That’s why it succeeded. It was not a forum, it was something new. Contribute a new question that was well stated and broadly applicable and get a response. You had to earn your answers.
It wasn’t flawless, but it was good and so good it fed all the AI agents with good materials.
It wasn’t a “help me!!!111!” site, it was an exchange.
I put lots of effort into narrowing my questions, eliminating obvious answers by demonstrating the work I put in, and editted all my questions like an email to my boss. I was almost always rewarded with good answers.
[–] Aurornis 1 day ago
[–] Skunkleton 1 day ago
[–] zahlman 23 hours ago
[–] joe_the_user 23 hours ago
Actually, talking to ChatGpt, I found this article: “Stack Overflow Isn’t Very Welcoming. It’s Time for That to Change.” ( https://stackoverflow.blog/2018/04/26/stack-overflow-isnt-ve...) and that actually shows what the SO admins didn't get (imo obviously).
They thought the problem was people not being nice. But that actual problem was that SO was "gamified" and it's gamified interface made it impossible to win unless someone was ultra-competitive (and I'd those types burned-out over time as well). People not being nice was real but it was side-effect of the situation (questions answered instantly so a would-be question answered had to be constantly racing).
It's easy to imagine how the admins didn't see the situation. Each up of intensity seemed to up the engagement and so seemed a good thing. The question of whether the process was sustainable didn't appear and when there were problems it was attributed to something else.
I think
[–] cowlevel 14 hours ago
Even though you didn't get a reputation score for that specific action as I remember, it still gave you the little "I'm helping" feeling.
The honeypots in the review queue trained you to err on the side of closing questions - you were punished for not closing a honeypot when other people had closed it, regardless of whether it actually deserved to be closed. Complaints about incorrect honeypots were frequent and the system was not improved.
[–] braiamp 11 hours ago
There's zero reward for closing a question. I know that because I casted 2 thousands of those, no reward ever came my way.
> The honeypots in the review queue trained you to err on the side of closing questions
That's also false, there were as many for either side in the audit queue. There's a whole list of post about people complaining[1]. Most of the time, the answer is the same: if you are not sure, hit skip.
For most the negatives audits, SO famously used spam rather than bad posts (which also taught the wrong lesson)
https://meta.stackoverflow.com/search?q=%5Bdisputed-review-a...
[–] joe_the_user 12 hours ago
Question-closing was a big complaint through-out SO's history. My guess is that the top people were unconcerned because someone who complained about it was by the way the system defined people, a bad user (since they were lumped-in with those who really did ask already answered questions).
[–] cowlevel 7 hours ago
[–] zahlman 23 hours ago
No, it absolutely very much was not. Even granting that you mean actions of the community (what we call "curation") rather than the small and extremely strained moderation team, way too much stuff historically got through, while people yelled at the OP in the comments and argued over downvotes instead of taking the action that actually matters: voting or flagging to close the question. Worse, people would post answers that acknowledge the problem with the question and asking OP in the answer text to fix things.
I know because I was part of that mess back then, because I didn't know better. That, in turn, was mostly because the entire set of ground expectations was poorly communicated, and the community wasn't given the tools to improve that communication (e.g. couldn't edit the help pages, had their feedback about the discoverability of site documentation ignored, etc.). But it was also because Atwood and Spolsky hadn't refined their intuition about why beginners' bad questions are bad. (Hint: it is not because most of their problems are easy to solve, or "beneath" experts.)
> when following a StackOverflow link from Google
Google got worse at search over the years, so you ended up seeing low-quality versions of the question more often. And the site got way too big (there is no reason why there should be more than three times as many publicly visible questions on Stack Overflow, as publicly visible articles on Wikipedia), such that search results could get diluted with irrelevant low-quality questions that happened to have a lot of keyword overlap with what you wanted. And misleading questions didn't get cleaned up, so you would search for X and find a question where the OP hadn't tried to analyze the situation at all and ended up asking about X when the problem was actually Y, and got the question closed as a duplicate of the main Y "canonical". (I know because I used to lose considerable amounts of hair trying to find the X canonical that I knew was there but stubbornly refused to turn up in searches.)
If "moderation" had actually "always" been "heavy" then you wouldn't have had this problem because this stuff would have gotten hidden from public view, one way or another.
> a moderator shutting it down
Again, these actions are overwhelmingly not taken by moderators, and phrasing things this way reflects a refusal to interact with the community on its own terms and understand its policies. Those who hold an outsider mindset are bound to end up feeling like outsiders.
[–] cowlevel 14 hours ago
[–] jrflowers 21 hours ago
You know that users are going to call anything that happens to their posts on any website “moderation” right, and after decades of that being the norm that’s what the word means
[–] braiamp 11 hours ago
[–] ChrisMarshallNY 1 day ago
Huh. I had the opposite experience. The better my questions got, the fewer responses I got.
I stopped bothering, after a number of questions went answered only by me (after I found the solution on my own).
It did reduce the insults, but that seemed to be the only gain that I got, for carefully articulating my questions.
LLMs aren't perfect, but they don't imply that I'm being lazy, for not already having the answer, and they do give me an answer, instantly (and, like SO, the answer is not always the best one).
[–] schneems 1 day ago
I have engaged with some younger devs who don't even bother turning the thought into a question. They make statements and expect you to do the work of inferring the question. Which I feel like they got from chatbots since it's faster and LLMs don't care if you make them jump through hoops.
> The better my questions got, the fewer responses I got.
There is a sweet spot of providing just enough detail but not too much in asking a good question. And even the best only “bat 300”. I've posted answers to my own questions plenty of times when I got no response.
I hear you on the hostility. Also a lot of people didn't experience how bad things were BEFORE stack overflow. The first time I tried asking a question on IRC I got mocked so bad I logged out and never tried again. Most forums were like that.
I gravitated to Ruby back then because it was one of the only forums people would be patient and empathetic with me.
[–] usedtobegood 1 day ago
I have repeatedly had this experience in trying to craft The perfect question for a consultant who was also a mathematics professor, in finding the essence of the question the answer becomes clear and I never even have to send the email
[–] basilikum 19 hours ago
[–] ChrisMarshallNY 1 day ago
Hear you on Ye Days of Yore. I never did IRC, but I did try UseNet, and it was basically a public toilet.
Everyone bitches about the Eternal September, but, in my opinion, it actually reduced the hostility and barbarism. Clueless dolts were a welcome break from the malevolent assholes.
[–] cowlevel 1 day ago
[–] noosphr 1 day ago
[–] cowlevel 7 hours ago
[–] noosphr 5 hours ago
[–] Pet_Ant 17 hours ago
The “answer my own question” feature was meant for exactly that. Had quite a few like that, and sometimes I still got better answers than I found
[–] zahlman 23 hours ago
Yes, that was part of the point. I hope you posted anyway (after checking for a duplicate, and assuming the underlying problem wasn't just a typo), and provided your own answer for it. That is an explicit part of how the site is supposed to work (but an incredibly poorly communicated one).
> Also a lot of people didn't experience how bad things were BEFORE stack overflow.
SO actually considerably improved on that axis over time. It's just that tons of people erroneously think that being expected to meet certain standards in asking a question (in particular, doing the necessary debugging work to locate and reproduce a problem, or the necessary contemplation to properly specify a task), is the same thing as yelling at people for not knowing answers to easy things. That goes both ways: people get upset when their question is closed, and people think they're helping by excluding easily answered questions because they're easily answered. In reality, the main problem is questions that are easy but complex, i.e. which really break down into multiple separate easy questions or require digging through a code dump to find a trivial error.
[–] yallpendantools 1 day ago
From my experience in Home Improvement and Bicycles, I almost always get in some form of contention with the mods over the boundary between "broadly applicable" and "too specific". I'd do the esr-minimum[1] of showing my homework by posting pictures of my WIP and citing Youtube videos I'm trying to follow but, apparently, this is already toeing the line. I honestly don't know what the mods expect of a good question there. Of course, there will be an appreciable degree of case-specificity in my questions because if the YT tutorials could already answer my problem, I would have no need to ask for help and advice!
Then we have sites like Literature SE which has been languishing in their sandbox since literally forever. Like most SE sites, the guidelines bid you to "avoid asking for opinions or open-ended discussions" and "be specific" but I find that untenable in a Literature site. Notably, sister site SF&F is comparably thriving but there is a fair amount of open-ended speculation going on in there.
Lastly, IMO, it just suffers because at its core its problem is human and social, not technical. In an effort for better curation we ended up with Computer Science, Theoretical Computer Science, and Computational Science. For some posters, finding the right SE site already introduces a huge wall to overcome for posting. This hampers a lot of the out group: students who might be seeking homework guidance or looking to supplement their own readings and lectures, amateurs who are just curious about something in the field. I've had a question in one of the Science SEs outside my field that was well-received by the community but one mod kept "tagging to delete" (I forgot the exact term) almost every year. This mod action has a prescriptive period and, boy, he kept at it. Honestly, I don't care in the end. I asked the question out of sheer curiosity and got a satisfactory answer. Yet I wonder how exactly they expect to foster a community this way.
[–] RHSeeger 1 day ago
And _then_ it became so much more hostile to users - and lost a lot of what made it so wonderful. It became very common that asking a new, good question would be marked as a duplicate because there was something it wasn't a duplicate of.
[–] mm321 1 day ago
What amazes me is that an LLM can actually make sense out of all this crappy training data and often produces meaningful answers. (maybe that's a sign that we already have AGI)
[–] eYrKEC2 1 day ago
[–] bluedino 16 hours ago
That's exactly what it turned into, though.
https://meta.stackoverflow.com/questions/320536/what-is-the-...
[–] slicktux 1 day ago
I detested it but quickly realized that I needed to put in work and learn and only when I understand can I ask valid questions to real problems…not just asking for someone to spoon feed me an answer.
I really appreciate that experience…
[–] ifixeditonmyown 1 day ago
Really? Because I only ever asked three things, technically two I suppose, and the only significant interaction I got from there was a guy saying they thought the third question was a duplicate but they only didn't mark it that way because they saw it was me asking again.
This dude straight up admitted to be on the hunt for marking things as duplicate and only put brainpower into doing not-that because something felt off in is gut.
Then someone else did.
No, he didn't actually answer my question. No one did.
[–] wbillingsley 1 day ago
If StackOverflow is being considered hostile now it's big ... well, Google the three stages of enshittification I suppose.
[–] BrandoElFollito 8 hours ago
After hundreds of thousands of rep points (it's a long time I haven't been there,) - this is simply not true anymore.
It was the case at the very beginning, and then there was a flood of useless questions, and the psychotic users decided to downvote everyone. Some groups (such as Golang) became a real sect. Meta was for unhinged psychopaths.
I loved SO and spent a long time there. At some point it became a 4chan. Shame.
Plenty of the nework sites zre still great though.
The moderation style in the latter years was very different than at the beginning.
It was always heavily moderated, but after some point it felt like there were more layers of moderation and rules and cultural expectations added every year.
Moderation went from being a background cleanup operation that quietly kept the site in line to being the primary thing I expected to see when following a StackOverflow link from Google. In the later years it felt like every other StackOverflow search result was going to be the question I had followed by a moderator shutting it down for some reason before anyone could answer.
The moderation wore me out and I wasn’t even posting questions, just trying to find answers!