18

There was a recent question Should SE switch focus on improving curation instead of creating new questions? which proposes one way of adapting to the massive drop in questions at Stack Overflow (like many major sites). But before we consider "do something", I think we better consider "do nothing" (which is usually far easier to implement).

We're presented with a trolley problem: "do nothing" or "do something" (and opting out of making a choice = "do nothing", so we must choose one or the other):

Stack Exchange's trolley problem in meme format

To make a better-informed decision as to whether or not we should pull the metaphorical lever...

Question: With a massive drop in questions, what happens if we maintain the status quo?

I really don't know. My current thoughts are:

  • Just because a site is small, doesn't mean it's bad. The question drop could just taper off, and we'll reach a new, perfectly acceptable "status quo".

  • Maybe there's a risk that sites will become "top heavy": lots of experts, but not enough questions for them to answer.

  • Somehow Stack Exchange will need to "pay the bills"; the ultra-responsive site we've come to expect at Stack Exchange comes at a price. But that needs views (right?). (Do we need to cut costs?)

  • We need to simultaneously satisfy:

    • Random people on the Internet (views, new users, questions, etc.);
    • Contributors (questions, answers, curation, moderation, etc.);
    • The company (server costs, maintainence, marketing, etc.).

    If we "do nothing" will all three be satisfied?

  • Stack Exchange is selling our work to AI mega-corporations. Is that alone enough to "pay the bills"? Is this long-term acceptable to users? Will revenue drop as questions drop?

  • With current trends, will Math.SE (or some other site) overtake Stack Overflow and thereby become the new Stack Exchange flagship site?

    screenshot from https://stackexchange.com/sites#questionsperday showing Stack Overflow with 209 questions per day, Math Stack Exchange at 54 questions per day, and Ask Ubuntu at 16 questions per day

  • Is "do something" even an option? E.g., if the Internet has indeed changed on a fundamental level, and if people have far-reduced interest in learning programming (and other topics) through a Stack-Overflow-like Q&A interface, then "do nothing" and "do something" might be the same choice anyway.

Improve this question
8
  • 9
    Don't get me wrong, but it seems somewhat pointless to me to discuss changes that we (the people discussing this topic here) cannot bring about. Only SE-the-company is in the position to change something. And even they don't have much control about alternative services which seem to be more attractive for potential visitors than the SE network. Commented Jan 12 at 10:10
  • 2
    Since you posted this on meta.SE maybe some statistics for the network as whole would be more relevant. SO is the flagship for sure and having problems for sure, but I don't really notice much of a decline on other SE sites. Commented Jan 12 at 10:16
  • 4
    @Lundin there is a decline in the rest of the network, though it's slower. Other sites I picked declined to ~1/5 of question volume while SO declined to 1/20. It's more difficult to show good graphs here, and my query is skewed by large spam waves (so Superuser looks pretty weird in it). Commented Jan 12 at 10:46
  • 8
    Aside from running "number of [...]" data explorer queries, I note that 10 questions per day used to be a minimum requirement for site launch, and now there are only 9 sites left that have 10+ questions per day. Commented Jan 12 at 11:09
  • 11
    What @DocBrown said with the addition of: none of the recent company announcements e.g. so.AI, design overhaul, comment changes, chat rep changes, SO opinion-based questions, feel like the company is interested in taking any community input. Negative feedback is handwaved away, even clear license violations (lack of so.AI attribution) were ignored; at best they act on minor feedback and make small adjustments to things they were already doing. Posts like these kind of feel like shouting into the void. Commented Jan 12 at 12:34
  • 2
    @Lundin "… but I don't really notice much of a decline on other SE sites." the English sites have been affected badly: Reason behind drastic decline in viewership and upvotes in ELU posts Commented Jan 12 at 18:40
  • 2
    @Lundin "I don't really notice much of a decline on other SE sites." See The number of questions has decreased 10-fold: where did users go and what can be done to prevent that?. It includes stats for each SE site: i.sstatic.net/4h76DtCL.png Almost all SE sites have a massive drop in number of questions. Commented Jan 12 at 18:55
  • 1
    Other than involving a choice between nothing & something, the post's options/decisions have nothing to do with the trolley problem. Your characterization of the trolley problem totally misses its point & trivializes it. Commented Jan 15 at 2:21

6 Answers 6

Reset to default
20

With a massive drop in questions, what happens if we maintain the status quo?

My best guess? Prosus sells SE to someone that just fires everyone and squeezes as much money as possible out of the remaining views. Or they do something similar themselves, reduce operations to the absolute minimum and jack up the ads.

I would assume that the page views also dropped significantly, but far, far less than the number of new questions. There's a lot of existing content that still gets views. But companies usually want to see growth, and the activity today is an indicator on how the page views that are important to revenue might look in the future. And the page views are also under direct threat as well from Google AI summaries.

Stack Overflow and the network are also important to the company for the prestige and the audience. Which in turn they try to use and sell Stack Internal. I've no idea how that's going, but I have my doubts that they're particularly successful here. A strong SaaS product could carry the public sites simply for prestige and advertisment, but I doubt that is the case here.

I also doubt that the AI companies will be willing to pay long-term here. Unless the licensing question is decided in a way that would force them to buy content like this. Otherwise I don't think they'll bother at some point.

But apart from that, I also think that the sites will suffer from the lack of activity. There could be a bit of an uptick in quality temporarily as the number of questions drops faster than the number of curators and qualified answerers. But those people don't necessarily stay around on a dead site. And a lot of other systems will fail with low activity, e.g. who will upvote the good posts?

Improve this answer
4
  • The peculiar thing is that everyone could already notice the decline trend before Prosus bought it. The big question is how anyone could ever think that anything but (targeted) advertising would work as a business model for this site - all byproducts like company sponsorships or the "Teams" thing aren't going to bring in money in the long term. Commented Jan 12 at 10:44
  • 4
    There's been speculation (far-fetched or not?) that the purchase wasn't even meant to bring in profit since nobody sane would buy SO in hope to make profit. But that was just part of the general "white washing" that Prosus is doing on behalf of their apartheid-associated parent company Naspers. More info: physics.meta.stackexchange.com/questions/13609/… Commented Jan 12 at 10:45
  • 1
    It's worth remembering that Teams was relatively newly-launched when the purchase happened and it was seen as a pretty major revenue stream. Microsoft Teams didn't add Q&A until 2022, so I think it was seen as a really viable product offering... now that it's part of MSFT Teams, though, I imagine it's struggling to compete. Commented Jan 14 at 15:06
  • Prosus basically bet a fair bit on the ed-tech market right around covid. If nothing else, we are probably in better shape than one of their other investments, a formerly massive indian ed-tech org called bijous IIRC Commented Jan 15 at 10:29
14

All things come to an end. The time to "do something" was when the line stopped going up. There's never been a vision (as far as I can tell from the outside) for Stack Overflow 2. The company and the community just want to keep twiddling and adding on to what exists without changing its fundamental nature.

From my perspective, the only place to start is from the beginning. Put what exists in maintenance mode and start building the next generation. Instead of asking how we can change what exists into something else, we should be asking "What is the goal of Stack Overflow 2? How does it make money? Who is its audience?"

A lot of the problem with trying to fix the decline is that the people who are really engaged don't want anything to change, but SO needs to fundamentally change to keep growing. By trying to "fix" it in place instead of building something new, you're destroying it for the core community while carrying a bunch of baggage that alienates a potential audience for the new thing. It's no wonder the company hasn't been having much success with these initiatives.

The company should stop shoehorning things in and start building.

Improve this answer
2
  • 2
    "Put what exists in maintenance mode and start building the next generation." — Yes. Ideally, on a different domain under different leadership that actually deserves it. Commented Jan 13 at 7:01
  • @KarlKnechtel Well now that the whole network is supposedly going to be branded as Stack Overflow, they could build it under Stack Exchange for maximum confusion lol. Commented Jan 13 at 17:04
7

I'll quote from @Mad Scientist's answer above

There could be a bit of an uptick in quality temporarily as the number of questions drops faster than the number of curators and qualified answerers. But those people don't necessarily stay around on a dead site. And a lot of other systems will fail with low activity, e.g. who will upvote the good posts?

I think SE is still the best Q&A site on the internet for expert answers and for its functionality. Period. I've considered other sites but SE is still where I'd want to ask questions and if I'm posting answers it would also be on SE (except the odd GitHub repo)... Q&A as a format isn't going away, the noob level questions that are likely 90% of the market are what's AI is good for.

To address the OP's thoughts:

Maybe there's a risk that sites will become "top heavy": lots of experts, but not enough questions for them to answer.

Where there are experts there'll be users wanting to tap into their expertise.

We need to simultaneously satisfy:

Random people, Contributors, The company

If we "do nothing" will all three be satisfied?

The company has been doing A LOT under the new management, in fact they haven't stopped seeking avenues for innovating and diversifying these past few years.

With current trends, will Math.SE (or some other site) overtake Stack Overflow and thereby become the new Stack Exchange flagship site?

Maybe, but what is the "flagship"? Several of the +120 sites on SE are likely the best in their respective domains... "Volume" and "activity" are transitory indicators, I've been active on SO tags that simply matured and the repo is saturated, few new changes warrant new Q&A but the knowledge is still there.

people have far-reduced interest in learning programming

It's a knowledge based job market that fuels SO in particular. IT and computer science are here to stay. And SO is still the "go to" for anything requiring real expertise - except in a minority of cases where companies and products host their own Q&A.

Improve this answer
5

With a massive drop in questions, what happens if we maintain the status quo?

As far as I know, SE's main sources of revenue are:

  1. Stack Overflow for Teams (SaaS)
  2. Data Licensing
  3. Stack Overflow Advertising

The number of questions plunging means a sharp decline in revenue 2, since data licensing is only useful to get access to content published after the latest data dump.

I assume the traffic is also taking a dive, which means revenue 3 is also sinking. Related: Is there a web traffic statistics page for SE and if so where is it?

Remains revenue 1 (Stack Overflow for Teams). The sustainability of Stack Exchange currently depends on it.

Improve this answer
3

I don't think doing nothing is an option. But what we do is important - do we focus on building out a community that thrives here, as we'd done early on, and invest back into the community we have or abandon it and try to do something different?

I've never really felt building and maintaining should be mutually exclusive to other pursuits. On the other hand, The fall's not just because of generative AI. We do have to deal with natural attrition, but there's also a certain sense of both not really getting much support on dealing with haters, and a certain stagnancy when it came to the community.

I don't think we can maintain the status quo - but I disagree we'd need to focus on appeasement of folks who're toxically insisting we're toxic or changing core design features in the hope it'll increase raw question counts.

We don't just need people, we need more people invested in what we do here, and sticking around, and not to lose them faster than we grow em.

AI doesn't help there - nor do some of the changes that are proposed lately IMO, without quite a bit more work to understand the effects it has.

Improve this answer
8
  • 6
    "I disagree we'd need to focus on appeasement of folks who're toxically insisting we're toxic or changing core design features in the hope it'll increase raw question counts" - a thousand times this. For me personally, the amount of questions is irrelevant, the signal-to-noise ratio is the main factor - and almost everything SE does continues to make this worse instead of better. Just having more questions when most of them are crap does not make me want to be an active answerer again. Commented Jan 12 at 12:40
  • but I disagree we'd need to focus on appeasement of folks who're toxically insisting we're toxic or changing core design features in the hope it'll increase raw question counts. Maybe not in the hope it'll increase raw question counts. But the issue persists. Recently there were a bunch of YT videos showing that particular statistic data.stackexchange.com/stackoverflow/query/1882532/… . And more than one of them went into detail of the toxicity abounding. I sure do hope the powers that be are aware and have a plan. Commented Jan 12 at 13:59
  • 4
    But is it? "My question was closed as a dupe" isn't toxicity. Its a system working as designed, with the parts of the system that were meant to mitigate that being broken, and all alternatives ignored. And there's been a focus on the unhappy folks for a while. Maybe its worth trying something different and not losing both groups of people. Commented Jan 12 at 16:04
  • 4
    @JourneymanGeek the perception matters here more than the reality or the intent. And I think this is an area SE should investigate as either a lot of people can't understand how the duplicates solve their issue, or we are actually making a significant amount of bad duplicate closures. Commented Jan 12 at 16:34
  • 1
    And spending some effort, and maybe even marketing on changing perceptions. I've a proposal for changing dupes stewing but I'm not quite happy with where it is yet tho. Commented Jan 12 at 22:57
  • 2
    @JourneymanGeek, "My question was closed as a dupe" - maybe not so toxic. But "My question was closed as a dupe, but it's not a dupe and none of those answers have anything remotely to do with my question anyway!" is kind of toxic. And before you ask me to provide a tome of examples, I'm merely posting the perception of new/seldom users. LLM's never close a question as a dupe. They might give the wrong answer based on very similar requests/questions because of the way they work, but they don't just close the chat box and say "Already been asked. Go look here. Good luck and goodbye". Commented Jan 14 at 17:54
  • 1
    "...maybe even marketing on changing perceptions" And I don't think marketing is going to help much either. I'm getting a bad vibe about how this is all going. This feels like trying to get a VHS producer to market the advantages they have over DVD's. Sure there might some arguments there, some perhaps very well made, but it's blatantly apparent who's going to win that one. The thing that SO has over all LLM's is new knowledge creation and archiving. But how many of those questions/answers have there ever been? I suspect it's always been a small percentage - hence the need for moderators. Commented Jan 14 at 17:55
  • They likely wouldn't get it indeed - but I'd rather one user who gets is and is in for the long run, than someone who's going to get immediately mad, or worse, never comes here cause they believed some random person on another site, who might or might not ever come here. Commented Jan 14 at 23:41
-6

Today I asked Claude a question about CI/CD in VoIP development. After a great discussion, I wondered: "What would Stack Overflow say?" CI/CD problem in the VoIP/RTC development

Claude predicted: 65% chance of good answers, 25% risk of closure.

claude

Reality: In under 3 hours—3 downvotes, 3 close votes, 1 delete vote. Zero explanatory comments.

Not just closed. Someone voted it shouldn't exist at all. This is why I ask Claude now, not Stack Overflow.

So my first suggestion is if you cast the deleted vote you need to explain why in comments.

Improve this answer
13
  • 14
    "Reality: In under 3 hours—3 downvotes, 3 close votes, 1 delete vote. Zero explanatory comments." so, in other words Claude completely lied to you. But it sounds like you blame the site for not conforming to the BS Claude said. Rather than condemning Claude for saying BS. Commented Jan 14 at 15:36
  • 7
    So what you're saying is, claud was quite innacurate Commented Jan 14 at 15:37
  • 10
    I'm surprised you took Claude at its word, given how wide and unfocused the question is. You should have the experience to know the question you posted isn't a good fit for SO. To paraphrase it, it's "I am doing something with VoIP and I need to ensure the audio quality. How can CI/CD help? And what unit tests should I add?" Which is really all over the place topic-wise. Commented Jan 14 at 15:38
  • 3
    The question you asked is marketedly off topic as a Q&A question. you may have had better luck asking it as a discussion, since a discussion is likely the kind of answer you're looking for Commented Jan 14 at 15:39
  • 2
    "After a great discussion". That means your initial prompt is not clear enough to provide an answer even considered good enough to you. Did you incorporate the subsequent clarification info in all following prompts to your SO question? Commented Jan 14 at 23:47
  • 1
    @VLAZ Everybody here is attacking instantly, but the reality is SO is a place where if you go to ask a question about a real problem you're facing you're more likely to face moderation than answering. People here know how to ask a good question that won't get closed, sure. But the default new-user experience is "if I go to an LLM, I get an answer. If I go to SO, I'm going to probably get closed, but I might get a better answer." Commented Jan 15 at 0:42
  • 2
    SO needs to be more willing to just help people. It's constantly shutting out new questions and enforcing quality standards that need to change. People were fine with just dealing with those standards before, but now that LLMs are an option, people are like: "I don't have to deal with this. I don't have to ask a super awesome, well-researched, laser-focused paper of a question. I have another option. That option is AI, and it doesn't harass me, downvote me, mark my question as a duplicate of outdated crap, or delete my question. It'll answer the best I can, and I have unlimited access to it." Commented Jan 15 at 0:43
  • To whom give the edit suggest to remove the picture, no I want to keep it, otherwise the comment does not make sense then "It is not toxicity when someone acts offensively (using AI on SO) and refuses to learn when presented with more information about their actions." That is a good one! Commented Jan 15 at 5:51
  • 1
    @Rumi-SEMustFollowtheCC-SA I mean, did you read the AI response, and did you compare it with the reality? Because the AI response misrepresented reality. It seems "helpful" only until you follow it. I don't know what else I need to say here. Evidence suggests the AI response wasn't helpful and yet you're the second person trying to argue that the world has to conform change to conform to the AI response, rather than the other way around. Commented Jan 15 at 6:01
  • If you want the (irrelevant) LLM content (that I edited out pending approval), in review either reject it or accept it (to document it) & roll it back or accept it with changes to keep only my other changes. But it adds nothing. Comments are ephemeral & there's no particular reason to keep comments about old versions or have them make sense; they have merely become No Longer Needed. Anyway if you considered them still useful you could just comment about what used to be there. Post history is reached via the 'edited' button. PS The name of a suggesting editor is on the review page. Commented Jan 15 at 6:47
  • Re LLM output images: Why are images of text, code and mathematical expressions discouraged? More: Why should I not upload images of code/data/errors? Commented Jan 15 at 6:49
  • You need to use @ if you want someone to be notified if there is more than one commenter. Commented Jan 15 at 8:59
  • 2
    It is never appropriate to be rude, whether directly or sarcastically, just do not do it, even to respond to it. meta.stackexchange.com/help/conduct Commented Jan 15 at 9:01

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.