand) needed day to day for all the AI stuff because it's infra. training data, not necessarily, and it can (even if it's not supposed to) be scraped without paying for it. Commented Jul 10 at 18:02
  • 5
    "The whole reason to jump on [the AI] train is to make money I thought. If that isn't even happening then why persist?" --- Ahem. Shouldn't that question be aimed at OpenAI, Anthropic, Google, and Meta? /sarcasm/not-really Commented Jul 10 at 20:30
  • 1
    @GregBurghardt It is aimed to SO, who keeps going for AI then downsizing, then going for AI, then downsizing. Commented Jul 13 at 14:26
  • Sorry, my comment was meant to be a joke, if not one based on reality. Commented yesterday
11

What value does the team believe they can possibly add with all of these new changes?

Let's imagine for a second that every feature implementation goes exactly to plan and the current roadmap is seen through to the end.

Why would someone come to Stack Overflow to ask an "open-ended" programming question, rather than literally any other programming-related discussion forum on the Internet?

If an answer to a question (traditional, "open-ended" or whatever else) might come from AI, why would someone prefer to ask Stack Overflow's model for that answer over anyone else's?

How are AI "agents" supposed to benefit from SOfA in a way that they can't get in any number of other places?

Why would humans whose vision for the site is constantly disrespected and ignored by the staff and owners, have any interest in continuing to participate?

In short: what, as they apparently say nowadays, is supposed to be the "moat"?

It certainly isn't the pre-existing Q&A knowledge base, because that was already Creative Commons-licensed to the community.

4
  • 5
    "Why would someone come to Stack Overflow to ask an "open-ended" programming question, rather than literally any other programming-related discussion forum on the Internet?" Stack overflow has always had the mission to solve problems that developers are having. Full stop. Problems and needs change over time and we want to be honest and survey what problems technologists are having and how we can better address them. 1/2 Commented Jul 13 at 13:36
  • 3
    As evidenced by the fact that these discussions current have to happen on other internet forums indicates we're not doing a good job solving that problem here in-house. So obviously we'd like to provide a space to allow for that type of discussion within our platform. The why is something we want to make the argument for through iteration. Hopefully it becomes a good place to bring those types of questions and that becomes the why. 2/2 Commented Jul 13 at 13:37
  • 5
    Bonus 3/2 - Also - want to point out for the entire history of the site, users have been trying to come to Stack to have these types of questions. And have felt consistently slapped away by the community when their question is closed for a variety of reasons. So there's strong signal that users want to have these conversations here. Now it's up to the community to figure out if they want to be a part of them as well. Commented Jul 13 at 13:42
  • 3
    @KyleMit beat me to it, but yes, people already come to SO to ask open-ended questions. Our product depends on having an active community and these users are trying to be part that, so we want to provide an appropriate home for this type of content rather than just saying it is not allowed. But we want to get it right, so we've spent years experimenting with exactly how that should work: discussions, opinion-based content, open-ended questions; all the while balancing feedback from meta, from user behaviour data, from user interviews, and from our own inutition as users too. Commented Jul 13 at 13:53
8

Why did you (plural) think you, rather than the community, have the authority to fundamentally redefine the contours of the Stack Overflow community by adding open-ended questions?

13
  • 10
    The Community team's intended to be the stewards of the network; stewardship includes both preserving what works, and making changes when we recognize a need to evolve. The community's still very much at the table here: Your ability to policy-craft and wield moderation tooling on Q&A is front-and-center with this kind of content now. You are well within your rights to narrow scope with your tools if consensus deems it justified - just like you always have been. Commented Jul 10 at 16:16
  • 9
    Sorry, that doesn't make sense. You clearly have decided that certain kinds of questions are now allowed the weren't previously, and the community (in your view) apparently lacks the authority to override that change. Or do you want me and my two friends to decide that the open-ended questions experiment is disliked by a 5:1 margin and act on that apparent consensus by indiscriminately closing every single open-ended question? Of course you don't, so you have abrogated the community's authority to decide it's own topicality rules, despite denying it. Commented Jul 10 at 19:16
  • 1
    (I want to be clear, I know that coordinated activity of that sort would probably be against other rules, and I'm not actually planning to do it nor do I have two specific friends in mind - it's a rhetorical technique, not a serious proposal) Commented Jul 10 at 19:30
  • 1
    rhetorical or not, that's exactly what they've opened this content up to. There's a lot of content that was flowing through OEQ's that were widly off topic and not getting curated at the rate they should be, now that they're normal Q&A they'll be dealt with rather swiftly Commented Jul 10 at 19:33
  • 6
    I think you're conflating two different kinds of authority. The company has always been responsible for the site's overall direction, including in this case changes to what kinds of content we want to support. Ultimately, this has always been the company's responsibility. The community's taken charge on putting that overall direction into practice via their tools and meta: deciding whether individual questions meet the standard for being answerable, well-scoped, and useful, building consensus, and evolving site norms within that broader framework. Commented Jul 10 at 19:34
  • 5
    We're changing the outer boundary of what's potentially in scope. It doesn't remove the community's role in determining what quality questions look like, nor does it prevent the community from concluding that certain formulations or subsets of those questions don't work in practice. Commented Jul 10 at 19:38
  • @Spevacus would it be accurate to say, the company is dictating the kind of content, but not necessarily the topicality of the content? Commented Jul 10 at 19:43
  • 1
    @user400654 Far more succinctly explained than what I said, yeah. Obviously we'll still be a part of some of these topicality discussions, but we're not going to dictate policy in those discussions unless we really, truly felt y'all were going in a damaging direction with it, which I don't expect will happen. Commented Jul 10 at 19:49
  • 6
    "The company has always been responsible for the site's overall direction, including in this case changes to what kinds of content we want to support"--Only because the company is holding the keys, not because of any moral authority. Commented Jul 12 at 0:31
  • 3
    @khelwood I don't think they ever claimed any kind of moral authority. Commented 2 days ago
  • 1
    The comments here nicely describe the power gradient. The company tried it without the community when open-ended questions were implemented without curation abilities but that failed somehow. The community seems to be still needed and that gives it some influence but not a major one and may also put it at odds with the company if views are not aligned. Maybe this co-dependence will turn out for the better of everyone, maybe not. For those preferring more direct influence there are alternatives in fully community owned Q&A platforms. Commented yesterday
  • 1
    @Spevacus "I think you're conflating two different kinds of authority." First of all, let me say that I get the general idea behind the comment, even if - let me say - it... doesn't feel that great worded like that. Anyway, that's not the point. Here let me just give you a little food for thought. While you are right in pointing out that, de facto "you decide the direction" the company may be a tad... overestimating its position in coercing unpaid volunteers to be the its subs. "The community's taken charge", you are missing a big variable there - they have to be willing to. Commented yesterday
  • 1
    @Spevacus that is just to point out that while you may still have enough support from a few users willing to curate "no mater what" I get the feeling that - imho - that is rapidly falling as more high rep users and mods request self deletion. Just a few day ago another big Python regular threw the towel. Feel free to ignore this but.. I suggest the company should do some internal reasoning to understand if this elongated bleeding of old regulars is sustainable in the name of the changes. You may find yourself with your newer content but no one left to curate it. Commented yesterday
5

What is your vision for human to human knowledge sharing or knowledge building in say five years from now?

Will human Q&A still play a role for generating knowledge? Will LLMs be asked first for everything? Will only chit-chat remain?

Try to be quantitative, maybe compare to today or five years ago.

1

I remember the unfriendly comments robot built in 2019 as one of the first ML applications, then called NLP, and introduced by Kevin Montrose and Jason Punyon. It was positively received and a success. It interacted directly with the content but left humans the final decision.

Why didn't you make more of these over the years? It should have become much cheaper and at the same time much more powerful to do so. There was also ongoing need for help with curation like duplicate detection, closing, reopening, tagging, better titles, ... lots of possibilities and little actually done even though you were leaders at one point.

Why was this avenue (ML based tools for improving the content) not followed further despite tremendous developments and potential there?

4
  • 1
    :thinking: i thought that was full of false positives Commented 2 days ago
  • @user400654: From what little I know of the project, that is indeed what I've heard SO mods (and some staff/ex-staff) say about it. (Though that's not to say a modern version of it couldn't be better.) Commented yesterday
  • :remembering: Can't confirm it. The official threads from then said it was a success. 100s of thousands comments got deleted, the rate of unfriendly comments could at least be halved. And false positives, you always have with every method. But maybe the internal evaluation was different. Commented yesterday
  • I can't speak to the full dev perspective here, but we have done more in this vein, particularly ML based anti-spam which has been quite successful. The trouble for most applications is finding enough examples of a desired or undesired behavior to train on, and that problem grows very quickly as the number of categories we want to identify grows. Commented 14 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.