Hacker News

points by kstrauser 1 day ago | 0 comments

OK, let's get this out of the way:

1. SO question numbers dropped precipitously before AI came along.

2. AI clearly finished the job.

I took the graph from this very page and annotated it with the public launch of ChatGPT in November 2022.[0] Yes, the decline hastened after that, inarguably. But also, it was already nearly 50% down from its June 2020 local maxima. This is also inarguable.

SO had been flatlined since 2014 and declining since 2017. It dropped off a cliff when, 2.5 years after its 2020 bounce, ChatGPT came along with the death blow. Even then, it looks like 4 months of ChatGPT speed-ran about a year's worth of traffic decline, and then the slope went back to approximately the pre-AI angle.

I contend, and I'm backed by facts, that AI is not what started the avalanche. It killed any slim chance of a comeback, but that wasn't the beginning of the end. The end was already well begun by then.

[0] https://imgur.com/a/DN5N90L

[–] antman 20 hours ago

SO was a governance disaster. It was the infighting it was the strike, it was bad auto moderation tools. And the Stack Exchange strike wasn't really about AI. AI was just the final trigger after years of management undermining the community that made the platform valuable.

Stack Exchange was built on an implicit social contract people contributed their expertise and thousands of hours of unpaid work because they believed they were building a collectively governed knowledge resource. The company owned the platform, but the community created its value.

After time, management became increasingly centralized and disconnected from contributors. Also the policing of questions based on geeky criteria and not what users wanted. The increasingly aggressive policing of questions, closing perfectly reasonable discussions, and incorrectly saying questions are duplicates made the platform unnecessarily hostile to everyone, qmd at least thats why i quit. Quality control gradually became useless gatekeeping.

So there were really more than one problems, corporate management treating contributors as free labor, and established contributors treating their moderation as authority over everyone else, there were bad tools closing questions. The platform had become extremely hard to use for posting questions, finding amswers, moderator work. For me it was useless. Incidentally the closing of “opinion” questions was the worst decision since it was the only thing that could make it distinct from AI produced answers.

For some of the events there is paper here with details https://arxiv.org/html/2512.08884v1

[–] nottorp 17 hours ago

> The increasingly aggressive policing of questions, closing perfectly reasonable discussions, and incorrectly saying questions are duplicates made the platform unnecessarily hostile to everyone, qmd at least thats why i quit.

This. Way before LLMs.

[–] matsemann 17 hours ago

> Also the policing of questions based on geeky criteria and not what users wanted.

This is wrong. People don't get that SO is moderated almost entirely by users. Users are those that down vote. Users are those who vote to close a question, and when a question reaches consensus it's closed. Users are those watching the review queue. Mods mainly did bigger stuff with additional tools.

So users wanted it how it was. And for good reason. If you actually participated on the site, like watching new questions arriving in a constant stream, it was obvious that most of it was low quality. People here will bemoan they got a question closed or no help, but that's to me mostly entitlement. The rules are clear, it was to create a Q&A site, not solve your specific question for free. To me that's not hostile.

[–] rpdillon 16 hours ago

> This is wrong.

This is exactly what drove me away. It's basically Pournelle's iron law of bureaucracy. SO got so overrun with people that were dedicated to Maintaining Order that the site itself stopped being useful. I feel like at least half the time I would go visit, the question that brought me there had been marked as a duplicate or closed. Pointing out that the fire hose is mostly low quality doesn't change this.

All those people fighting to keep SO pristine now preside over a kingdom of nothing.

[–] matsemann 15 hours ago

Why did it matter if the question you visited was closed as a duplicate? Still you would find your answer by following it. People wrongly think a duplicate was something wrong. They probably could've made it clearer.

[–] Sohcahtoa82 14 hours ago

Because as literally dozens of people have already pointed out, it often wasn't actually a duplicate.

"How do I do X in Python?"

"This has already been answered in <other question>. Closing as duplicate."

And then you look up that other question and it's "How do I do X in PHP?"

Ask how to prevent XSS, and someone will flag it as a duplicate of a question asking how to prevent SQL Injection. Ask how to encode something in Base64, someone will flag it as a duplicate of a question about how to encrypt something with AES.

And the worst part, is that these moron moderators pat themselves on the back for keeping SO clean.

[–] rerdavies 15 hours ago

It was trivially easy to close dup a question that looked like it was a dup from the title, but much harder to actually read the contents of the questions, and realize that they were not. And you got points even if you were completely wrong. SO would have been much better if there were no close dups at all.

And even if the questions were actually duplicates, far too often the answers would be so out of date that it would be better if the question (and the answers) were allowed to start again from scratch.

And I agree: the "opinion" close was toxic as well.

A tragic gamification failure, that contributed significantly to my drift away from SO.

[–] shagie 14 hours ago

> And you got points even if you were completely wrong.

One of the oft repeated ideas is that you got points for closing questions.

You don't. The only ways to get rep on stack overflow is described at https://stackoverflow.com/help/whats-reputation

    You gain reputation when:
    
    question is voted up: +10
    answer is voted up: +10
    article is voted up: +10
    answer is marked “accepted”: +15 (+2 to acceptor)
    suggested edit is accepted: +2 (up to +1000 total per user)
    bounty awarded to your answer: + full bounty amount
    one of your answers is awarded a bounty automatically: + half of the bounty amount (see more details about how bounties work)
    site association bonus: +100 on each site (awarded a maximum of one time per site)

You do not get any points for closing a question.

[–] kstrauser 13 hours ago

That's not quite right. Deletionists campaigned to be elected as moderators based on the number of questions they closed. For example: https://stackoverflow.com/election/14

They did not directly increase the integer counter of their reputation score. They did use it to explain why they should be granted increased human reputation as a moderator.

[–] ryathal 12 hours ago

When they added review queues there were badges tied to closing questions. The queues made a lot of existing problems a lot more public and prominent.

[–] jodrellblank 14 hours ago

> "SO would have been much better if there were no close dups at all."

Do you spend much time on Quora? Or AskReddit? Do you like that they are endlessly 'the same' questions over and over and over?

[–] OroPla 11 hours ago

How would you even notice without being chronically online? I visit these places and see users complaining about questions repeating, but never notice myself, since I don't read that kinda stuff every day.

[–] chuckadams 15 hours ago

"We're not answering your question because it has words another question has, you're wrong for saying it doesn't answer your question, and you're wrong for thinking we were wrong."

That's a bit clearer. Bit long to engrave on a tombstone though, but SO will remain unburied because after all, most coffins are pretty similar.

[–] Bratmon 14 hours ago

The percentage of questions closed as duplicates that actually

1. Were actually duplicates of existing questions

2. That had a correct, up to date answer

was definitely less than 1%

[–] braiamp 13 hours ago

That's a lie, you know why? Because I've actually counted it, it's less than 20% wrong. And it's mostly because a rule of thumb got grossly misapplied: if you want to know if two questions are duplicate, look at the answers; if they are mostly the same then they likely to be duplicate.

It got grossly misapplied because it became "look at any answer, if one of the answers it the same, then they are duplicate". Note how likeness became overlap and likelyhood became certainty. Two questions where restart might solve the problem became duplicates under that absurd policy. Your example on the other hand I didn't find. I didn't find anyone claiming that two questions are duplicate of one another in different programming languages unless the language is not relevant.

[–] Bratmon 13 hours ago

> Your example on the other hand I didn't find. I didn't find anyone claiming that two questions are duplicate of one another in different programming languages unless the language is not relevant.

Did... Did you just mentally mark my comment as a duplicate question and copy/paste an unrelated answer?

[–] inetknght 15 hours ago

> People wrongly think a duplicate was something wrong.

No. A duplicate is fine.

What was wrong was when "duplicate" questions were closed in favor of other more-upvoted questions with less-correct answers, or questions that were marked duplicate which were in fact not duplicate.

[–] rerdavies 15 hours ago

Combined with the problem that more correct questions, that were likely to produce more correct answers were closed for further comments when they were incorrectly close-dup-ed. So often you would find the correct answer in the brief window between when the question was asked, and when a member of the self-appointed moderator class had incorrectly close-dup-ed the question. So frustrating!

[–] jodrellblank 13 hours ago

> "a member of the self-appointed moderator class"

It isn't self-appointed, everyone is a 'moderator'. And 'everyone' could vote to re-open closed questions.

[–] ezconnect 14 hours ago

The problem of SO is the duplicate question is years apart. The answer is now different because software version changes along with the APIs. Cancer killed SO, the aggressive policing and killing of noobs who want to contribute.

[–] johannes1234321 12 hours ago

> So users wanted it how it was.

A group of active users wanted it that way. Some other group of users didn't necessarily understand the power mechanisms, others would would have liked to be active users were driven away.

It's however tough to know what a quiet diverse majority would have liked.

[–] matsemann 11 hours ago

What matters is however the active users answering questions.

[–] johannes1234321 8 hours ago

Well, if they drive away interesting questions or interesting answers that matters, too.

The problem is that measuring the unknown is tough.

[–] rerdavies 15 hours ago

I think you also need to distinguish between actual users and members of the gamified moderator class: an entire class of SO users whose primary involvement was gratuitously closing questions so that they could get more points, or picking up cheap points by re-formatting the one or two variables you missed in your answer. Often with no real evidence that they had anything but very modest technical skills. Those people. If you don't want to call them moderators, so be it. But I'm not sure what you would call them if not "moderator", whether they had some magical shield that nobody paid attention to or not.

Easy to spot. Just spend some time browsing through the activity of super users with astronomical point scores. It's pretty obvious that they were doing what they were doing to get points, rather than to make the site better or easier to use, many of whom had quite unimpressive technical skills.

The obvious toxic effects of gamification was a substantial contributor to my personal drift away from SO. But what really killed my interest: the toxic effects of close-dup, which produced rapidly accelerating content rot. It is true that, at some point, they started applying penalties to old answers; but that seemed to have no real effect on the toxic effects of "close-dup".

e.g. Questions raised against Android 14 being close-dup-ed against questions that were originally raised against Android 4.0, where the top thee answers, each of which had 1400 upvotes, was that "you can't do that", even though you absolutely could on Android 14.

That and the fact that I had discovered (as I think many others had) that you actually got much better answers if you appended "reddit" to the start of you google search queries, and skipped over the SO answers altogether.

My final point score on SO: somewhere around 8,000 I think. So not a casual user. I'm pretty sure that about 3,000 of those points were given for an answer that applied to Android 4.0, that stopped being accurate in Android 9.0, and I (shamefully) never got around to updating for modern Android tools.

[–] matsemann 11 hours ago

You have multiple comments here in my thread where you wrongly describe how SO works. It's quite meaningless to discuss with you when you lie about points being awarded for stuff it was not etc.

[–] benibela 11 hours ago

but now the same moderators are on reddit

[–] g8oz 16 hours ago

There are real differences and experiences between heavy users and occasional users, voters and moderators, question posters and contributors. Unless you're careful it's very easy to develop a culture of gatekeeping that is off putting to a broader community. SO wasn't able to successfully steer between the sometimes competing interests of these groups.

[–] ghaff 19 hours ago

You've described many things over the past few years. COVID is also one of them. Stuff that was just basically holding on anyway based on history and momentum just couldn't handle an additional body blow. (And, in a number of cases, didn't really care to.)

[–] JensenTorp 15 hours ago

The toxicity (we know best) was there from the start when they refused to let you create a local account and forced OpenID.

[–] rerdavies 15 hours ago

My personal hate-on for SO: forcing all answers, INCLUDING the code to be licensed under an awkward CC license. Yes, you were "allowed" to dual-license the code in your answer under a more permissive license that doesn't require attribution.

I tried adding an explicit dual license to the code in my answers; but moderators would any answer that tried to do that.

Not so much that I care about the copyright of my code; but more a question of whether I would find a lawsuit from SO on my doorstep because I had neglected to credit SO for three lines of code that somebody ELSE had contributed to SO. A very real problem, given that SO management after the takeover had shown itself to be not really trustable.

[–] LtWorf 19 hours ago

For me the main problem was that the level was so incredibly low.

Already when I was writing my master thesis I could get very little out of it.

All of my questions were too specific and went unanswered.

My most voted answer explains how to install header files on debian.

[–] justauserid 1 day ago

I had stopped going to it quite some time ago.

1. It was hard to ask any question. Sometimes it felt if a few characters match it will be closed off as dupe. Besides no thought was given to the fact that same question might change entirely from pov of the answers for a platform or SDK across, say, a decade (?).

2. I also realised reading the documentation (that is if available) and looking though github and people's blogs helped better.

3. But bigger takeway was that due to relative uselessnes (if not downright hostility) of SO I started just breaking down the problem and solving on my own, talk to colleagues, or look for solutions for broken down problems and then I think I used to come across SO in months and that too most of the time when a colleague will refer to something.

This was before LLMs or COVID.

[–] 10111001 23 hours ago

> 1. It was hard to ask any question. Sometimes it felt if a few characters match it will be closed off as dupe. Besides no thought was given to the fact that same question might change entirely from pov of the answers for a platform or SDK across, say, a decade (?).

> ...

> 3. But bigger takeway was that due to relative uselessnes (if not downright hostility) of SO I started just breaking down the problem and solving on my own, talk to colleagues, or look for solutions for broken down problems and then I think I used to come across SO in months and that too most of the time when a colleague will refer to something.

Yeah, SO's culture meant that it was always my absolute last resort to ask a question there. And whenever I did I had to be super-defensive from the get go. It was either people trying their hardest to close it, or "redefine" a hard question into an easier one they could answer (e.g. ignore the constraints that were key to my question).

I understand a lot of that came from their desire to create a curated list of questions, but in retrospect they took the wrong approach. They should have let people ask and answer questions freely, even dups. Instead of hunting for dups to close, they should have had their moderators link them together and author the "canonical question" from them.

[–] somenameforme 22 hours ago

Yip, LLMs were the executioners but SO had already long since put itself on death row for exactly what you're mentioning. I think sites like FlipCode were in many ways the spiritual predecessor of SO, and the questions were more of a part of the community rather than the definition of it. People need to feel free to communicate for something like SO to be successful, and they'd turned themselves into an absurdly user-hostile site where it seemed like people were competing against each other to see who could find a reason to close questions, rather than looking for a reason to answer them.

The irony is they were constantly looking for reasons to mark a question as low quality, but the moderation efforts were themselves extremely low quality, like absurdly so. Things would be marked as dupes when they had no relationship whatsoever, completely inappropriate questions would be asked antagonistically like 'provide a minimal replicable code example' for an abstract/theoretic question, and stuff that stood a chance of fostering a healthy discussion was just randomly closed.

It's perhaps a good example of the dangers of using volunteer moderators. It just never works out, because doing a good job of moderating is a thankless, time consuming, and painful job. So you disproportionately end up with people doing it for power over others or some perception of status seeking. In either case, these are the last sort of people you want having any authority whatsoever over others.

[–] vintermann 21 hours ago

A big part of the picture was that they'd gamified the heck out of it. Community moderation can work, somewhat, but I've never seen gamified moderation work. Gamification is the big killer of collaborative communities.

[–] ChrisMarshallNY 20 hours ago

I think you got it in one.

The one thing that always struck me, was that most (not some) of the high-score individuals had 1-digit question counts (with that digit frequently being "0"). It was clear that asking questions was a "demerit." These were folks judging questions, that had no idea how to ask a question, themselves.

Also, I found that many were not ones that worked on shipping software. It seemed that their vocation was SO. That's very much like the kid that spends all his time, playing Counterstrike, and giving advice to soldiers in theater.

I'm really big on apprenticeships. We become masters, by moving through being juniors. It gives us an extremely solid base, and elicits acknowledgment from others.

Journeymen can often be demanding and irascible, but they command respect, by demonstrating that they have been in your shoes, and that's really important. You suffer the slings and arrows, because you know that there's a clear path to where they are at.

I have friends that are former US Marines, and they all have Boot Camp stories. These stories all feature downright abusive Drill Sergeants. But they also all profess immense respect for these dudes, because they came from years of experience, and they often showed boots how to do something, by demonstrating it, themselves, at a high degree of skill.

> "Critics are like eunuchs in a harem; they know how it's done, they've seen it done every day, but they're unable to do it themselves."

-Brendan Behan

[–] rerdavies 14 hours ago

> We become masters, by moving through being juniors.

Yes this! I found it painful to watch abuse being heaped upon obviously junior programmers. Points should have beem award for mentoring those on the path, not gratuitously rejecting their questions.

[–] tsss 20 hours ago

Community moderation at that scale just doesn't work, period. It always attracts the worst office chair nazis who enjoy lording their petty power over other users.

[–] phatskat 22 hours ago

As someone who spent a lot of time answering and moderating (not full-on moderating, but editing, voting, reviewing first questions, etc) in the WordPress SE, it was jarring sometimes to see the attitudes that the real mods brought in.

However, when you spend an hour looking at low-quality answers, questions that have he's been asked like ten times already, etc, I get the sense that the people doing the moderating were trying their best - some of them, at least. No doubt there are mods there who just enjoy the miniature lower trip a la overzealous reddit mods, and I can't imagine the stress of trying to handle the amount of traffic some of the exchanges got.

[–] greengreengrass 18 hours ago

> "redefine" a hard question into an easier one they could answer (e.g. ignore the constraints that were key to my question)

> I understand a lot of that came from their desire to create a curated list of questions

I think both of these are true, but I'm not sure that many of the contributors (or even many of the moderators?) _really_ cared too much about building the "curated list of questions".

I'd posit that part of the problem was also its competitiveness – contributors wanted to earn rep, so they ignored constraints or answered the question they really wanted to answer. Some of the tags I used to frequent had a few contributors who seemed to do nothing but answer SO questions, so there were never any good questions left for a mere mortal with other commitments in their day.

I am generally unconvinced that reputation-based systems like SO's are actually useful. Contrast SO with a project mailing list or issue tracker – where I've had some of the best support – where there's nobody tracking rep or keeping a tally, and everyone is there to simply help and share their knowledge.

> Instead of hunting for dups to close, they should have had their moderators link them together and author the "canonical question" from them.

I did plenty of free moderation over time on the review queues. I have no idea today why I invested so much time in that, for it was a thankless task and ultimately didn't feel I could move the needle on the bigger problems.

[–] alex7o 22 hours ago

Almost feels like training an llm on that caused it to ignore certain constraints that make a problem hard and solve an easier version of you problem instead.

Also a lot of communities do their own QA on discord where they can modererate themselves.

[–] kamaal 22 hours ago

>>Yeah, SO's culture meant that it was always my absolute last resort to ask a question there.

Have seen this every where, even in case of people. If people are dreading to even approach you for something, its over for you. You can't get just the good things to pass through a filter. No such thing exists.

In many ways you need to be open for everything(apart from outright wrong things, like abuse) including stupid things, to get the good things.

[–] bityard 17 hours ago

On most forums, ending up with power-tripping asshat moderators is purely accidental.

In reading the site rules and meta discussions, SO always seemed to have been explicitly designed to invite, cater to, and directly nurture them.

[–] mort96 21 hours ago

My biggest issue with SO was that all the answers were outdated. In the beginning, if you found an answer on SO it'd be still relevant because SO was relatively new. But then as it got older, so did the answers to questions.

The ridiculously strict "no duplicates" policy guaranteed that the old, increasingly outdated answers would remain the only answers forever. In retrospect, I believe a site like SO wants to encourage some rate of duplicates, to ensure that answers remain fresh.

Just yesterday I looked something up on SO and the answer was 14 years old. Would've been a pretty relevant answer back when I first started using SO, but not now.

They really killed themselves through strict moderation. It's incredibly sad, it was a good resource while it lasted.

[–] ceejayoz 19 hours ago

I think SO really needed a "vote to unaccept / mark outdated this answer" or proper supercession process.

[–] rcxdude 18 hours ago

I think the intended approach was basically to just edit the top answer to update it. One of SO's issues I think was that it couldn't really decide if it wanted to be a Q/A site or a wiki and so subverted the expectations of both.

[–] mort96 18 hours ago

I think you're right that that was their expectation/intention. I think it failed because experts don't just randomly go through old questions and make modifications. What prompts knowledgeable people to share their knowledge is new questions asked by new people.

[–] ryathal 12 hours ago

I think the problem is that the creators, and by extension the moderators were bought into the idea of a giant wiki of all programming knowledge. The vast majority of the users just wanted a Q&A site. This led to the infighting and lots of counter intuitive rules as the gamification part conflicted with the goals of being a wiki.

[–] kstrauser 16 hours ago

That approach was untenable because then it still attributed the answer to the original author, even when the content was updated. That could cause real-world issues. “Hey, why did you say that Python is untyped on SO? I thought you knew something about it.” “I never said any such thing.” “Here’s your answer where you say it!” “WTF.” And sure enough, there’s some dumb answer with crappy grammar and mistakes, still attributed to you.

That was a terrible, awful plan, and I always reverted edits to my answers that went beyond fixing typos or updating old links.

[–] ourmandave 18 hours ago

Some sort of version number would help. I use .NET and if the most current answer is on a question asked 15 years ago, I'll never see it.

[–] lloda2 20 hours ago

To me, the way the site was organized was absurd. It was difficult to answer questions.

The one time I tried, I couldn't answer directly because I had no reputation, and I couldn't comment either, but I could edit the question (!) which someone else had already done, answering the question in the question. It was an incomplete answer, so I added another edit with a complete answer. Then the mod who approved my edit also added a newline to it, so it was them and not me who was credited (unless you went to the 'history' page).

I gave up after that.

[–] bulbar 1 day ago

> 1. It was hard to ask any question.

Totally. There were so many safeguards in place, for a newcomer it is quite hard to contribute.

They also never understood a core principle of such platforms. The noise is not noise but what carries such platforms. They even had it quantified, some of their most upvoted questions are off topic because they are opinion based. Bad and good questions (and answers) come hand in hand.

Well, they were the backbone of professional IT/SWE, so of course they did a lot of things right obviously.

[–] ceejayoz 19 hours ago

> They also never understood a core principle of such platforms. The noise is not noise but what carries such platforms.

But it's also what wrecks such platforms.

SO didn't arise in a vacuum; it was explicitly a reaction to how shitty ExpertSexchange had gotten by then with an anything-goes moderation policy.

[–] kortex 13 hours ago

The solution was the one they arrived at way too late: have a "starter zone" for entry level stuff.

[–] eithed 21 hours ago

The most hilarious moment for me was when my question that was 5y old was closed as a duplicate of one that was couple of months old. I provided an answer on how to work around the issue; the new question didn't have an answer, but it had more points. From what I read the original issue was still there. At that point I was rarely visiting the website either way, but the absurdity of the situation was staggering.

[–] matsemann 17 hours ago

It's only absurd if you think the question being closed as a duplicate was calling you out for doing something wrong. This was how it intentionally worked: group everything, and provide a canonical answer. Duplicates were good, the idea being that more ways to ask the same question could lead more people to find the answer. The other question just happened to have more people find it (hence more points), but someone finding yours could then be redirected to the more popular variation of the question.

[–] rpdillon 16 hours ago

He just explained that that question didn't have an answer. Redirecting people to a question with no answer isn't helpful.

[–] tacone 21 hours ago

I think they tried too hard to structure things and keep everything clean, which becomes difficult or even outright deleterious when the numbers are large.

They should have accepted a certain level of mess instead of being so strict. After all, even good popular answers become outdated sooner or later.

[–] xtracto 1 day ago

I hated asking Qs in SO, it felt terribly snobby. However, I used it a lot to search for answers. The alternatives were crap.

First, there was some DevX? or similar site (either Indian or used by Indians a lot), the quality was pretty bad. At some point medium had good SnR , but then it got enshittified. Then a bunch of SEO crap sites came in (geeks4geeks and whatnot) and it was the end of it. Only S.O maintained certain quality, for existing info. Asking was still pretty terrible.

The "web" [http] is currently in a deplorable state, ads are horrible and rampant, the majority of content is pure SEO crap [now AI generated]. So LLMs are a breath of air to get answers.

Ive said it before but, the internet is really dead, for a lot of different reasons [death internet theory and public services automated hacking is the nail in the coffin].

[–] lexandstuff 21 hours ago

I remember Experts Exchange somehow managed to have every question one could ever ask, but no actual answers - even with a paid account.

I was glad when S.O. put an end to them.

[–] pnut 19 hours ago

good ol' expert sex change

[–] zahlman 1 day ago

> The alternatives were crap. … Asking was still pretty terrible.

The alternates were crap precisely because they didn't have the restrictions on posting new questions that everyone complains about. Asking was terrible because otherwise you get the other forms of crap. That's what everyone knew going in and it was a common observation that specifically motivated creating the site in the first place.

[–] rerdavies 14 hours ago

Ironically, in my experience, the alternatives were crap because the people giving the answers were point seekers. Microsoft Tech Support forums come to mind, particularly. No problem with the questions. But significant concerns about the skills of the people giving the answers. Not sure why that is, and why SO managed to get something right when it came to getting quality answers. For a while, anyway.

[–] DuncanCoffee 21 hours ago

> It was hard to ask any question. Sometimes it felt if a few characters match it will be closed off as dupe. Besides no thought was given to the fact that same question might change entirely from pov of the answers for a platform or SDK across, say, a decade (?).

I used to end the questions with "It's not a duplicate of AAA because BBB is different. It's also not a duplicate of CCC because DDD". It DID force you to check for already answered questions, but the automated hints while you typed was often good enough to spot duplicates.

In the end, imagine wasting half an hour to post a four-lines issue... No thanks.

[–] lovich 21 hours ago

I did the same and would still have a question closed as a dupe, with a link to the question I had already explained was different from my situation.

[–] matsemann 17 hours ago

Do you have a link?

[–] lovich 9 hours ago

No, I used my real name for stack overflowing and not willing to link it to this account

[–] rerdavies 14 hours ago

I've seen that too. Not that common, but it certainly happened more than it should. You get a point for close-dup-ing whether you're right or not.

[–] cwnyth 12 hours ago

You do not get points for closing questions, and you get negative points for downvoting. Where did that lie come from?

[–] thephyber 20 hours ago

I always found that the SO search was frustratingly bad and users NEED to use search to find the authoritative question for your needs.

And even if I find the right existing question, the accepted answer might not be the one I need. Sometimes the question asked accepts the wrong question. Sometimes the current right answer didn't exist at the time of the answer acceptance.

SO was a good first approximation of a good Q&A site. It didn't evolve with the times.

[–] NetOpWibby 22 hours ago

+1 to the outright hostility.

Good riddance.

[–] locknitpicker 1 day ago

> It was hard to ask any question. Sometimes it felt if a few characters match it will be closed off as dupe.

I feel this is the critical bit. People are quick to dismiss this problem as a moderation problem, but I don't agree. I believe it was a query problem, and the absolute lack of support for tying both questions and answers to specific software versions, or range of software versions. So you'd see answers from ages ago marked as the right answer to a question, even though it was using some deprecated version.

Questions on how to address the same issue when new versions were out where closed as duplicates because literally the question was asked before and answered as well. But the answer nowadays would not be applicable.

Then to sidestep this problem you see the phenomenon where ancient questions end up having multiple answers, with newer answers even out voting the accepted answer. And even those would be deprecated as well.

So SO was designed as a system where a single question can only possibly have a single accepted answer. Ever. And anyone asking similar questions would by design be routed to he question that was asked and answered. But the system was designed over the concept of static information and static notion of truth. Which the world definitely doesn't follow.

[–] f6v 21 hours ago

I agree that SO became very hostile at some point. But I also moderated a bit around 2014-2019 and the vast majority of questions in my field were very low effort. People would just dump their code with no context and ask why it doesn't work.

So, the place was a shit hole.

[–] baud147258 21 hours ago

As someone who spent some time participating in the filtering of SO's questions, considering the quality of some of the new question, I understand why the volunteer moderators pushed SO/SE for those question quality control tools and a push for users to use them. But sadly, unlike the upvote/downvote mechanism, the incentives to use them resulted in hasty decisions (close as dupe), which had an adverse effect on the site's reputation.

[–] shevy-java 22 hours ago

Well, 2) is often a honey pot trap since some projects have no documentation, or only so little documentation that spending time trying to figure out a project is a waste of time. There is a reason why SO got important in the first place. Horrible documentation was one reason.

[–] mschuster91 22 hours ago

> I also realised reading the documentation (that is if available) and looking though github and people's blogs helped better.

For the Javascript ecosystem in particular... it also massively helped that projects finally started to work on their documentation. The churn rate was insane and documentation or working examples took an afterthought for far too long, that is what drove so many people to SO in the first place.

[–] neya 23 hours ago

> 1. SO question numbers dropped precipitously before AI came along.

This was single-handedly the work of arrogant self-righteous moderators who would close random useful questions as not useful just to prove they were doing something. A lot of us long timers got annoyed at selective enforcement of rules and never looked back. It's sad because it was actually a fun place to hangout. Too much moderation always kills the vibe.

[–] btilly 1 day ago

I contend that shitty moderation was why it flatlined.

I enjoyed answering algorithm questions. Unfortunately, every so often, people who weren't into algorithms would show up. And somehow summoned each other. To close good questions.

It was clear that they didn't understand the topic. They couldn't recognize what a good question, or a good answer, was. I wound up going away, coming back every few years, go until my next bad experience, and leave again.

Each time, the population answering those questions was smaller.

I believe that a lot of good people left like I did. Then just didn't come back.

[–] rerdavies 14 hours ago

You need to go deeper. The problem wasn't shitty moderation; it was shitty gamification.

[–] Strilanc 1 day ago

If you look at the data on a log scale, and include other stack exchange sites, it's apparent that the serious decline is starting after ChatGPT launches: https://x.com/CraigGidney/status/2115664509042053377/photo/1

One way that stackoverflow does stand out from the other stackexchange sites is that its half life appears far shorter. On a log scale plot it has a steeper descent than the other stackexchange sites. I'm not sure what to make of that. Another interesting thing is that the math stackexchange initially declines more slowly than the others for a year or two and then decays pretty similarly. You can also see examples in some of the other sites of their decay beginning before chatgpt releases (e.g. drupal). You can also see some sites decaying a decade earlier (e.g. expressionengine). I think those outliers are from their subject being too specific and going out of style.

I got the cross-site data from data.stackexchange.com with this vibe-coded query:

    DECLARE @sql NVARCHAR(MAX) = N'
    SELECT
      site = [$site_name$],
      month = DATEADD(MONTH, DATEDIFF(MONTH, 0, CreationDate), 0),
      questions = COUNT(*)
    FROM PostsWithDeleted
    WHERE PostTypeId = 1
    GROUP BY DATEADD(MONTH, DATEDIFF(MONTH, 0, CreationDate), 0)
    ';

    EXEC dbo.sede_ineachdb
      @SQLCommand = @sql,
      @IncludeMainSites = 1,
      @IncludeMetaSites = 0,
      @IncludeMainMeta = 0;

[–] rerdavies 14 hours ago

So much for the long tail! The rate of decline after the advent of AIs is sobering. From 181,000 questions a month at the absolute peak, to less than a thousand a month now!

[–] Rochus 20 hours ago

That's interesting, thanks! The peak for most topics was around 2015-2020 apparently, and then came a "continuous" decline. If it really was a major event that caused the decline, wouldn't we expect to see a volatile plunge, like on the stock market?

[–] famouswaffles 15 hours ago

Today with Opus, Sol, etc there's basically no reason to visit the site, but when chatGPT launched it was powered by GPT 3.5, which would realistically only be able to answer a (minor) chunk of all possible StackOverflow questions.

[–] Rochus 13 hours ago

There are many potential explanations, including the policy changes in 2018. LLMs are definitely a concurrence to the service, but since the decline started years before they were really useful, other reasons might be relevant. I don't know to what degree. I just wondered why some of the forums had significant peaks after the GPT line.

[–] LaurensBER 23 hours ago

The issue was that Stack overflow never was enjoyable to use.

Asking questions was difficult and any difficulty questions I might ask would not get a response anyway.

Answering questions was a gamble, sometimes you would get rewarded, most of the time you would not.

[–] jonwinstanley 21 hours ago

Right when it first started it was so much better than what had been there before, and the way they assigned points was designed to encourage lots of people to chip in with answers.

Asking questions did become painful later on. But it was the default site to choose whenever you googled a programming issue for years after.

[–] anticensor 1 day ago

https://data.stackexchange.com/stackoverflow/query/1948078/c... shows the situation even better.

And, StackExchange has adjusted stale/inactive question deletion scripts as a result of this change: https://meta.stackexchange.com/questions/419988/we-ve-rework...

[–] fuzzythinker 20 hours ago

Nice. But about 10 years late. They should have seen for 2 years from 2014 to 2016 the damage done from their deletion policy.

[–] tuveson 1 day ago

I like that you can see a Covid bump in there and then it immediately craters lol

[–] kstrauser 1 day ago

Right? It's not a very subtle spike, is it? Even then, if you erased the bump, the graph is pretty close to a straight downward-sloped line from 2016 to late 2022.

It's kind of sad: I bet there was some excitement around the SO office. Is this it? Are we on the way back? Have people remembered that we're here and now they're returning? Alas, no, they weren't, and they weren't. That must've been rough.

[–] dpark 1 day ago

“Has our strategy of being openly hostile to anyone asking questions finally resulted in the increase in questions we always dreamed of? Did our long term bet on alienating the community finally pay off?”

[–] wildzzz 1 day ago

That's probably why the number of new questions asked dropped off. Once someone has asked "how to fizzbuzz in python", that's pretty much it. You'll get the textbook answer, the "why are you fizzbuzzing in the first place" answers, and then an answer a couple years later using the new python fizzbuzz built in library. You have a massive compendium of questions and answers and for most people, that's good enough. I don't think that stack overflow was necessarily less popular in visitors pre-LLMs but they certainly killed it off quick.

[–] zahlman 1 day ago

Yeah, the newcomers asked worse questions than average. There was a huge flood of people presenting like they thought they had a difficult question about how neural network simulations work, but actually had code full of multiple elementary errors (completely unrelated to the domain, just learning the programming language kinda stuff) and no apparent willingness to read error messages even to the extent of "it says TypeError, I guess that must mean something is wrong with a type".

[–] jmyeet 1 day ago

This was all too common across tech. I know of many products at the time that, say, had their usage quadruple in Covid (from a very large baseline) and nobody in leadership seemed to ask the question "is this permanent or just an artifact of people being stuck at home?" Clearly the answer was "no".

[–] smsm42 1 day ago

In my experience, SO became useless many years before AI showed up on the horizon. I stopped actively participating about 2019 and about 2022 gave up on it completely.

[–] hollowonepl 21 hours ago

Agreed, I found SO and Quora and similar sites with general or specific question/answer based communities weak long before AI.

AI replaced Google Search in many ways more directly on arrival than those.

SO to me was weak as it had outdated answers to outdated questions (yet related to search queries) that nobody maintained well.. and the community more focused on patronising to push their point of view than being essentially objective in many ways.

that was a total killer to me and I started preferring to come back to books, manuals and researching topics myself than short cuts like SO.. and that research is heavily optimised during AI/LLM era, yes, but that's just one of many nails to SO's coffin.

[–] Izkata 6 hours ago

> SO had been flatlined since 2014 and declining since 2017.

During this peak popularity period, they were actively trying to stop people from flooding the site with low-quality "do my homework"-style questions.

[–] drumhead 13 hours ago

With SO the experience always felt like you were being chastised for everything you did. It was quite off putting. AI coming along and just giving you working solutions without any attitude was just a more pleasant way to work.

[–] saonnet 21 hours ago

Often overlooked but Discord(and similar communities) are also a big reason why people no longer ask ques. on SO.

[–] nitbard 21 hours ago

No surprises here. However I recently tried to register global account on Stack Exchange network, just to get to some organic conversations and meet people... and I couldn't. Gateway errors, "try again in a few minutes", etc. I did quick research and it seems that I'm not the only one. Maybe I just run into one of the bad days.

[–] Kurtz79 18 hours ago

The graph makes it clear; the question I have though, between the start of SO decline and up to when AI became able to answer SO-like questions (or remove the need for them altogether), where did people go to find an answer?

[–] kstrauser 11 hours ago

Me personally? Reddit. For the languages and technologies I worked with, their subreddits were far friendlier to reasonable questions, even FAQs, even if they weren't pristinely composed. There was also a recency bias that I appreciated. If you brought up a subject that was asked last year, people would link back to the earlier conversation while continuing the current chat given new developments. An example might be:

> Q: What HTTP library should I use?

> A: Last year everyone liked Requests [link to chat], and that's still good, but if you're using async you might look at aiohttp which is newer and not quite as polished yet but has much better async support.

Those were the answers that would've kept SO relevant: here's what we all thought last time this came up, but here's how the situation has evolved since then.

Reddit came with its own problems, of course. The programming subreddits hadn't (yet?) been as weighed down with those issues when I weaned myself off going there.

[–] 1718627440 19 hours ago

I don't get why there is actual complaining about that outcome, because that seemed what the actual intention was. SO made it clear, that they do not want to be a Q&A forum, but a collection of novel and interesting questions. Naturally, a lot of newly asked question were already asked in a similar fashion, which for SO philosophy meant that they could be merged. In addition, they gradually increased the "quality" level new questions need to be have, to be kept.

The declining number of questions is hence not a side-effect or a mistake, but just means that they reach their end-goal. It's actual a measure of success. (Now you can disagree, whether that philosophy was reasonable or sustainable in the first place, but given there choice, this is the logical conclusion and end goal.)

[–] nomilk 22 hours ago

A downward trend in the number of questions doesn't imply the site was in decline, since there's only so many questions you can ask about a given technology, so would-be question askers find the answer via search engine rather than having to create a new question on the site.

Tangental, but I tried plotting number of questions by month including deleted questions. It shows similar trends to questions by month (so nothing particularly interesting): https://imgur.com/a/NLLB3OK

Query: https://data.stackexchange.com/stackoverflow/query/edit/1969...

[–] sharpshadow 18 hours ago

Copilot in VS Code with OpenAI Codex came out in June 2021.

[–] tayo42 1 day ago

Where else were people getting answers and help from between 2018 and AI getting good in the mean time? As I remember it there wasn't really an alternative.

Maybe it was more consolidation of tech and better docs over time.

[–] geraneum 23 hours ago

Maybe they visited the website and the existing questions were enough? These stats seems to be questions asked. Maybe that’s the reason.

[–] jamesfinlayson 1 day ago

I was using StackOverflow pretty heavily in that time. I was doing PHP, Java and Groovy at the time, and PHP's documentation is pretty good. JavaDoc generated from source not so much...

[–] kstrauser 15 hours ago

I think a lot moved to subreddits. You could ask a decent question there and actually get answers, even if it didn’t meet SO’s purity requirements.

[–] Macha 18 hours ago

Github issues

[–] cubefox 23 hours ago

> SO had been flatlined since 2014 and declining since 2017.

I'm fairly certain this early decline came from the fact that many common questions had already been answered, and therefore were available via Google, so hardly anybody asked them again. In the beginning there were just way more unanswered questions.

[–] shevy-java 22 hours ago

Yes that was my impression too. SO already died before the rise of AI.