Live data from Hacker News

What AI did to stackoverflow in a graph

data.stackexchange.com

591–600 of 615 posts

Re: What AI did to stackoverflow in a graph

#591

Earlier quoted context omitted.

> It's just not that important to me for a dead site. I get it, that means "just trust me bro" is in play. See, this is the part where I'm actually getting insulted by the invocation of the "bro" language. No; what's happening here is that my lived experience as a curator of the site (which was quite stressful) is being denied, and it is supposed that evidence to counter me is not necessary. This isn't about "winning…

> a community, and communities are supposed to have a right to self-determination. Your comment can be boiled down to this. And yes, I completely agree with you that a site has the right to operate however it pleases. But that does not mean a site has the right to be free from outside criticism. Exercising your self-determination in a way that is obviously noxious will get called out, which is the right of an outside…

> Exercising your self-determination in a way that is obviously noxious

But we weren't, and it wasn't.

People were overwhelmingly trying to be polite and factual about what needed to happen for a question to be reopened, the entire time I tried using the site. The yelling and cursing overwhelmingly came from people who imagined that they were entitled to do things that were explicitly against policy, simply because of UI affordances.

We are not the ones who created those UI affordances. We had no control over any detail of the reputation system or the review queues. We had no way to show you the things you needed to read except to link them to you, and we were constantly confronted by people who interpreted any kind of hyperlink to policy in a discussion as inherently rude. We had no control over the onboarding flows that failed to show those things to you. The only access we had was to produce a mini-faq (https://meta.stackoverflow.com/questions/251225) out of Q&A on the meta site (which really should not itself have used the Q&A format IMO, but that is also something that was completely out of our control).

And for the cases where people got out of line, there was a flagging system all over the place and moderators who took complaints seriously; and there was also a meta site where people could come to have policy explained in more detail. But it was genuinely rare that people who actually knew better, actually got out of line.

It was far more common for people to complain loudly that they were hard done by because someone downvoted their question (which genuinely merited it, per the purpose of downvotes on the site) and didn't leave a detailed comment explaining why (which is not an expectation ever imposed on Redditors or anyone else, and is an action protected by policy because i) there's hovertext on the down arrow summarizing the general reasons for using it; ii) people who leave those comments often get abusive replies). Or to complain that they were hard done by because their question was closed for being "too easy" (which is not why it was actually closed; an actual closure reason was explicitly labelled in the interface, but a more accurate summary of the typical close reason is "you've identified the symptoms of a problem in your code, but thus far you have not actually distilled this to a question; you just want someone else to make your code work again"). Or to complain that someone had the sheer audacity to pick through a problem description, isolate a single apparent cause, recognize that beginners encounter that issue constantly, and provide an immediate link for free to the existing "canonical" Q&A for it. (After all, why should they have to adapt code that was written against some other abstract example reconstruction of the problem, to their own code? Why should they have to figure out how that pattern applies to their hundred-line code dump, and where?)

> Your perspective, frankly, doesn't matter. It won't rehabilitate the reputation of SO. I'm sorry this hurts as you've obviously invested a lot of time there.

You don't understand. I am not trying to rehabilitate the reputation of SO.

I am trying to rehabilitate the reputation of the Q&A site format.

And I am trying to get people to stop slandering me and other people I made friends with and cared about. In particular, to stop ascribing false and hurtful motivations, and to have some empathy for what we put up with, and to consider the possibility that what we were trying to accomplish might actually be a worthwhile thing to accomplish that serious people might actually genuinely care about.

When people talk about SO, overwhelmingly they talk about me and my friends. And they repeatedly incorrectly describe me and my friends as "moderators", which comes across as sheer apathy.

Of course my perspective matters to how people are treating me on the Internet based on their perception of things that I did.

> Sites die when the experience isn't something end users enjoy.

I feel like I'm banging my head against a wall here. The site was constantly full of people who were fundamentally not apparently interested in the site's goals, not apparently interested in joining or understanding the existing community, not apparently interested in anything but the service they incorrectly believed the site was there to provide. And the constant implication of this discussion — because it always boils down to "the site died and people hated it and hate you and that's bad and reflects badly on you" — is that I'm somehow supposed to feel bad about the fact that they eventually figured out they were in the wrong place and left. (Or perhaps more accurately: LLMs finally provided they place they really wanted to be the whole time; before that, some made do with Reddit I guess.) After I went out of my way to keep my cool volunteering my time against an unending horde, knowing that attempts to "correct the record" off-site would be at best sneered at.

No, I don't feel bad about that. Because feeling bad about that is logically equivalent to rejecting the right of self-determination that you just acknowledged.

Re: What AI did to stackoverflow in a graph

#592
post #546

Earlier quoted context omitted.

Would you be willing to show questions you attempted to ask? I'd be happy to help explain how policy works/worked there.

We know how it works. That's why we all eventually gave up on helping people there (years before I knew what an LLM was.)

If you knew how it worked, you would have understood that answering questions is only indirectly about "helping people", by hopefully building something that third parties can make use of later. The objective is not "the OP's problem was solved", but "the OP's question was answered" (which requires that there is a recognizable question extracted from analyzing the problem; and that there is one such question; and of course, those third parties are better served if they are all routed to the same place where the best answers are already written.)

If you were there to volunteer troubleshooting, or the programming equivalent of "tech support", you were in the wrong place. If you were interested in going back and forth with suggestions to help debug a program, you were in the wrong place. All of that is supposed to happen before the question is posted. The question is about why a specific thing works a specific way, or how to accomplish a specific atomic objective.

Re: What AI did to stackoverflow in a graph

#593

Earlier quoted context omitted.

It is taken very seriously in Stack Overflow editing policy that edits are not supposed to go against the author's apparent intent. But it does involve removing a lot of things that you might want the answer to say, but don't meet the guidelines for what answers are supposed to contain. When you join the site, you agree to Terms and Conditions that, among other things, grant a Creative Commons license to the communit…

But the points system was fundamentally inextricable from the site’s operation, and the officiousness and arrogance it rewarded meant it could never be as good as anybody hoped it would be. The people making the content decisions were often misguided, and vastly less qualified than the authors they meddled with… but were rewarded for their actions all the same.

> But the points system was fundamentally inextricable from the site’s operation, and the officiousness and arrogance it rewarded meant it could never be as good as anybody hoped it would be.

Er... yes.jpg ?

You seem to be under the impression that I'm trying to defend all the site's design choices. I'm not. I'm trying to defend a community that did the best it could with what they were given, after all sorts of requests were repeatedly ignored or denied (or preemptively judged unreasonable and not even made) and then new ownership came in and tried to force an unending torrent of even worse ideas on us, even undoing the few things we actually liked about the design.

> The people making the content decisions were often misguided, and vastly less qualified than the authors they meddled with… but were rewarded for their actions all the same.

Again, qualification in the subject matter is entirely orthogonal to understanding of what the site is trying to accomplish, what the policies are, and how those policies were designed to align with the objectives.

That was the part we did have control over. But we were making a good faith effort to carry forward the oral tradition of Atwood and Spolsky's vision.

Re: What AI did to stackoverflow in a graph

#594
post #546

Earlier quoted context omitted.

We know how it works. That's why we all eventually gave up on helping people there (years before I knew what an LLM was.)

If you knew how it worked, you would have understood that answering questions is only indirectly about "helping people", by hopefully building something that third parties can make use of later. The objective is not "the OP's problem was solved", but "the OP's question was answered" (which requires that there is a recognizable question extracted from analyzing the problem; and that there is one such question; and of…

I believe the disconnect here is you're describing how some wanted it to work vs. how it played out in the real world. In practice, it became an unrewarding place to contribute high quality original content to.

Re: What AI did to stackoverflow in a graph

#596

Earlier quoted context omitted.

meh, in my experience good quality LLMs are significantly more correct than Stack Overflow ever was and have access to your code and machine to help you diagnose issues.

It's about 50:50 for me. Often times they completely fuck things up and then want to rewrite rather than correct the original issues. That leads to debugging their rewritten code and an afternoon down the drain.

you must be very unlucky

Re: What AI did to stackoverflow in a graph

#597

Earlier quoted context omitted.

> a community, and communities are supposed to have a right to self-determination. Your comment can be boiled down to this. And yes, I completely agree with you that a site has the right to operate however it pleases. But that does not mean a site has the right to be free from outside criticism. Exercising your self-determination in a way that is obviously noxious will get called out, which is the right of an outside…

> Exercising your self-determination in a way that is obviously noxious But we weren't, and it wasn't. People were overwhelmingly trying to be polite and factual about what needed to happen for a question to be reopened, the entire time I tried using the site. The yelling and cursing overwhelmingly came from people who imagined that they were entitled to do things that were explicitly against policy, simply because o…

I'm ending here. You are too emotional and now you are accusing (I assume myself) of slander.

Grow up.

Re: What AI did to stackoverflow in a graph

#598

Earlier quoted context omitted.

Mod vs non-mod-user-with-edit-privileges is a semantic difference from a user perspective. Nobody gives a shit what the internal labeling system looks like or hierarchy among the people with edit privileges. I’m not going to litigate my case in front of a clique of other officious hall monitors just for the privilege of making that site better. I don’t care if it was against the rules for me to be annoyed by their ob…

> Mod vs non-mod-user-with-edit-privileges is a semantic difference from a user perspective. If your lack of interest in how the existing community works is that profound, then you could hardly be said to have a good-faith interest in joining the community in the first place. Which means you're actually upset that the site doesn't provide the service you expected it to . Sites are not required to provide services sim…

You’re trying to lawyer your way around this very plain fact: in the minds of the vast majority of people who knew about it, it was a site to ask questions about relatively obscure topics, and for knowledgeable people to answer them to disseminate their knowledge. You make it sound like users were supposed to consider using the site as their primary goal, and consider the subject at hand as secondary. If you read what people were saying here in good faith (or really what damn near everyone outside of that insular community has said about it for like a decade and a half) you’d realize that you are in the extreme minority. You can cite all the governance concepts rules you want, but you can’t make users care about something they don’t, and that’s the problem. There was a group of people operating the site like the site itself was the reason it was popular. It was popular because it had a lot of information people wanted, like Wikipedia is. And there are a hundred billion ways it could have pimped its best questions to Google without having the layer of obnoxious, arrogant people playing question police.

Yes, the points system was a mess because of the way it made people behave. You can’t look at it in a vacuum or pretend the negative behavioral consequences were somehow different from the purer intentions of the site. When someone that has permission to change what you say and they abuse that, it doesn’t matter if there’s some justification in the site concept — we’re just going to be bothered by the bothersome people, and tell other people about how annoying the experience was. It could have have been structured to prioritize stewarding people’s knowledge, like Wikipedia, but it wasn’t. There could have been a much greater barrier to being able to meddle with other people’s words but there wasn’t. They could have chosen to not reward people for doing exactly that, but they didn’t. Instead, in interface and in the audience of privileged users it attracted, it prioritized gamification, popularity, the ‘erm aactually’ guys that were way out of their depth, and a whole bunch of other shit that everyone else rolled their eyes at. It was successful in spite of the obnoxious governance and self-important pseudo stewardship of other people’s knowledge, not because of it.

If SO was even vaguely still relevant to the conversation, I’d still be telling people that it’s time for a big ego deflation and giant reality check. But it’s not relevant. If preponderance of users were really enamored with the community and governance and rules and site structure and all of that shit, rather than it simple having information on obscure topics and issues where you could get a question answered by someone who knew what they were talking about, then everybody wouldn’t have jumped ship literally the moment they had an alternative. Wikipedia didn’t lose anywhere close to that about amount of traffic. Search results are also barely relevant these days, comparatively, but I’ll bet the people there now are just as overzealous about policing questions for the same justifications.

Re: What AI did to stackoverflow in a graph

#599

Earlier quoted context omitted.

> Mod vs non-mod-user-with-edit-privileges is a semantic difference from a user perspective. If your lack of interest in how the existing community works is that profound, then you could hardly be said to have a good-faith interest in joining the community in the first place. Which means you're actually upset that the site doesn't provide the service you expected it to . Sites are not required to provide services sim…

You’re trying to lawyer your way around this very plain fact: in the minds of the vast majority of people who knew about it, it was a site to ask questions about relatively obscure topics, and for knowledgeable people to answer them to disseminate their knowledge. You make it sound like users were supposed to consider using the site as their primary goal, and consider the subject at hand as secondary. If you read wha…

> You’re trying to lawyer your way around this very plain fact

It's not lawyering. I'm extremely well aware of the impedance mismatch. It wasn't created by the community. It was created by the site software failing to align with the site's goals.

> in the minds of the vast majority of people who knew about it, it was a site to ask questions about relatively obscure topics, and for knowledgeable people to answer them to disseminate their knowledge.

Well, no. "The vast majority of people who knew about it" were asking beginner-level questions, as is to be expected, because every field of human endeavour is full of neophytes. They believed their questions to be obscure due to a lack of expertise. They often were obscure; not because of the subject matter but because of a lack of attempt to debug the code (and, often, a lack of general awareness of the concept of debugging). And they were often not coherent questions at all, or else they were multiple questions arising from multiple problems with the same code sample.

But it was a site to ask questions and for knowledgeable people to answer them to disseminate their knowledge. That is not even slightly inconsistent with anything I've said here.

The problem is, "this is a place to ask and answer questions" is not the same thing as "this is a place to show what's causing you a problem and get someone to help you, specifically" and also not the same thing as "this is a place to help people solve their problems one-on-one".

> You make it sound like users were supposed to consider using the site as their primary goal, and consider the subject at hand as secondary.

Yes, because they were. Just like they are on every other website, e.g. this one. When people come to HN and post things that are counter to the guidelines, other community members use downvotes, flags and polite replies (and, in some cases, emails to the site staff) to address the issue. This much is obvious and I don't understand how it can be a point of contention.

> If you read what people were saying here in good faith (or really what damn near everyone outside of that insular community has said about it for like a decade and a half) you’d realize that you are in the extreme minority.

I have been reading it for most of that decade and a half. I have been flooded with it again and again and again. I literally cannot find an off-site discussion of Stack Overflow that isn't full of it. I am intimately familiar with the argument, and you have said nothing new here.

What I have said over and over again, which everyone outside "the extreme minority" seems unwilling to accept, is that this does not matter. If people come to a place and don't care about how things work in that place, the place is not in any way obligated to change to accommodate them. The people should leave of their own accord, and they should accept that they misunderstood the place, and accept that the place has a right to sovereignty, and find a different place that does work the way they want.

There were countless places that did work the way the typical new SO user wanted. SO existed specifically because of a belief that those places are incapable of producing a useful knowledge base, and cause experts to burn out on repeating basic advice. (SO use patterns demonstrated to me that actually there are more than enough people around who are happy to repeat basic advice and operate a free help desk for beginners; if they get upset, it's because the beginners still fail to understand things after they were explained. But those people were actively interfering with the site's objectives. They were far more destructive than any beginner could ever be.)

> And there are a hundred billion ways it could have pimped its best questions to Google without having the layer of obnoxious, arrogant people playing question police.

You had over a decade and a half to try proposing some on meta.

> Yes, the points system was a mess because of the way it made people behave.

Specifically, it enabled the people I describe in the above parenthetical, yes. But you present this argument as if you think people were downvoting and closing questions, and otherwise trying to uphold policy, in order to gain reputation points.

That is the literal exact opposite of how the system worked. You got reputation for basically everything else, which then enabled you to take those actions. You got 0 reputation for voting to close something, or for "dupe-hammering" it with that specific privilege. You got either 0 or -1 reputation for downvoting something.

You could potentially get 2 reputation points for editing something, if you didn't already have unilateral edit privileges and your edit was accepted (and because the people who could accept your edit could also edit unilaterally, and were doing all of this specifically to try to make the site better by getting good edits in place ASAP, they were therefore incentivized not to give you that reputation.) Plus there was a cap on total reputation you could get that way; you still had to ask and answer questions to reach the unilateral edit privilege.

By contrast, you got +25 for an upvote and accept-mark on an answer, no matter how clearly the question violated site guidelines; versus -2 from an annoyed curator downvoting you for not understanding that you weren't supposed to answer the question and were supposed to help close it instead. (Plus you got a five minute window to post if you'd started, even if the question got closed immediately. Plus you could do this unilaterally while three curators would have to agree to close except for duplicates. Plus only a small fraction of the userbase could cast close votes, while everyone could answer.)

> There could have been a much greater barrier to being able to meddle with other people’s words but there wasn’t.

That would have been an actively bad thing. The editing was by well-considered design, specifically to make the site better. The edits people complain about overwhelmingly do, objectively, obviously make the site better and I'm frankly tired of trying to explain the obvious. I have, over and over, for years, tried to point people at the policy explaining this. Nobody ever even tries to engage with it.

You may have noticed that Wikipedia frowns on signatures on main-space pages, and expects citations for claims, for example. SO policy is roughly analogous.

> If preponderance of users were really enamored with the community and governance and rules and site structure and all of that shit, rather than it simple having information on obscure topics and issues where you could get a question answered by someone who knew what they were talking about, then everybody wouldn’t have jumped ship literally the moment they had an alternative.

Nobody ever tried to say they were, certainly not me.

What I'm saying is that the people who didn't understand this stuff, shouldn't have been there in the first place. Because simply getting questions answered is not how you build a useful Q&A reference, and that is what the creators of the site wanted to have. Which is amply demonstrated by looking at a random sampling of questions (you can do this yourself on data.stackexchange.com with an SQL query, if you like).

I am genuinely happy they found the alternative(s) they wanted the whole time, and should have been using the whole time (well, as they became available).

Re: What AI did to stackoverflow in a graph

#600

Earlier quoted context omitted.

> Exercising your self-determination in a way that is obviously noxious But we weren't, and it wasn't. People were overwhelmingly trying to be polite and factual about what needed to happen for a question to be reopened, the entire time I tried using the site. The yelling and cursing overwhelmingly came from people who imagined that they were entitled to do things that were explicitly against policy, simply because o…

I'm ending here. You are too emotional and now you are accusing (I assume myself) of slander. Grow up.

Just throwing this out there: the post you're replying to had a litany of things like "we had no control over the fact that the site's owners and maintainers implemented a completely different UX than our interpretation of the policies would allow for".

The site's builders put one set of UX patterns in place. The site's visitors who created things like that linked mini-faq tried to put another in place. When the two are in obvious conflict, as they clearly frequently are, the people who wrote the mini-faq etc. claim it's because the people who built the site are wrong, as is every other non-personally-invested user of the site.

You're not going crazy, or if you are, at least you're not going alone.

Post reply on HN