[dfabulich]

The graph proves the cause of their decline was AI, and not aggressive question moderation.

Ask yourself: in what year did it become difficult to ask questions on Stack Overflow? 2014? 2016? 2018? 2020? Aggressive question-closing was part of their design from the very beginning. Their high barriers to question-asking was the cause of their rise, as their primary user was never question writers: it was Google, and anonymous Google users. The whole thing was an SEO play from start to finish.

It's fun to imagine that their aggressive moderation was the "real" cause of their decline. It feels so gratifying, doesn't it? Finally those assholes got their comeuppance, because of their bad behavior!

But that's not why they failed. They failed because SEO businesses can't survive when AI answers the question directly, without referring you any traffic.

(The same thing is happening to Wikipedia, BTW, which is also aggressively moderated.)

[eterm]

You're right, the main decline was AI, but it was on a downward trajectory anyway.

This graph shows a distinct change pre-dating AI, starting 2014, there's explosive growth which suddenly stops around then.

A soft decline which carries on until Covid caused a temporary reversal of that.

The soft decline then continues at a pace around where it was, until November 2022, when it suddenly accelerates to its death. That's ChatGPT of course.

But the site was already in decline, against the backdrop of vastly increased software developers and software development, because of hostility.

Software developers used Stackoverflow despite the hostility, because there was no alternative.

The early growth wasn't caused by the moderation, because the early moderation was a lot softer.

[Izkata]

Yeah, this is much more accurate to what was actually happening. During that high period they struggled to get people to stop posting low-quality "do my homework" style questions - despite what people on here say the barrier for entry was extremely low and those made up the vast majority of what was posted.

I've maintained that if they handled this AI-caused decline well, they could return the site to its better days before the flood of people who didn't know what they were doing, offloading the bad questions while getting still getting all the good ones. I'm not sure they're even trying.

[pygy_]

> The same thing is happening to Wikipedia, BTW, which is also aggressively moderated

[citation needed]

Well here it is, and you're wrong: https://en.wikipedia.org/wiki/Wikipedia:Statistics

The article creation and edits curves are stable. The former growing at a slightly declining pace, which is expected since the amount of knowledge is finite. The latter is literally flat.

The monthly page views are in decline on mobile (from ~5.7 billions at the peak in 2024 to 4.5 billions currently). They are stable YoY on desktop at ~3.7 billions, and have been rising in the recent months.

StackOverflow is dead, the WP community is thriving, even if the page views have declined a bit.

SO had a moat because of its mass, but the place was a cesspool.

[Viliam1234]

> The article creation [...] growing at a slightly declining pace, which is expected since the amount of knowledge is finite.

Not just because of that, but they also made the process of creating new articles very hostile.

Decades ago, when Wikipedia started, it was possible to create a short article, and as long as no one objected against it, it stayed there, and people could later expand it. That's what "wiki" was supposed to mean.

Today, you need to create an article in a separate "Draft" namespace, then a random Wikipedia editor will judge it, and if it is not perfect (e.g. does not have enough references), it cannot be published. And if you don't fix it by a certain date, it gets automatically deleted. (Rather than leaving it there for someone else to hopefully improve.)

I tried to make an article for an author whose books my kids liked. His books were translated to many languages, he won a few awards, and now some company has bought movie rights for the books. Alas, I have failed to establish notability sufficiently in my short article, so it was rejected. The editor didn't even have to argue that the author is not notable (which would be silly, a google search clearly shows otherwise), only that my article failed to establish it. So the official policy now kinda says that it is better to have no article rather than an imperfect one.

Well, I am not getting paid for producing perfect articles for Wikipedia, which means there is no way for me to contribute anymore. Too bad, I have created a few articles long ago.

[pygy_]

I've had my share of battles on WP to preserve content that was dismissed as fringe by people who didn't understand what they were looking at (see e.g. https://en.wikipedia.org/wiki/Wikipedia:Files_for_deletion/2...).

The author you mentioned may not have met the inclusion criteria at the time, or maybe you had the misfortune to fall on a zealous admin who gets dopamine kicks from that kind of behavior. That being said, if a cursory Google search was enough to establish notability, it shouldn't have taken much time and effort to add links to reliable sources and save the article.

WP is very sensitive regarding inclusion criteria for biographies of living persons, because of the flood of vanity submission they have to filter out. The notability criteria are not perfect, but the provide a good enough balance to limit the spam (https://en.wikipedia.org/wiki/Wikipedia:Notability_(people)#...).

In my experience, toxic behavior isn't as rewarded as it was on SO, and there are venues to discuss issues.

[bombcar]

The old Wikipedia had more information on obscure Pokemon than major politicians, and you could add PowerThirst as an energy drink and not have it reverted for weeks.

There was something good and glorious about that, but that era is gone. Wikipedia and Stack Overflow are done, finished, cooked - what comes next?

[wikiiiisksklhll]

Quora had a lot more arrogant mods, but a much shorter peak when it was useful. Now it’s not even worth for AI models to train on its content, it’s even worse than expert-sex-change, the site that was so bad in incentivized SO’s creation.

I still remember when quora was launched and people claimed it was more welcoming for open-ended questions, when it was obvious from day one it was the opposite, extremely hostile to any newcomers. It actually more appeared to be mostly a community of assholes that was expelled from SO.

[wikiiiisksklhll]

Q&A sites in the age of LLMs are dead.

[pwdisswordfishq]

> if it is not perfect (e.g. does not have enough references)

Having references for claims is not "perfect"; it's basics. Which is especially becoming more relevant in the age of ever more rampant misinformation.

Take a look at the list of hoaxes below and wonder what would you have done to prevent them. With Wikipedia being open to editing by anybody, I don't think you could come up with anything better than the system currently in place.

https://en.wikipedia.org/wiki/Wikipedia:List_of_hoaxes_on_Wi...

[dfabulich]

https://diff.wikimedia.org/2025/10/17/new-user-trends-on-wik...

[ConorSheehan1]

And what was the alternative to stack overflow before AI?

People were desperate for an alternative, and ran with the first semi-viable one that emerged.

AI is arguably worse than SO for precision. It will hallucinate, and agree with you when you're wrong. But stack overflow is so unwelcoming to newcomers that people switched anyway.

[psd1]

> People were desperate for an alternative

Sure, why not. They were looking for a community that would let them speak. But that's like complaining your product review on amazon got deleted. It may be part of the amazon experience, but it's not the point.

[8n4vidtmkvmk]

90% of my over 250k rep is from asking questions.

Surely there's some value to asking a well worded question that many other people also have.

And yes, I've had my share of questions closed for some BS, including many in recent times closing 10 year old questions as dupes. Brilliant.

[throwaway2037]

    > The same thing is happening to Wikipedia

Can you explain? I am read-only there and it seems better than ever. Even in the era of LLMs, Wiki is still an awesome source.

[dfabulich]

Wikipedia's traffic (and donations) are collapsing. https://diff.wikimedia.org/2025/10/17/new-user-trends-on-wik...

People don't like to think of it this way, but Wikipedia is funded by ads. Except, the ads only advertise one thing: requests for donations. If people don't visit Wikipedia, because the AI regurgitated Wikipedia's answers, they won't see Wikipedia's donation ads, and they won't donate.

Over time, if current trends continue, more and more people won't even know about Wikipedia. They won't have any "warm feeling" towards the project, because they never go there, they never use it, they never see what good it does them.

[dannyw]

Doesn’t WMF hold hundreds of million and growing on their foundation balance sheet, and raise $150M+/year through donations when hosting expenses are $4M/year?

Obviously they need staff and more costs than just hosting, but something isn’t adding up for me, so I stopped donating.

[throwaway2037]

I feel exactly the same. I think I saw a blog post on HN with a deep analysis of their financial state of affairs. They don't need my money. They can run for a long time by (1) reducing unnecessary headcount (the org is way too big) and (2) carefully managing their endowment.

[bsder]

> Ask yourself: in what year did it become difficult to ask questions on Stack Overflow? 2014? 2016? 2018? 2020?

The year that Google made "open source contribution" a checkbox on your annual review. That's when almost ALL these types of sites went to dogshit. And then everybody followed Google which then weaponized it even more.

Once Google made Fake Internet Points(tm) worth actual money via your annual review, all holy hell broke loose over the sites.

[ButlerianJihad]

You’re actually on to something here, and it’s a phenomenon beyond Google that has engendered tons of blogspam and AIslop and manual LinkedIn slop.

Many professional certifications require bearers to earn CEUs. One way these may be earned is by blogging or doing demonstrations. So if you see a bunch of entry-level techbros doing really boring blog entries or posting to LinkedIn, you should know that they intend to earn CEUs for their particular professional certifications, such as CompTIA, et. al.

It’s not their fault... but Lord, how insufferable it can be!

[bombcar]

Some school somewhere made participation required or heavily encouraged - understandably perhaps; it was a great signal before it got eternal Septembered into worthlessness.

[casey2]

It's the same thing. Why did AI compete so aggressively with them? It's because their system was one that produced confident misinformation from hobbyists while gatekeeping actual experts.

The same thing is not yet happening to wikipedia as you can see with the pageview tool. You may be confusing a covid bump. At most any drop is within an order of magnitude.

[firesteelrain]

You aren’t downvoted on Wikipedia. And people have been complaining about heavy handed editors and mods there for a long time. Same with SO.

It’s pretty much a meme now.

[dannyw]

Wikipedia at least has a culture where (most of the time) if you’re objectively rude or mean, especially to newbies, you’re at least shunned a bit.

Strict moderation etc isn’t a bad thing, but the environment and culture you mould is what matters.

[marcosdumay]

Ouch, great, you should show that protocol you used to get causation out of an observational study to the entire scientific community! It's unclear what area will give you a Nobel Prize for that, but several of them will rush to do it first.