73

Stack Overflow Live Q&A YouTube Link: https://www.youtube.com/live/6-3J7G_XX0s


On February 26, 2025, from 3:00 to 4:30 PM EST (20:00 to 21:30 UTC), Stack Overflow CEO Prashanth Chandrasekar will be holding an AMA (Ask Me Anything) session for members of the Stack Overflow and Stack Exchange communities. We would appreciate it if you would join for this session, which will be hosted on YouTube. Learn more about this event from Prashanth’s blog post here.

Why You Should Attend

Live Q&A Session: Prashanth and other leadership team members will answer your questions. A presentation on the future of Stack Overflow. An opportunity to ask questions, share feedback, and hear from members of the Stack Exchange leadership team.

Session Agenda (90 minutes total)

  • Presentation - approximately 30 mins
  • Questions submitted in advance - approximately 30 mins
  • Questions submitted during the session - approximately 30 mins

How to Participate & Attend

You can view and participate in the session on the Stack YouTube Channel. You can submit your questions ahead of time either as an answer on this post or via submission on this form. You may also submit questions during the Q&A portion of the AMA.

Submitting a question is not a guarantee that one or any of your questions will be answered. You don’t have to attend to submit a question. Feel free to submit questions even if you won’t be able to attend.

Save the Date:

  • Date: February 26, 2025
  • Time: 20:00 UTC, 3:00 PM EST, 12:00 PM PST, 11:00 PM GMT+3, 7:00 AM AEDT
  • Platform: YouTube
  • Question submission: Use this form

We are looking forward to hearing from you and answering as many questions as we have time for. If you have any additional concerns or feedback, please leave them on this post.

We will continue to monitor this post until the AMA has concluded, but the best way to make sure your questions are submitted for possible inclusion in the AMA, especially during the week of the event, is to fill out the question submission form.

35
  • 219
    dunno if yall have forgotten, but you've kinda got a Q&A network here, it could even be organized by the tag you just created. Commented Feb 6, 2025 at 15:28
  • 133
    It could even just go on forever, with the CEO and "few others" being able to answer whenever they want, wherever they want, with no real time crunch or necessary need to have an answer immediately or within the next month. Commented Feb 6, 2025 at 15:53
  • 12
    Wouldn't it be better to hold this in the morning US time? Given that largest parts of the SE user base is split between 3 time zones: Americas, Europe/Africa and India. Commented Feb 6, 2025 at 15:53
  • 4
    @TylerH unless the questions are closely related I'd be for separate answers, then it's easier to vote on them and later select the popular ones. Commented Feb 6, 2025 at 16:22
  • 173
    Since I don't think this will get answered anyway I'll just throw this out here instead of filling that form: Over the last 4 years the CEO basically has had zero activity on the network (Other than a comment replying to a joke). Over this period of time there have been many discussions have happened and the companies relation with the community has been at a low point. Why does the CEO not use their own platform and why doesn't he address the community here but rather chooses external platforms to do so? Commented Feb 6, 2025 at 16:44
  • 16
    @AbdulAzizBarkat A comment replying to a joke about an event with users that was then postponed and never happened. Commented Feb 6, 2025 at 19:26
  • 15
    In Indian Standard Time, the meeting will commence at 1:30 AM. Note that the date in India and Australia will be February 27, 2025. Commented Feb 6, 2025 at 19:29
  • 100
    If SO's leadership had any interest in answering questions from the community, they would simply log in to meta occasionally and join in the discussions. I don't exactly know why they're doing this "AMA" but it's not to answer actual questions. Any predictions about how many "questions", not submitted in advance, will get "answered"? I guess 5. Commented Feb 7, 2025 at 5:57
  • 14
    @SteveBennett And here I am not just predicting how many, I think I may also know whose questions will be answered. Commented Feb 7, 2025 at 8:22
  • 6
    @SteveBennett it's just a show. Might be fun to watch, but I really and honestly don't expect it to be anything more than just fun. (i.e. no hope something useful will come out of it.) Commented Feb 9, 2025 at 9:27
  • 5
    @NoDataDumpNoContribution I remember it. It ran for 45 minutes and it was announced as it started. I found out about it 40 minutes after the start. It was never re-scheduled or in any way repeated. Not even all things that were raised there were given staff response. Even though some of the points that were raised after the 45 minute window did get a response, others didn't. I vividly remember how it didn't serve at all as a boost of confidence for Collectives and I took it as a red flag at the time. In retrospect, I feel it was justified. Commented Feb 10, 2025 at 13:03
  • 12
    @Sinatr You can submit your questions ahead of time either as an answer on this post or via submission on this form. It is perfectly reasonable for someone to want to ask a question without the general public knowing they were the one that asked it; they may not want to deal with the comments. We all know the questions are going to be filtered so it matters not that some are asked privately. Commented Feb 10, 2025 at 18:13
  • 9
    @Sinatr an external form that doesn't require login allows access to people outside of MSE Commented Feb 10, 2025 at 19:22
  • 37
    "A presentation on the future of Stack Overflow" - Why do I get the icky feeling that this is the main reason for this "AMA", and the "Ask" and "Anything" parts are just tacked on to draw people in to the presentation? How much of said presentation will involve convincing us that AI isn't evil? Commented Feb 12, 2025 at 0:17
  • 8
    CEO says one of his biggest mistakes is not engaging with the community on Meta. I'm guessing this doesn't extend to him actually doing so, starting tomorrow morning Commented Feb 26, 2025 at 20:43

56 Answers 56

215

AMA Question:

Why don’t you ever participate on meta or the rest of the network like the original founders did, in order to be able to get feedback from the community directly?

Answer by Prashanth Chandrasekar (CEO):

Prashanth: got it who asked this by the way that that...

Philippe: We actually got that asked by a number of people beginning with Starship but we got similar questions from Curious Danny, journeyman geek, wizzwizz4, Bobble

Prashanth: Okay right, it's a popular question. I recognize some of those names, Starship, thank you also for your very thoughtful questions and answers throughout the years. I've enjoyed reading them on meta.

So just so you know, I may not be very active on meta but I read everything. In fact I will often tell my team that this is one of the things that I do late at night right before going to sleep, which I'm not sure that's a good idea in retrospect, but it's but it is definitely something that I do. I read pretty much all things about the company, anybody that knows me knows I'm fairly detail oriented that way.

And the general sort of point here is that I actually love engaging with our community, and I do it quite regularly. And I do it, let's say number one, in real life. I'm constantly talking to community members everywhere I am in person. Number two, community members who are working at companies that we are working with, I often engage with them as we talk about obviously not only our public Stack Overflow but also Enterprise Products. And I often, by the way, share a lot of what I hear in these sort of areas with the company, and I write an email every week on Fridays to the whole company. And I've been doing it you know for five and a half years, and I've been at the company without missing a beat and oftentimes I will say this is what I heard from users this week, right. And you know it's up to us to do something about that.

And so just so you know that I absolutely love engaging with all of you and I think but more broadly speaking I think my job has you know multiple dimensions as a CEO. I think it's different than being a founder of the company that started, or founders, that started this platform, with that sole mission. For me it's like talking to customers because we have a business, obviously talking to the community, we have talking to partners now in the context of these AI topics, it's talking to employees to make sure they understand that we're changing our direction in this new world, talking to investors. So it's you know it's a balancing act ultimately in terms of like the time that I allocate towards being, let's say, specifically on the meta platform. And I also, by the way, rely on as I mentioned previously, an excellent team with Philippe leading it, and his community management team that spend a lot of time doing that on a minute-by-minute basis, so then I'm in constant touch with Philippe all the time to to understand you know what's happening what's not happening and for us to make decisions off that. So you know they've got my trust, my ear constantly, and a lot of my support. And I don't want to come across a sort of overly micromanaging or distrusting their efforts by just being in the details like you know 24/7 in their work.

So I joined the company October 1st 2019, I literally walked into also a fairly like you know very complex community situation, there was a very negative situation that was underway, and it was a very tricky time. Also I think for me to sort of engage at that moment, and I think it sort of set me back actually personally, just reflecting on it and being able to just be having this sort of conversation with every community member through even meta. And you know I didn't want to jump in without fully understanding the nature of what was going on, and how things worked etc. So I chose instead to basically let my team sort of you know at that time to manage that and I sort of focus on the other aspects of you know what I mentioned around running the company and so on. But I probably could have and should have probably engaged more fully when things did calm down after that original introduction, if you will, and I'll just sort of reiterate that even though I may not be very active on meta I'm constantly reading all your comments there and I'm constantly obviously like talking to all of you in you know in all these other places that I described that we engage with community members.

19
  • 3
    Out of curiosity, were Jeff and Joel ever active on any site other than SO or MSE/MSO? Commented Feb 6, 2025 at 20:02
  • 24
    @V2Blast yes: stackexchange.com/users/1/jeff-atwood?tab=accounts Commented Feb 6, 2025 at 20:16
  • 12
    @V2Blast Jeff was active on SU, SF, Webmasters and many others Commented Feb 6, 2025 at 20:16
  • 17
    @V2Blast Joel too:stackexchange.com/users/4/joel-spolsky?tab=accounts Commented Feb 6, 2025 at 20:17
  • 8
    I had interactions with Jeff on Super User and meta Super User - even disagreed on things politely Commented Feb 6, 2025 at 23:34
  • Interesting. I should have said non-tech sites, but even then both of them have a decent amount of activity on some of those too. Commented Feb 6, 2025 at 23:47
  • 48
    Both Jeff and early Joel were deeply plugged in an engaged at the ground level. I can’t imagine anyone in current upper leadership directly engaging on any meta, let alone taking ownership of more controversial decisions Commented Feb 7, 2025 at 3:43
  • 13
    To be fair, Spolsky was also very inactive on the sites when he was CEO. I don't recall him doing anything at all after that "take a stand" post when Trump was first elected in 2016. Commented Feb 7, 2025 at 11:46
  • 12
    @user9517 Yes, but "interact with your community" is part of "building a community" playbook Commented Feb 7, 2025 at 14:41
  • 5
    And whether or not it's common for CEOs is completely beside the point, especially on this platform - the question still stands. Commented Feb 9, 2025 at 11:39
  • 1
    > "interact with your community" is part of "building a community" playbook That's the community teams job the CEO just 'makes it so'. Commented Feb 10, 2025 at 8:56
  • 1
    I think people are missing the context of the network. This isn't some random CEO, but the CEO of a company that has done great harms to the community it claims to be a steward of, and for which it wants to regain their trust. Well, regaining trust is fundamentally a personal thing, and for many of us, the company can't begin to regain our trust until the head of the company, the CEO, shows us that he personally is committed to regaining our trust. Being active in our communities would be a great way to show that the CEO does actually care about company-community relations. Commented Feb 26, 2025 at 1:29
  • 4
    Summary of his answer to this question based on what I understood: "I do engage in meta community, I read it before going to sleep, I engage with community in person - with our customers, business partners, etc., I talk to members of community in person at the places with do business with [...]". (He said other stuff I can't remember). But as far as I understood he primarily sees business partners as his community. At least that's the feeling I got (English is not my first languages, I might have misunderstood things). Commented Feb 26, 2025 at 21:32
  • 2
    @Starship youtu.be/6-3J7G_XX0s?t=4869 And here is his activity for side by side comparison: stackexchange.com/users/16260110/… Commented Feb 27, 2025 at 2:41
  • 2
    @V2Blast came across this looking for something else on root access apple.meta.stackexchange.com/questions/589/… - so a very small sample of Joel on AD Commented Mar 11, 2025 at 11:40
172

AMA Question:

Why do you seem to regularly implement features or experiments to Q&A that have overwhelmingly negative receptions when announced, instead of adjusting (or canceling) those features/experiments to address the concerns of your product users, or instead of implementing the features your products users are actually asking for? Whose feedback are you listening to instead, if not the most active users of your products? On a related note, roughly how many employees are there currently dedicated to developing and testing your primary product, Q&A?

5
  • 11
    Don’t know if it’s the primary topic of this AMA question (I’m guessing it is) but maybe highlight the Answers Bot, the announcement for which has gotten well-over 200 downvotes in the last day and which was protested by basically every user. Commented Feb 6, 2025 at 17:53
  • 5
    If we're going to focus it on a specific topic, it might be better to focus it on a non-current one, otherwise it's just treading the same ground as probably dozens of other questions. (for that reason it may be better to just not) Commented Feb 6, 2025 at 17:56
  • 3
    @controlgroup An AMA does not have a primary topic, since it stands for "Ask Me Anything". Commented Feb 10, 2025 at 14:52
  • Even without talking about AI, the new Stack Overflow home page was horrid. The announcement for it is among the 50 worst-scoring posts in MSO history. I’d be interested in hearing about the decision-making process behind that particular feature. Commented Feb 26, 2025 at 1:05
  • @Anerdw Mr Krabs voice: "we like money" Commented Feb 26, 2025 at 12:34
160

AMA Question:

What is the company doing to attract and retain experts to answer questions?


Here is my summary of the answer. You may have a different take if you read the transcript or watch the video.

Over the years it has gotten more difficult to both ask good questions and answer them on Stack Overflow. The way people interact has evolved over time (implying that things that worked before may not work today).

The company is interested in adding different mediums and types of content, like chat, informal discussions, or videos with the purpose of helping people get started or meeting them where they are. The Staging Ground is an example of helping people create better questions and connecting them with mentors, which has resulted in better/more answers.

There are a lot of things to look into - it is intimidating to ask questions on Stack Overflow and the company is focused on helping people graduate from that first question asking experience to more regular interaction.

So my take-away from this is that the company believes that they can convert the people asking questions into people who answer questions and isn't doing anything to attract people to come to the site explicitly to answer questions. I think y'all are down a rabbit hole. As you said, times have changed. Converting question askers into highly engaged answerers just isn't going to work the way it did in the past.

I look for information on Stack Overflow all the time, but all the questions I need answered are already there. I have almost no need to ask new ones, because at this point in my career my questions are so niche and specific to what I'm doing I'm not going to waste my time shouting them to the general public.

What are you doing to incentivize people like me to answer questions? Why would I participate in the Staging Grounds? So what if there are good quality questions I know the answer to? I'm not looking for them. I'm looking for an answer. I'm not searching the SE network unless I'm already familiar with it. I'm using a search engine that is going to have AI slop as an answer at the top of the results, then a bunch of links to SEO-savvy sites that are going to be more AI-generated garbage, and maybe links to Wikipedia and Reddit after that.

If I were completely new to the network, what would motivate me to answer a question? The company's mission to help Google and other giant tech corps train their AI with good human-sourced data so they can put it in some developer's IDE and charge for it doesn't inspire me. It also completely bypasses/obsoletes any existing answers on Stack Overflow, so there's no hope of building something of lasting usefulness, especially now that the data dump is sequestered on the company's private servers and access to it has been restricted.

The live stream spent a lot of time (at least in the parts I skimmed) explaining what's in it for the company - how great the people working there are, how rewarding the work is, how bright the future is for the company if it embraces AI... what's in it for me? What makes this site worth donating my time to?

17
  • 76
    Remember when experts on SO/SE enthusiastically promoted the network to their work colleagues and friends? Now when I tell people about SE it's with disclaimers and warnings. Commented Feb 6, 2025 at 19:33
  • 11
    @PM2Ring Come to think of it, it has been a long time since I’ve mentioned to anyone that I was a moderator on the network or encouraged anyone to visit. Commented Feb 7, 2025 at 12:35
  • 1
    @PM2Ring Very much so, unfortunately. In 2018 I had a story full of positivity and now it's full of "we're losing users left and right". Commented Feb 8, 2025 at 19:33
  • 3
    [...if anything...] Commented Feb 16, 2025 at 12:44
  • SO seems to have become a ghost town. "do my homework for me" type questions or repeats of basic easily found questions, snarky comments on old accepted answers, and no (or very low quality, AI?) responses to questions. I check in, but without any real enthusiasm or expectation of help or things worth contributing on. Commented Feb 24, 2025 at 7:46
  • 1
    @PM2Ring I still consider SO a great resource for very specific questions. AI is much easier and usually reliable for really simple questions, but when you get specific it becomes very unreliable. I think Github issues have also become significant competition for SO, even though the Q&A format of SO is far more pleasant to deal with than sifting through a giant comment chain. Commented Feb 24, 2025 at 19:31
  • ... perhaps better phrased as "What is the company doing to stop detracting and driving away experts who answer questions?" :-( Commented Feb 26, 2025 at 15:43
  • Answered at 41:39 Commented Feb 26, 2025 at 22:06
  • I am addicted to answering questions. "Reputation" helps keep me on the site answering questions. However, I noticed that the graph of my reputation is sagging and it seems like there are fewer and fewer questions I am interested in tackling. Maybe, as mentioned in the AMA, it is AI?? Commented Feb 27, 2025 at 16:07
  • @RickJames They've already hooked you - I'm asking what their plan is to get new blood in when you inevitably burn out and move on to other things. Was your first interaction with SE a question or an answer? I think a lot of the old guard started just by finding answers, then by browsing the site just to read it, and then after lurking a while jumped in to answer something themselves. Or, they started by asking a question and had a positive experience with the people, and started looking for more ways to gain reputation because they wanted to engage with the site more. Commented Feb 27, 2025 at 16:14
  • @ColleenV - Yeah, that's a pretty good summary. I discovered Stack* from a Google search about 15 years ago. I soon found that I could Answer some of the questions while having very few Questions to ask. I now keep several 'search' tabs in Chrome and check them for new issues daily (or whenever I am bored with other stuff). I am retired; this is my main hobby. I don't remember if I started with a Question. Commented Feb 27, 2025 at 16:23
  • @ColleenV - And... I used to troll MySQL.com's Q&A. Once I saw how well designed Stack* was, I have abandoned answering questions for them. Commented Feb 27, 2025 at 16:24
  • @ColleenV - Oh, this string of comments brings up a "bug" that I am working around. [I assume it is still a bug.] Notice that you got a link to the last of my comments here. The workaround I use is to hint that my Comment is one of several, so you will think to look "up" to see the rest of the string. Commented Feb 27, 2025 at 16:31
  • @ColleenV - And... Most of my time is on StackOverflow; that name and "StackExchange" mean nothing; it took me a long time to understand which questions belong in which "forum". Oh, what is the generic term if it is not "forum"? Commented Feb 27, 2025 at 16:33
  • 1
    @ColleenV - And, surprisingly, no sign of burnout [yet]. I have noticed some of the 'regulars' disappearing. I don't know how to encourage 'new blood'. I bias my upvotes toward newbies (when appropriate). I did get chased away from [php] by one or two belligerent contributors; I try to tread gently on newbies to [mysql], where I spend most of my time. Commented Feb 27, 2025 at 16:38
128

AMA question:

Why, in your mind would anyone take the time to verify an AI answer rather than writing an answer or asking an AI themselves? And why would anyone ask a question on Stack Exchange to get an answer from an AI when they could ask that AI directly?

13
  • 3
    Hi again. In this case, their outward explanation could be that, it takes a lot less time to read an answer and correct it if needed, than it takes to write one up of your own. And as far as asking a general AI, that AI would be one who hasn't had his answers vetted by humans, therefore might be hallucinating a lot more than the SE AI. Also, since the training set of the SE AI would be SE alone, there would be a lot less noise as far as getting it trained -- so its answers would become more coherent more quickly. Commented Feb 7, 2025 at 17:11
  • @Alex But why don't I go to chat.openai.com and ask the AI myself and fact check it yourself. The SE AI will hallucinate just as much as any other AI because it just is that AI but in SE Commented Feb 7, 2025 at 17:13
  • 1
    I don't mean to be pushy; feel free to let me know if you disagree with my assessment here or on any other chat; but the idea is, that the "just as much" is not accurate. Human editing will teach the AI how not to hallucinate. The goal of this experiment (me having CS research background) is to see whether or not an AI algorithm of what not to say on top of the current AI algorithm of what to say, can in fact lead to good answers. But to know what not to say, they need data of humans editing AI hallucinations. And they chose SE to get that data. Commented Feb 7, 2025 at 17:18
  • p.s. this is also how I came to suspect that their end client here is not openAI, but someone highly politicized. Commented Feb 7, 2025 at 17:24
  • 3
    @Alex AIs are currently trained on about half the internet. Our edit is an insignificant blimp in a massive dataset likely ignored. Commented Feb 7, 2025 at 17:47
  • 3
    @Starship That's only good for unsupervised learning. A set of (candidate answer, correctness score, edited answer) tuples is worth more than an unlabeled dataset many times the size. Which isn't to say that's what the experiment is, but it would be a meaningfully different dataset. Commented Feb 7, 2025 at 18:38
  • @Ray Still, it isn't nearly as large as many other datasets out there. And remember that robo-reviewers will exist on both sides. Commented Feb 7, 2025 at 18:39
  • @Starship Yeah, I'm not saying the exercise would actually generate quality data. Just that size is only one of many relevant factors when it comes to how useful a dataset is. That's why RLHF is such a major part of a lot of training approaches: labeled data can make a big difference. Commented Feb 7, 2025 at 18:42
  • 1
    @Starship I'll rephrase what Ray said in my simple language. Getting all data on the i-net is simply the first "pass" in approximating what humans are expecting. The next step in predicting what humans are expecting requires AI to generate its approximation, and then have humans "guide" it for a second pass over the same data. AI cannot get a second refinement in its results without some feedback over what its first "pass" has done wrong. This is where the CEOs are hunting for a good (refinement amount) / ($ spent) ratio. And SE just happens to have a lot of people willing to work for cheap. Commented Feb 7, 2025 at 19:51
  • The second question has an obvious answer: "It's about as easy and you get some sort of answer almost immediately and with no 'judgement'." Commented Feb 26, 2025 at 15:44
  • @einpoklum Its more difficult (limited topic, takes longer for AI to answer, can't reply to prompt, etc.) and regardless it needs to be better to get someone to switch Commented Feb 26, 2025 at 15:46
  • @Starship: Do the topics not cover the large majority of user questions? Also - questions can remain unanswered forever, and often do, either because they're not well-received or because they require very specific expertise; the second reason is a problem with AI as well, but the first is where AI does "help". Commented Feb 26, 2025 at 16:10
  • @einpoklum Evidently not (source: how many questions I flag/close as off topic per day). Commented Feb 26, 2025 at 16:13
112

AMA question:

This tweet really gets it 100% right

You know what the biggest problem with pushing all-things-AI is? Wrong direction.
I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes.

Most of the AI features we've seen from the SE devs recently seem to be going the wrong direction - towards content creation and curation. But that's the main thing that keeps most of us interested and passionate about our sites.

This problem isn't limited just to SE - most capitalistic uses of AI seem intended to turn humans back to menial labour. But could SE be part of a reverse trend? Do you have a vision for AI that would help free up human time for the interesting creative side of Q&A?

11
  • I think this question (and the quoted tweet) deserves its own t-shirt Commented Feb 11, 2025 at 22:07
  • 9
    "most capitalistic uses of AI seem intended to turn humans back to menial labour." Because capitalism as a system demands maximum labor output at all times. To capitalism, AI should free up more time for work and act as a force multiplier so the work is more productive, not the opposite. Commented Feb 12, 2025 at 15:23
  • In particular, when I tell AI to do my laundry and dishes, I don't want it to throw the dishes out the window, put my cat in the washer and then cover it all up with shameless lies. Step #1 of pushing out AI is to make it intelligent in the first place. Commented Feb 14, 2025 at 9:29
  • Capitalism just means that "the means of production" (factories, machinery, computers, employees, etc.) are in "private ownership" (as opposed to being owned by the state) for "profit" (so you don't need to live paycheck to paycheck). It is unrelated to economic output. Commented Feb 24, 2025 at 10:17
  • 2
    @Ocean That’s a very textbook definition. It’s not useful for thinking about how capitalism works in the modern world. The profit motive is an undeniable driving force for a lot of corporate behavior. That’s not to say the profit motive is universally evil, but it definitely exists, and it’s definitely tied to capitalism. Commented Feb 26, 2025 at 1:08
  • 1
    IMHO given the current state of art, it's easier/cheaper/more reliable for AI to generate images, sounds and writing than for it to do laundry and dishes. It's not a philosophical choice, but a pragmatic one. Commented Feb 26, 2025 at 1:09
  • @Ricardo Just as it's "pragmatic" to do away with the arts entirely? Of course for the text that GenAI produces for programming questions it's of questionable value, meaning that often it's wasting electricity for zero usefulness. Not pragmatic either. Commented Feb 26, 2025 at 1:26
  • @curiousdannii I'm personally in favor that technology helps us to do less manual work and empower us to do more intellectual, creative and moral work. My previous point was to suggest the possibility that it may seem like we're not going on that direction, not because people working on it doesn't also want it, but because that's the current state of art. Commented Feb 26, 2025 at 1:35
  • @Anerdw Pursuing profit at the expense of other people isn't a consequence of capitalism. Look back when there were nobility and peasants. Pursuing profits at the expense of other people was worse then. What capitalism does is to enrich a population. On the other hand communism impoverishes a population. Commented Feb 26, 2025 at 19:53
  • 1
    @Ocean Nobody brought up communism, and nobody's saying capitalism is universally immiserating. It just sucks at certain things. That's factually undeniable, and it's true of every economic system. Commented Feb 26, 2025 at 21:21
  • @Anerdw The issue is that you are blaming capitalism for something capitalism is not responsible for. You're using capitalism as a scapegoat for human nature. Commented Feb 27, 2025 at 10:44
106

AMA question:

Do you use any Stack Exchange sites yourself? Your network profile shows little activity, but do you perhaps use various sites in a read-only capacity? If so, what do you think of the user experience for primarily read-only users?

1
  • 3
    Network-wide: 132 votes and no helpful flags, approved edits, nor enthusiast badges, so little passive activity too Commented Feb 26, 2025 at 19:41
81

AMA Question:

Why now?

You've stayed silent over so many things that have happened that have caused strife in the community over the last 4 years.

So why now?

Answer by Prashanth Chandrasekar (CEO):

Yeah, it's a really I think it's a it's a great point I think that this is a very as Jody mentioned previously it's it's a it's a particular moment in time where this is an inflection point for not only Stack Overflow but for the entire software development programming coding sort of space in community, right. So our goal is to engage with all of you to build the future of what Stack should be together with you as I mentioned already. And you know we want to be even more transparent about our thinking. That's part of the reason why we wanted to do this in this particular moment in time, and also we want you to be involved in building that future with us as we talked about. So this is like my way of you know explicitly stating that, to all of you, to say please engage please work with our team engage with the feedback engage with the mechanisms we've provided you know whether it's in research or surveys or through meta or anything else, or even joining our community management team, to make sure that we can build it the way we all feel very happy about to add value for the future.

8
  • 13
    ... why not now? Commented Feb 7, 2025 at 2:04
  • 29
    @starball - Thats fine too, but I'd love to know what the motivation is given everything that they've chosen to ignore Commented Feb 7, 2025 at 9:13
  • 7
    @Sayse - The CEO had a spare hour in his calendar? Commented Feb 10, 2025 at 0:42
  • 3
    @Sayse There's a saying "Trust the no, don't trust the why." So, if this question gets answered, would one trust the answer? Commented Feb 14, 2025 at 17:35
  • 2
    It's because it's not for us. There's some Important Person or other who needs to be impressed and this is just a show for them. Commented Feb 23, 2025 at 21:11
  • 1
    I got the feeling is going to be like presidential elections: you are going to assist only to be told later they gave you a listening attempt, while implementing anyway any bs they want to do Commented Feb 24, 2025 at 0:40
  • @Robotnik and that spare hour can be used for other activities. So why this activity in this spare hour? Commented Feb 24, 2025 at 15:40
  • Cause he needs to talk about AI, apparently. Commented Feb 26, 2025 at 20:18
60

AMA question:

What mistakes do you believe you made since taking the CEO position? What would you have done differently if you had the knowledge/experience you do now?

Philippe, at 38:45: This comes from CPlus, one of our users and CPlus asks what mistakes do you think you've made since taking the CEO position? What would you have done differently if you had the knowledge and experience that you do now?

Prashanth Chandrasekar: Yeah, it's fair question. As it relates, there are two things that come to my head.

The first one is really probably have tilted more on fixing this content creation issue much earlier. And so maybe opening up these lanes a lot sooner, because you know the content as the world sort of creates documents in the Library of Congress on Stack Overflow, all the knowledge on programming, the incremental new things to be documented. You know there's kind of a difference in the volume that originally started out when we start out with buses you know in the future so we need to create new ways of engaging people as we've talked about especially the next generation of users who care for other mediums, so I wish we had explored that a lot sooner than at this point in time when we have to work with a few other things so I think the timing of that could've been much better. Alongside that would've been investing in the public platform you know a lot more consistently over the past several years, even as we focused on the Teams, Enterprise business. We did that because a focus and wanting to double-down resources on a thing we said was important at the time. But at the same time, you know I can't help but think you know maybe we should have allocated more to do all the things we're talking about now. You know, back then a little bit more balanced way. The other, probably, mistake was I think overhiring in 2021. And this is, you know, the pre-[?] era that you know the economic cycles ... all sort of overheated mode an then it all sort of ultimately corrected itself in 2021. And so much like a lot of other companies, we had to do personnel cuts, and I think that one in particular I think was very difficult for me [?] as people who know me best, including at the company, know that I'm a very approachable or try, aspire to be an approachable leader. So I generally know pretty much everybody in the company, so when we had to do that, it felt really, you know, gut-wrenching for me to go through that and that was difficult, so that was certainly a mistake.

And then finally, I think you know doing this sort of thing a lot more often. I think we have not necessarily created more uniform way for me to directly speak to you; I've relied on our community manager team or relied on my in-real-life engagement with community members. I think these sort of forums always I wish we had spoken, we engage more including in forums like Meta, you know where I know there's a lot of great engagement and some of our best power users exist there so I wish I had spent more time, but I've had to sort of balance a few different things and this is where it's [?] up and so looking forward to the future.

Philippe: Thank you.

5
  • 2
    Answered at 38:39 Commented Feb 26, 2025 at 22:01
  • did you type this transcript yourself? there are parts which look wrong. did you copy it from a generated transcript? Commented Feb 27, 2025 at 2:23
  • @starball Looks like an editor did. Commented Feb 27, 2025 at 2:53
  • 2
    @starball I typed it myself. Although I had YouTube captions enabled to keep track of where I was, I didn't copy-paste them. Instead, I went through the video, pausing every handful of words to write down what the speakers said and rewatching sections as needed. But no, it isn't 100% perfect. I didn't copy it into the answer, but in VS Code (where I transcribed this), I had written a note that it isn't perfect, my notes are in square brackets so [?] and ... were words I didn't make out and that I skipped filler words (e.g., "um"). Commented Feb 27, 2025 at 4:11
  • Now that there is a transcript available, would you like me to replace @cocomac version with it or keep this post as is? Commented May 11, 2025 at 21:51
59

Is there a reason the company doesn't make use of the community it fosters here? It's always touting how great/full of experts the community is, yet, both historically and certainly recently, it's never really been a part of the company's processes outside of surfacing bugs and marketing for SaaS.

49

AMA Question:

Instead of implementing AI generated answers, have you considered using AI to other - and IMO, more useful - things? Such as:

  • give a better welcoming experience to new users
  • improve the Ask a Question page, with instructions on how to improve the question and make it on-topic, etc
  • detect spam and rude/offensive content

Note that those items don't necessarily require AI to be done (and some already exists, such as spam detector bots), but since you're willing to get on the AI hype, maybe you should consider using it on these things.

7
  • 6
    i mean... they already believe they're doing 1 and 2... Commented Feb 6, 2025 at 18:17
  • 3
    Also regarding 3: meta.stackexchange.com/q/406353/1422281 Commented Feb 6, 2025 at 18:33
  • 7
    They tried letting the AI rewrite questions to clear up issues. Instead the AI tried to answer them or hallunicated code into the question OP never wrote. Commented Feb 8, 2025 at 19:35
  • 2
    I would add "give recommendations on legibility" to the list. Like helping with phrasing and wording for non-native speakers. Commented Feb 11, 2025 at 19:15
  • 1
    I would welcome AI-assisted curation tools. Any attempt to sharpen up the Related list in the side panel would actually help people find "related" content. "Focussed Content" tool might show me pages that require editing/curation which relate to content that I've recently edited/curated. Improved searching tools without relying on SEDE queries which timeout or require me to prove I'm not a robot multiple times to complete execution. I recommend decoupling the "success" of the platform from the page visits and upvotes. SO needs better eggs in fewer baskets; now, there's too much noise. Commented Feb 24, 2025 at 23:11
  • 2
    Because useful AI will cost money. They are here to earn money, so AI intended to use us, not other way around. Commented Feb 25, 2025 at 6:07
  • For the record, I would like to say the Stack Exchange does not seem to be actually opposed to using AI to detect bad content (they support others making AI bots to detect bad content, but don't make them themselves) Commented Feb 25, 2025 at 12:59
49

@t=1:17:49:

  • How would you describe the relationship between SO Inc. and the most deeply engaged users of the public platform over the past couple years (such as those who consistently try to engage with the company on meta)?
    • Is it what you want it to be?
    • What do you want it to be, and what do you think it will take concretely to get that relationship there?
  • What have you heard from those users of their answer(s) to those questions?
    • Does your answer differ? Why do you think that is?

Answer by Philippe (@t=1:18:13):

I would describe it as- it's got some rough spots for sure, but we've got a lot of really positive collaboration as well. I've been a part of a number of sessions where we've been openly and very frankly discussed lessons that we've learned so that we can do knowledge transfer to avoid repeating mistakes from the past where we can.

We've also got a number of initiatives either running or being planned that examine things like where we communicate, how we set our expectations, and how we convey things like the deprecation of a feature, or tests that are intended to provide valuable data that we need, but which may not be really popular with some parts of the userbase. And so, it's worth pointing out that in some of those things, we're making changes that frankly are pointed to the new contributors that are joining us. We're doing some necessary revisits of processes and tools and the ways of communication that we have because we believe that with strong growth, we have an opportunity to improve on those things. So some things that might not be built for a power user can seem like a waste of time, or can impact community trust, but that's because the feature isn't meant for you. It's meant for somebody else. Over time, of course, we want those newer users to become the power users. So some of the changes that we've been implementing and that will be implemented are disruptive change. So by their very nature, some folk aren't going to like them. Honestly, if we don't make some sort of disruptive change, though, we could continue to see the number of questions drop.

So, yeah. We test out some things that aren't real popular. Some of them we keep, some of them we don't; we change some things that some folks maybe wish we hadn't, and all of these can lead to tension, and trust me when I say that I do not relish that tension. Nobody on either side of it enjoys it, but I do believe that as we get better at understanding the why behind why we do things, community members will come to believe in our good intent, and that we can gradually regain some trust from those who lost it, and we can get better at delivering features that will meet needs, even when the need isn't totally appreciated yet. And I believe that that will grow confidence in the company as a good steward for the network. But it is trust that is won inch by inch and foot by foot.

And Prashanth added (@t=1:20:21):

Sometimes I describe what we're doing is a little bit like we're governing [inaudible] of about a hundred million people. It is quite complex in that ultimately we need to listen to do that successfully- we need to listen to people, but we need to listen to people from various perspectives- not just a certain group of people, especially the diversity we have on the website: the power users, and then the newer users, etc., ultimately, in the spirit of serving the whole. That's the complexity of what we are doing- serving a very large community. That's easier said than done.

I think Philippe's response doesn't do justice to the fact that changes or features can be made with design for UX that help new users and don't frustrate power users for whom the change or feature is not intended. The recent comment experiment is a good example (and from Philippe's words, it sounds like his response is written with that context in mind, along with other recent changes like testing the stacks editor). The comment change had a purpose and an audience. In intent, it may have been a "[change] pointed to the new contributors that are joining us", but in implementation, it was deployed to everyone, and made UX worse for many people whom it didn't need to in a way that was heuristically avoidable. (It didn't even just affect general commenting either. It temporarily broke commenting in review queues, several of which are there to... serve new users (maybe there's a moral in that mini-story)).

Hurting your power users' UX to attract new users isn't necessarily necessary. And if you want those newer users to become power users like you've claimed in your answer to the question, then that would be a sadistic and vicious cycle conveyer belt human wildfire to orchestrate: make a change to attract some large number of new people, and burn some large number of those you previously attracted; repeat, but with a bigger appetite. Obviously from the your words, it doesn't sound like you want to create that kind of growth. But as I think the comment experiment is demonstrating, intention and implementation are two different things, and even if you can get us to "believe in [your] good intent", that doesn't mean you will have gotten us "librarians" to stop expecting it to mean bad outcomes for us / the library even if those bad outcomes are avoidable with some thought and effort and you would want to avoid them.

I'd also like to say that (me being a "power-user") some of the ideas I have about improving the network that I'm most interested to see considered are designed specifically to improve the user experience for non-power-users. And to my memory, whenever I express those ideas, the company doesn't seem to see it or take it into consideration. I don't feel like Prashanth's response does justice to the fact that there are power users who are keenly interested in improving the experiences of new/non-power-users, and doing so in a way that considers the complexities of serving a diverse set of parties. To me, it reads like a lecture on something I already believe.

but I do believe that as we get better at understanding the why behind why we do things

Who is "we" here? The company? Why would you need to get better at understanding why you are doing something? Are you referring to the company and the userbase? I'd find that phrasing patronizing and mildly heavy handed.


Details of this particular discussion aside, I appreciate that you held this event and prepared answers to our questions- including tough ones. Thank you :)

1
  • "understanding the why behind why we do things" Is this like meta-meta? Anyway I think it all simply comes down to doing the right things at the right time. And there Prashant is exactly right: it's easier said than done. Commented Feb 27, 2025 at 7:56
42

Most of our interactions with the CEO, and most senior staff seem to be stage managed and filtered to an extent at best (with exceptions), and rare or non existent at worse, and even if there's good coverage of topics, it lacks a certain directness. We don't really know most of the decision makers as we have in the past. Both quantity and quality of these interactions have value for mutual understanding

What're the blockers for more direct interactions with the community for senior staff, when the community is supposed to be at the center of what SO is?

1
  • 1
    This answer expands upon this comment on the question. Commented Feb 8, 2025 at 10:40
42

We've seen significant turnover in our community team due to downsizing, and I recently found out that the previous CPO (who the moderators had spoken to previously) is no longer with the company.

To an extent communities need stability, and over the last decade we've not really had much of it, with cycles of downsizing, voluntary and involuntary staff turnover, deep and repeated expressions of community unhappiness, and related negative effects.

What would you do, going forward to ensure a more stable, safe and predictable platform for the community and your staff?

1
  • 9
    In particular it is baffling how well-liked and competent CMs are fired, only to hire new junior ones some months later. Seems very short-sighted to me. Commented Feb 14, 2025 at 9:35
37

AMA question:

Will you actually take the time to personally read every one of these answers, or is this just another empty scheme attempting to 'gain the trust of the community'?

To prove that you, personally, have read this answer, please comment below accordingly.

6
  • 10
    Of course they will read all the questions. Because they want to choose those which suit them the best -- even if it means that they choose a low-vote question. Commented Feb 7, 2025 at 16:27
  • 3
    @Alex To be fair the vote isn't everything. It's biased against non-mainstream opinions, I typically also look at the extremes either way, just to know what's not popular at the moment. And why shouldn't they choose the questions, if it's their show? Ignoring the event is one option. Commented Feb 7, 2025 at 18:56
  • 22
    Well, its' me again. :-). I'm reading it (and I'll be onstage for the event itself, so I'm actually reading them all myself.) Commented Feb 10, 2025 at 14:13
  • 1
    Initially, I thought this was too pointed. Then I thought about it, and decided it's not. It was not you, @security_paranoid, who put any of this in doubt. It's no trivial matter if you ignore feedback and be opaque to people who you're supposed to lead. Commented Feb 11, 2025 at 10:05
  • 1
    @Philippe cheers, I'm very… relieved ;). This question has achieved its initial purpose (i.e. because you clearly have read it), so whether or not it is actually used "onstage" is up to you lot. I will be watching the event either live or afterwards, so I will find out then. Commented Feb 11, 2025 at 10:43
  • 5
    While I haven't been active here in the past, I am reading the questions and intend to be at least a little more visible in the future. I too will be at the event with Philippe. Commented Feb 24, 2025 at 3:18
37

AMA question:

Philippe said last year that on one significant change relevant to the community he had no decision making role.

That got me wondering - who is a high up decision maker in the company who does actually engage in the community? You have only posted one comment on the network since 2020. What expectations do you have for your executives and directors to be a part of the network they control?

36

AMA question: Was my joke really that funny?

Over the past four years, the only activity of network account 16260110 "Prashanth Chandrasekar" has been joining sites, upvotes, and this comment:

Haha @wizzwizz4 - you've cracked the code! Jokes aside, I'm really looking forward to seeing as many of you as possible IRL and having a great conversation.

I have been baffled by this for months, now. Do you have a policy of not using Stack Exchange?

  • If so, why break that policy here?
  • If not, why not weigh in on discussions on Meta? That would go a huge way towards restoring a healthy company-community relationship: even if you ignore 95% of what we're saying, we would at least know that you thought about it, and deliberately disregarded it.

And, most importantly: did you actually find it funny, or were you just being polite?

4
  • 8
    Whatever happened to that event anyway? Commented Feb 10, 2025 at 19:52
  • 8
    @cottontail cancelled/postponed with no new date given. The only explanation given to planned community attendees was that the date no longer worked for the SE attendees. Relevant answer Commented Feb 10, 2025 at 22:42
  • He's not breaking the "not using SE" policy, since the AMA is on youtube. Commented Feb 25, 2025 at 12:37
  • @Gloweye "Here" refers to the response to my joke. Commented Feb 25, 2025 at 18:39
35

AMA Question

With the announcement for the new Answer Bot feature, the staff member says

We conducted research and facilitated discussions with Stack Exchange community members and moderators about this concept throughout the latter half of 2024, to both get initial feedback from various types of users, and to identify communities that saw value in testing it out.

How many people were surveyed, and what was the fraction of mods/staff to regular users? How does that apparently-positive feedback connect to the fact that the announcement received well over two hundred downvotes and only a dozen or so upvotes within the first day (with -500 downvotes as of 20 Feb 2025), with many prominent users directly disagreeing with the concept of the experiment, including apparently some of the mods who want to do the experiment to get the facts that it’s a bad idea?

0
32

AMA question:

What will you do to bridge the (increasingly wide) gap between SE Inc. and the (online) community? In what concrete ways will you try to steer both groups into the same mutually beneficial direction again?

32

In October 2023 - the company decided to pivot towards AI, and perhaps not coincidentally, we had staff cutbacks.

From the perspective on the ground, we've had a few attempts at deeply unpopular features, and nothing that really feels successful.

What would you consider to be the criteria for the AI experiments to be considered a failure, and on the other hand, what would you consider to have made all this worth it?

Answer by Prashanth Chandrasekar (CEO) and Jody Bailey (CTO):

Prashanth: got it no thank you journey man geek I've enjoyed your engagement in meta and I've seen you know a lot lot of thoughtful comments in there over the over the many years so thank you again for your thoughtful question

I think if you think about the context if you go back this a little bit in 2023 the company had like these two big enormous challenges right the first one was the rise of gen and chat gpt's launch in November of 2022 that had sort of an immediate impact on our traffic since it was back then a you know direct substitution to what we you know what we stood for and what we did for for users and the second big issue that we were facing is sort of the sign significant economic headwind that was in the post pandemic era in at the end of 2021 and you know customers were pulling back on spending on SAAS software on advertising on you know spending on job boards Etc so our staff cutbacks we had unfortunately had to do them to correct this over hiring issue that we had I explained previously relative to this demand sort of slowdown and as I mentioned previously very very brutal for us to go do that and you know not you know definitely not one of my happiest moments of the company given you know all the people that we affected there now

we also had to respond to those AI heads so I you know I personally whether it's correct or incorrect we set a goal at the company to say okay look this is a real it's a real sort of shift that's happening at the end of 2022 that this is like a few months after that so we set a goal to say we're going to come up with our own AI response what is our response to AI and let's do it within a very short period of time you know let's really have a tiger team that focus on this so we organize you know 10% of the company Etc and we set this goal of announcing our AI strategy and implementation six months out or so at the WeAre developers conference in Germany and the general goal was not to leverage AI for just AI sake just to be very clear but was it to the whole point was to leverage it in a way that number one solved user and customer problems and number two it would Delight you in ways that you didn't sort of expect so both those were in our minds and we tested these AI features that we built we built several that integrated into things like IDE into into Visual Studio code into slack into Microsoft teams and we did this both on the public platform as well as on Stacko for teams our private Enterprise product and net net what we saw was that the on the public platform where they were there were actually you know the search and summarization features were actually relatively well received but several other features theoretically you know solve customer problems but like surfacing content in in their workflow like through those tools I mentioned or being able to ingest content from external places but not everything was you know was was you know well received like you know people said okay they like this and they didn't like it so we basically retired in response to your feedback the team decided to retire the public platform AI functionality because it just wasn't I suppose a big enough problem to solve for you folks it wasn't you know meaningful enough and so we listened to your feedback and that's what we did and it was a right decision and we started focusing on other workflow items you know for example what are some other problems that people care about so for example the question asking experience was something that people have talked about a lot given the new user experience that we described that both Philippe and Jody have talked about and so we've now integrated of course AI assistance in the stage ground area for users and now even Beyond staging ground we're testing that so all questions are getting sort of an AI assistant and that's been relatively well received by many of you so especially the newer users seem to like the fact that they can place have this sort of safe sandbox to be able to craft a question that doesn't have you know human input and then they can present their best put their best foot forward

so that's one example of us of of doing something that is pivoting kind of our focus and then e thing with the answer experience you know we want to also be able to say how do we test AI in that context is that useful when people are writing answers you know people are getting used to getting prompts on how to you know even let's say you write your email in Google and Gmail you know you're getting fed you know sentences as you're completing sentences so should you be able to do that that's a question for us we want to be able to see would you find that valuable so we're constantly going to test and you know make make this site you know solve your big problems basically and and it really sort of help you know get unstuck you know any of your sort of workflow items you know to be able to ask questions get recognized find jobs and so on and if AI happens to be the answer to that solution great if not we're not sort of married to the concept necessarily right so we are we're not going to shoehorn AI where it's not needed we use it wherever it makes sense and especially if you find it to be valuable that's the most important thing yeah I

Jody Bailey: I might chime in there too that you know we we do make mistakes not that that I need to tell anybody that right so we've recently run an experiment of answers with AI right where we were using AI to provide an answer with the idea that people could edit it to provide an answer and the the resounding feedback was that that doesn't work for for our question answering population that it's more work to edit an answer than it is to get help with an answer so you we've run the experiment I think it's still running just to get feedback but but we clearly you know the approach there for a number of reasons is not the right approach and so you know we're not going to force that we're not going to shoe horn as you said we'll go back to the drawing board and and try to figure out you know how can we help with the the answering process and does AI play a role in it and if so what is that role

31

A lot of fanfare gets made for new features, while old ones (and/or longstanding bugs with them) languish. On the other side of the coin, there's an entire section of our community who are pretty dedicated to improving things for the whole community - there's plenty of SEDE queries and userscripts floating around alongside bigger efforts like Smokey and the like.


AMA Question

Is there any appetite for giving the community more power over customising any level of the public sites?


Some examples of pain points I've seen raised in the past:

  • The ability to submit site designs - both for sites stuck on the SE default AND changes/bugfixes to existing custom designs. This could be accomplished through Github merge requests or similar - I.e. still keeping control of the repo (and the power to accept or deny changes with SE staff).
  • Fixing or enhancing other aspects of the site, e.g. site tooling, new user onboarding, Ask question help etc. Setting up tag blocks and warnings ourselves.
  • The ability to set up things like collectives at a community level.
  • Customising more of the help center/FAQ/Tour messaging
  • Changes or additions to auto-comment systems (eg duplicate text, LQ review queue messages, Community Bot messages etc)
  • Tweaking site settings (like number of close/reopen votes required, amount of available flags per user, reputation levels for feature sets etc). - Currently possible through contacting CMs but (IIRC) most of these still require Dev or DBA time to make the changes.
2
  • "Currently possible through contacting CMs but (IIRC) most of these still require Dev or DBA time to make the changes." – Yeah, though it depends on the setting – many of the ones that the company would be likely to customize per-site are editable directly by CMs. And the ones that CMs can't edit, they can at least request the ability to be able to edit themselves, if there's a good case for changing the setting. Commented Feb 21, 2025 at 19:45
  • @V2Blast yeah, I guess what I'm saying is how much of that control can be handed over to the community (via new privileges) or the mod teams? There have been sites where requests to tweak stuff has gone through community discussion, been tagged status-review for action, then languished for months or years at this point - when what you're saying it boils down to a flick of a switch or an addition to a form. Surely SOME of that can be offloaded from the CM team's bucket of tasks :) Commented Feb 22, 2025 at 12:48
30

AMA question: Why is attribution optional?

Lately, the company keeps saying "attribution is non-negotiable", and then using / endorsing systems that fundamentally cannot respect attribution. What's up with that?

Answer by Prashanth Chandrasekar (CEO):

got it thank you wizz wizz for the question that's that's an important one I think the attribution is absolutely very important we consider it as is completely vital it is a non-negotiable as we engage with Partners but it's also a matter of you know us holding them accountable when they on their road map and as they it does take some time for partners to incorporate that into their AI tools as we do these Partnerships with them and where they get licensed access to the data as an example so it's still in sort of an early infancy stage despite the fact we've struck multiple of these Partnerships they're all in the process of attributing so as an example one recent example is with Google Gemini Cloud assist that which is now attributing Stack Overflow content in the in their IDE right as you write code through that through that tool you will see the sources back to the Stack overflow Content that was was used you know originally anecdotally also I would just say that you know our head of marketing Eric Martin shared this with me earlier today is that the with open AI we now you begin to see things like sources within there and you can actually see the links again to Stack overflow if you're in and we have actually seen a nice increase in the traffic that comes to our site from from chatgpt actually and you know obviously the opposite is happening on search for all content sites because that's just a it's a different world for search but as gen AI and search trade off on you know you know getting access to users is sort of the primary screen that people spend time on to get access to information on the web we're literally seeing that happen on our site where traffic seems to be coming now from both places people are coming definitely from Google but also our search engines and people are definitely coming from now our some of our knowledge as a service integration so it was interesting to see the chat GPT traffic through this attribution point so it is a long way of saying wizz wizz it is a a work in progress and it is is happening as we keep working with each of these Partners to to hold them accountable to that requirement thank you

4
  • I guess they thought the systems would somehow learn that and demanding it would set themselves apart, and when they learned that they can't, they regretted the move and don't know how to backtrack without looking bad. But let's see their explanation. Commented Feb 10, 2025 at 6:48
  • 3
    @NoDataDumpNoContribution I have told them, a few times, that I know a way to introduce genuine, information-theoretic attribution to transformer-based LLMs. (I've described it in some detail.) What with the DeepSeek optimisations, it's feasible to do this. However, this wouldn't solve the "they don't actually work" problem, so (unless SE wants to try an actual LLM use-case), the main purpose of doing it would be political: rebuilding bridges with the community, challenging the narrative around copyright to maintain SE's "sell training data to AI companies" business model… Commented Feb 10, 2025 at 10:15
  • 2
    @wizzwizz4 Now I'm very curious; where have you described these improvements? I'd love to read more about this. Commented Feb 12, 2025 at 3:28
  • 1
    @Draconis I think just in moderator-only spaces. I haven't done the research yet (I'd need funding, and/or a compute donation), so I can only be confident about my would-definitely-work approaches, which have caveats like "statistics I don't quite understand is required to extract provenance information" and "model size scales with number of attribution buckets". I think these issues can probably be solved, but… I'm not confident speaking in public about things I haven't actually checked. Commented Feb 12, 2025 at 9:32
27

AMA Question

A significant number of valued members on SO–and other Q&A sites–have left the network since 2019. A few have been replaced by equally qualified and experienced users but active participation appears to be limited.

What can be done to encourage "old" experts to return or for new ones to participate on Stack Exchange sites?

2
27

Bringing up something I brought up in 2023

Is there an actual plan for community growth as far as getting people to stick around? Sometimes it feels like it's a marketing term rather than an actual desire to do better. The community is people, not AI.

It feels like the focus is on new signups, rather than reconciliation with the community, encouraging people to stick with the community and building healthy communities that are self sustaining.

What are your views on the 'deeper' parts of the community funnel, attracting former users back and what would you do to improve the way things are done now?

1
  • 6
    Well yes but I think retention is missed out here and pretty important Commented Feb 9, 2025 at 10:36
27

AMA Question

Since the rise of large language models, the number of new questions asked on Stack Overflow has dropped from 6,000 per day in February 2023 to just 2,000 in February 2025.

Given this shift, are there any strategies Stack Overflow is considering to encourage users to interact with real people, rather than relying solely on LLMs? And if the trend continues, with fewer questions being asked, what do you see as the future of Stack Overflow?

Answer by Jody Bailey (CTO):

Yeah I I feel like we've kind of talked about this a few times already you know the you know the three-lane highway that that Prashanth mentioned right providing different ways for humans and people and primarily technologists when we talk about Stack Overflow for them to interact you know trying to identify different mediums meeting people where they are you know trying to help keep engineers in flow all of these things are are areas that we're exploring in order to you know create a space for people to interact more you know we talked at the very top you know pant talked about bringing humans and Technology Ai and humans together in community right and so that's a big part of our mission is how do we actually get people talking to one another in order to get better Solutions better answers etc so we've got a number of things that we're focused on and you know we're going to be continuing to work through those as we go.

5
  • 3
    e.g. will you train LLMs to ask "well-formulated" questions in order to increase user interaction? Commented Feb 10, 2025 at 14:57
  • 1
    Users stopped dropping before that though, so I'd be hesitant to blame the AI hype for it. The AI hype might have lead to a decrease in easy to answer beginner questions, but the user and activity drop is consistent among new users and veteran users all alike, including moderators. Commented Feb 14, 2025 at 9:44
  • 1
    We want the question rate to drop so the number of new questions stops overwhelming the curators. Commented Feb 15, 2025 at 23:08
  • 1
    New questions asked is the wrong metric. At the least, you should be filtering by some quality metric, in addition to whether the question receives a useful answer. A better metric might be something along the lines of "questions that receive an answer upvoted by someone other than the asker", as it captures whether the question provides value to more than one person. Commented Feb 21, 2025 at 0:14
  • Perhaps "fewer questions being asked" attests to the benefit of the site -- Most novice questions have been answered and are easily searched for ! Commented Feb 26, 2025 at 19:46
25

Some meta-commentary:

We will continue to monitor this post until the AMA has concluded, but the best way to make sure your questions are submitted for possible inclusion in the AMA, especially during the week of the event, is to fill out the question submission form.

Why? You have a perfectly-good Q&A platform which, on top of already providing the requisite infrastructure, allows the community to self-select the questions they find most important. The question submission form is by its very nature worse as an AMA tool - the only "benefit" it provides is preventing community input so the company can pick and choose which questions they want to answer instead of what the community cares about.

Relatedly, please don't ignore the scores on these questions when picking which ones to answer. It doesn't need to be strict adherence, but I really don't think you should answer a -1 question above a +30 one unless there's no other option. Picking and choosing the ones that fit your marketing goals over the ones the community actually cares about amounts to disingenuous non-engagement that will seriously hurt my trust in company leadership going forward. There's been so little communication from higher-ups since well before I joined the network; this AMA cannot replace genuine engagement with marketing or publicity.

TL;DR: This needs to be an actual AMA - not just a chance to talk about your shiniest new AI tchotchke.

10
  • 4
    I agree that its best for questions to be submitted on this post so that post score can be taken into account. The key part of the sentence you quoted is "especially during the week of the event". What I was trying to express was that if someone leaves an AMA question on this post say, the night before or the morning of the event, I can't guarantee that we will see it. It takes time to go through posts with lots of answers. In the hours leading up to the event, it will be easier for us to see questions that come in through the form. People can also ask questions during the event. Commented Feb 14, 2025 at 15:47
  • 3
    I'm sorry, but an AMA is not supposed to be a struggle session. If the most popular question was "what color underwear are you wearing?" do you want them to pick that over a question that made them think of something meaningful to communicate even though the meta community didn't like the way the question was phrased? These questions are opportunities to communicate, not test questions that have to be answered exactly the way the querant would prefer. Commented Feb 14, 2025 at 16:00
  • @ColleenV I agree that this should be an opportunity to communicate, which is why score matters. If "how have/haven't you been considering community input in feature rollouts" is +30 and "AI seems really cool; how are you going to bring its awesomeness to SE in 2025" is -5, I am very certain that the company would have things that they think are "meaningful to communicate" regarding the second question even though the community at large doesn't really want that to be answered. Commented Feb 14, 2025 at 16:18
  • "What color underwear are you wearing" is obviously silly; that falls under the "no other option" to me. But more importantly, that wouldn't rise to the top anyways because the MSE community knows they do and don't find important. The community is much better at judging what communication is and isn't important than the company will be; there needs to be some trust in MSE to self-regulate. Commented Feb 14, 2025 at 16:21
  • I think what I'm really getting at is given Prashanth's and the rest of leadership's track record with communication, I am not particularly optimistic that this will be carried out in good faith. This post was an attempt to discuss what good faith does and does not look like in the context of a Stack Exchange AMA. Commented Feb 14, 2025 at 16:25
  • 2
    "what is seen by management as the single biggest danger to continued success of the platform?" is an interesting question that could have an insightful answer but isn't getting upvoted. We all know the issues around post score and have argued about it a lot right here. The questions they choose to answer will say a lot, but it's unreasonable to demand that they forgo answering questions with negative scores even if they have something interesting to say about that topic. The score is information but it's not a replacement for looking at the content. Commented Feb 14, 2025 at 16:40
  • @ColleenV Sure, not a demand. I edited the answer to reflect that. But it really does matter. I want this AMA to be engagement, not a publicity stunt. Commented Feb 14, 2025 at 16:58
  • 4
    I think most people are hoping that it will be substantive, but we've been burned before. Commented Feb 14, 2025 at 17:00
  • 3
    @ColleenV - I'm pretty sure this is just going to be a bunch of corporate double-speak and an announcement about them dumping AI crap onto every answer Commented Feb 15, 2025 at 19:36
  • 1
    @ColleenV Answering questions with negative ratings is the same idea as implementing features with overwhelmingly negative votes -- with the same motivation: because it benefits them the most. Commented Feb 24, 2025 at 16:10
21

AMA question:

How long do you see, financially, SE/SO lasting? Is bankruptcy/retirement a realistic future, or is 'the plan' to keep the company running indefinitely?

5
  • 5
    This is my favorite answer so far. I worry that based on the way things are going, it might not last for as long as I'd like. Commented Feb 8, 2025 at 15:11
  • 1
    Do businesses plan to retire/go bankrupt? It often simply happens despite all efforts. Commented Feb 9, 2025 at 9:12
  • @NoDataDumpNoContribution which is why I asked if “bankruptcy/retirement is a realistic future”, but also, some sub businesses do decide to retire, maybe, say, if the parent business feeding the sub business is losing money to it. Commented Feb 9, 2025 at 9:42
  • And even if a business doesn't want to retire/go bankrupt, they will still have to plan for it. Commented Feb 9, 2025 at 9:43
  • 27
    No CEO will be dumb enough to say “yeah my company’s about to fail, that’s my plan” Commented Feb 9, 2025 at 16:22
19

In your opinion, why do people volunteer their time to curate and answer questions on Stack Exchange? Does restricting access to the data dump impact people’s willingness to donate their expertise and time to a for-profit company?

2
  • 1
    can why a large and diverse group of people do an activity/activities be a matter of any one person's opinion? as it's currently written, this seems like a question to the community (the source of the answer) rather than the CEO. Commented Feb 10, 2025 at 17:54
  • 5
    @starball I would like to know what the company's understanding is... the question has already been asked of the community many times in several forms. Commented Feb 10, 2025 at 18:08
16

Not a question but a comment. You say this event is "for members of the Stack Overflow and Stack Exchange communities", but I don't think it is. Instead I think this is a show being put on for someone else - an Important Person with money or influence who needs to be impressed - and it's for them, not us.

I think you know what to do if you want to avoid giving this impression, as other answers have explained it very well, on this question and many others. You probably know this already and don't care, but you're not really fooling anyone with this, and it appears incredibly disingenuous.

15

I am not interested in seeing some CEO of some company, even though I am using their site on a daily basis.

The first hour (2 parts) could be published on Youtube, there's no really good reason to do it in stream. The only reason to streaming it at all is for the last half a hour, that could be something interesting, though there is always a chance of scamming (prepared questions are posted by associated peoples in chat and only those are chosen).

This whole event sounds way too pompous, which reminds me of Putin's direct lines, where "beloved president" solemnly descends to a talk with plebeians, finally forcing regional administrations to do what they are supposed to do (but they don't, since they are not elected but rather assigned from the top), promising to slap them on their wrist and that such things will never happens again... /facepalm. I hope I am wrong and whole thing would not become a circus.

This event is a strange thing for Q&A oriented site. I don't see any reason preventing a CEO from participating in meta-discussions, answering questions or just giving his comments here and there. Of course he is very busy, but probably something goes very wrong right now, so finally there is 30 min of his time reserved…

I'd like to read a comment here, under my post or a similar question, regarding such an unusual move, rather than seeing talking face on youtube. Why doing this event at all? Why now? Why not before?

2
  • 3
    This should be a comment, not an answer. Commented Feb 10, 2025 at 14:56
  • 3
    It's true that a question needs to be asked at the start, otherwise your very good points remain as critical comments. How about putting in bold, AMA question: Why should I be in interested in seeing and listening to the CEO of this company, given how pompous the CEO comes across as? (If you don't post a question, you have no chance of receiving an answer.) Commented Feb 10, 2025 at 17:43
15

A while ago Philippe asked the community a series of questions: https://meta.stackexchange.com/questions/tagged/community-vp-questions

What insights from the responses to those questions, if any, did you find useful in planning future development of the public Q&A platform.

4
  • 6
    @Starship Slate went out of their way to explain how the discussion wasn't likely to result in any changes, so I don't see the point. Also, Slate is not the Community VP. Commented Feb 11, 2025 at 13:28
  • Yes, but I would like to know if they learned anything from it Commented Feb 11, 2025 at 14:23
  • 2
    @Starship I'm not sure why you'd expect leadership to be at all aware of that discussion though. If you want to know if anything came out of it, you should ask Slate. As far as I can tell, none of it got outside a very small group of mods and Meta users. Commented Feb 11, 2025 at 15:50
  • 2
    @Starship "if they learned anything from it" Maybe that their community still doesn't understand them. And we would probably try to if only they would educate us. But their plan is only known to them so far. I tried to read the blog posts for a while but I only remember "AI is the future of community" or something similar from it and gave up. Commented Feb 13, 2025 at 22:17

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.