The number of questions has decreased 10-fold: where did users go and what can be done to prevent that?

Well, the 'solution' varies on whom you ask. To me, the big issue is over time we lost our 'core'. We've had various incidents over time that caused core users to leave, and we lost the momentum we had early on.

SE's solution over time has been to try to 'change' parts of the core platform. I'm not a fan of some of these changes, but I vaguely understand where it's coming from. My worry is the attempt to make it more 'familiar' to people used to other plaforms may erode its unique selling points.

SE has been looking hard outside for solutions, when to a certain extent to solution is here. We need new users but an engaged, happy core base is essential for keeping them there.

More recently, there have been attempts are re-building ties with current users and refreshing aspects of the current platform. We've not yet seen the fruits of this but I'm guardedly optimistic.

To me this is an essential part of the problem - SE has focused a little heavily on trying to change the platform for people who aren't part of it, but sometimes a little too little (in my view) on the people who are on it.

To me, I feel like we need to look a little to the past and find the ways and resources to not just 'try to tweak the system for newcomers' but actively attract new users. Old SE had a few advantages - Jeff and Joel had their fanbases and that fed into SE and it was new and exciting. That's a river we can't cross again (or a badger we can't lick twice). We don't just need to get people, we need to get people who 'enjoy' the network as it is.

To me, there are three parallel paths we need to restore in the network.

1. Better community/company relations - this is probably a longer term thing. We need to think 'outside' new user growth. We need to think about retention and re-attracting old users. We also need to ensure that we have less 'dramatic' events initiated by the company, and if dramatic events happen anyway, to be able to mitigate the damage.

   We need to build trust on both sides, bridges and a better understanding of how users currently on the network will react to various initiatives

2. We need to market SE for what it is. I'd say I'd like to see the same energy put into marketing Teams and legacy careers into marketing Q&A - but really I'd like to put in more directed energy into it. Unlike the commercial products, we're not selling to new users - we're trying to convince current users to stay, and for new users to buy into our model.

3. Active recruitment of new users, and getting our current/legacy userbase invested in it. I've advocated in the past for a revival of team chaos in some form but I don't think on the very short term, we have the resources for it. User growth isn't a design problem to me - it's a effort/direction problem.

This question erroneously assumes that it is worthy to "prevent" this apparent decline. You point to data that suggests that the question volume used to be 10× higher. But you do not establish that it should be 10× higher.

No site has any intrinsic reason to be — let alone to have a particular level of engagement.

Knowing where would-be askers are going is of little importance. And the proposition that there are even would-be askers out there is an assumption.

I use this platform as long as it makes my days better. When it doesn't, I won't. And if that time comes, I'm not going to complain to Stack Exchange or this meta community to change anything to keep me.

I think it's a natural consequence of not allowing duplicate questions: the "repository of information" fills up.

With questions having accumulated over decades, effectively all the "good questions" have been asked many times over already. So new questions tend to be highly specific, of narrow interest, and unable to generate pearls nor attract experts.

I recall this happening on a short time-scale for Atheism.SE; they just ran out of questions that the then-community cared about:

Yet sometimes a subject just isn't dynamic enough to facilitate a boundless supply of questions. --- Robert Cartaino

Is this major drop in question rate an existential crisis, or should we consider it "mission accomplished"? I'm not sure. Either way, I expect the peak question rate was unsustainable.

I think user volume is a problem for the company, not MSE, to tackle. Both because the company has more user data, analytics, and capital to do things like commission surveys, which we don't have. But secondly because fundamentally the goal of MSE should be to be a voice for current users that the company can use as a reference, not to act as the company's board of directors.

While I understand wanting to try and solve that problem ourselves, I don't think we can and I don't think it's particularly productive to try.

You argue about the number of users and present data about the number of questions. That is not the same. Is there a proportionality between the two? If so, is it constant over time.

Okay, I agree that a ten times decrease probably means that much less users ask new questions. But what about the users that only read questions and do not ask questions, because they don't need to, because somebody else already asked them. Have they decreased too? In the same way, i.e. also tenfold decrease? I don't think so.

Sorry, I do not have ways to tell you where the users went, neither cross-site tracking nor surveys. Maybe the company has (or Google), but this knowledge might not be publicly available.

I agree with your possible options. I think they all make sense and likely it's a mixture of them all. However, the exact proportions might remain unknown.

It's partly AI because there is an increase since 2022, but it's not only AI because the demise started much earlier (2018). And AI should not forget that it's based on human generated content, not the other way around. AI may cut off the basis it's standing on.

It's surely some saturation because you can't ask the same questions multiple times (without us noticing). If you look at the most often visited questions they are probably quite old. Low hanging fruits and all that. But Wikipedia should have the same effect and Wikipedia is still growing, while for us the slowdown seems stronger. My gut feeling would be that we haven't yet finished the mission, that the available knowledge is much bigger than what is contained in the data dumps and there are millions and millions of questions that could and should still be asked. Only we have not many people currently to ask them.

It might be other sites (Reddit, GitHub, Codidact, ..) but also only to some extent. Younger colleagues of me seem to not be so keen on SE anymore, they seem to... like to go back to forums, where you get help without voting or closing but also without the quality content from here. It's just anecdotal, so might not be significant and I don't understand it either. I would prefer SE any day.

Now what we could do about it? How much needs to be done about it? If for example the mission is accomplished, nothing would need to be done. I don't think it is, so I would like to SE to continue.

And I think SE is on a good way with improved chat, articles, discussions, comment threads, games, staging ground, ... The real problem is that it's 10 years too late. Any new feature will only be used by so many fewer people, it's hardly worth developing them. 10 years earlier, it might have been a game changer, but the past success made everyone complacent for too long. That often happens. SE might simply not be the best anymore.

I don't think much more can or should be done. Already so much is done. Let's see how the integration of new features looks when it is finished. Apart from this I would only recommend a bit more friendliness. Obviously we need to court question askers more with the increased competition, if we want to keep them. That might feel difficult for some, but I would sell it as a cultural thing: be nice, assume good intent, forgive mistakes, collaborate to polish content together and gradually. A bit like when a new site starts. And that's what we would do: restart SE with a better experience for everyone.

For me personally: it's a mixture of most of the time I find the solution here already, some of the time AI helps me and solves my problem and in the remaining cases I know how to ask a question here but I had many years training. Others may not. No idea what they do instead. Maybe they don't have a good place to ask currently?

This ("the number of questions has decreased 10-fold over the entire Stack Exchange network") might indeed be a problem as you pointed out. But it may or may not reflect the overall health of Stack Exchange.

I suggest to measure instead the health of Stack Exchange by the number of visitors who have gone away with something useful. For example, I would be looking at the increase in either of these across time:

• number of visitors
• {number of visitors} * {time spent on the Stack Exchange sites}
• number of visitors who copied something from the Stack Exchange sites^[1]
• sum of Google page ranks across all Stack Exchange pages

^[1] Relevant to some, but not all, Stack Exchange sites, for example for Stack Overflow, but not for Physics.SE (thanks to PM 2Ring for the comment).

A broken voting and reputation system

One issue that contributes to a decline might be that the voting and reputation system doesn't work well in an ageing database of questions and answers:

Is the voting and reputation system sustainable? How can we improve it or maybe it should be replaced?

New questions and answers arrive in large ocean of older questions which have already acquired a lot of votes.

This makes the incentive from the kudos-based system less important and it is less attractive for newcomers to collaborate on the website.

It is not just a decline in new questions

The decline is probably not just in the rate of new questions. There is also a decline in the interaction with older questions/posts. We can't see this in terms of views (that's not in the public database), but at least in terms of votes we can see this: the total number of votes that is being given to posts of more than one week old has almost disappeared completely.

The majority of voting is from the interactions when questions are still fresh and appear on the homepage.

Older questions

Similar graphs like these show that older posts get relatively more of these few late votes.

For example the graph below. Posts from 2008 get more late votes (and their rates even increase like fine wine). But posts from 2014/2016 get fewer late votes and have a decline as they age.

This is partly because the older post lived in days with more activity and less competition from other posts.

But also if we compare them at the same time, then old posts acquire more of the late votes.

This can be due to a bias (people searching will look more often to posts with already many votes). But it can also be that older posts are simply better because it is the low hanging fruit that nowadays cannot be asked any more.

This repeats the idea from the start of the post, that it has become less interesting to make additional posts/contributions because the database of questions and answers is already very full.

My previous answer was deleted for bringing in personal issues, but here is another answer of mine, concerning a different perspective.

I feel that many Stack Exchange sites fell into a bad cycle of topics with questions and answers repeating certain themes, without expanding beyond.

For instance, in Retrocomputing.SE people discuss only Western retrocomputing. Questions asking anything about Soviet retrocomputing will remain unanswered, and if you answer a question from a wider perspective, you will get criticized.

Existing topics promote reproduction of questions and answers on that topic. People reading a question or answer tend to ask tangential questions, etc.

I have noticed this on Math.SE: after making several answers on hypercomplex numbers, I've spotted more questions, related to my answers.

That said, some sites got cocooned in their usual topics and anything transgressing them is frowned upon. For instance, on Philosophy.SE one of my questions on solipsism was deleted on the basis of solipsism not being philosophy. In Mathoverflow.net the topics of math analysis are not welcome, and things related to set theory and topology are preferred.