Almost every single comment on that thread confirms it - this is a self-inflicted damage. You can rarely ask a question or answer one without being overzealously or rudely moderated. Questions are marked as duplicate if they are vaguely similar to an earlier question, even if they're not the same or if the old question is outdated. And nobody thought it worthwhile to rectify this problem despite loud complaints from the majority of its users.
And that's such a shame - LLMs and code copilots don't fully satisfy what SO can do. SO answers often contain deep insights and wisdom. I frequently add SO reference links in my code as comments. It's hard to elicit such explanations from LLMs, and it's hard to get a permalink even if I manage to ask all the right questions. SO is a good product lost to hubris.
zahlman•8mo ago
> You can rarely ask a question or answer one without being overzealously or rudely moderated.
Yes, because most of what is "asked" is fundamentally not compatible with what the site is trying to accomplish.
> Questions are marked as duplicate if they are vaguely similar to an earlier question, even if they're not the same or if the old question is outdated.
In practice, the similarity is much more than vague; I have seen countless complaints over the years about duplicate closure that were completely ridiculous, along with many people rejecting the idea entirely without even trying to understand the purpose.
The standard for a duplicate is well established and clearly laid out:
> And nobody thought it worthwhile to rectify this problem despite loud complaints from the majority of its users.
Because it is not a problem and the complaints come from people who do not understand the site's purpose or design, who are fundamentally trying to use the site as something that it is explicitly designed not to be.
> LLMs and code copilots don't fully satisfy what SO can do. SO answers often contain deep insights and wisdom.
Yes.
And in order for SO to work that way, it must filter out the "questions" (requests for help) that aren't suited for a searchable Q&A database, which are better addressed by an LLM or a discussion forum.
Deep insights have value when they're attached to searchable, well-written questions that properly frame the information provided in the answer.
Answers are searchable when the answers to fundamentally the same question are in the same place.
> SO is a good product lost to hubris.
What are you talking about? The curators these days feel the exact opposite of hubris; the good content is buried under a mountain of crap and the prospect of fixing this is despair-inducing. The only silver lining is that the declining question rate means less time spent on turning away even more crap.
A useful reference that is focused on programming doesn't need to have more than three times as many questions as Wikipedia (which is about literally anything) has articles. But SO has that many. The fact that people have stopped wanting to ask questions is pretty close to the opposite of "a problem", honestly.
goku12•8mo ago
And that's such a shame - LLMs and code copilots don't fully satisfy what SO can do. SO answers often contain deep insights and wisdom. I frequently add SO reference links in my code as comments. It's hard to elicit such explanations from LLMs, and it's hard to get a permalink even if I manage to ask all the right questions. SO is a good product lost to hubris.
zahlman•8mo ago
Yes, because most of what is "asked" is fundamentally not compatible with what the site is trying to accomplish.
Stack Overflow is explicitly not there to provide help (https://meta.stackoverflow.com/questions/284236/why-is-can-s...) or deal with urgent problems (https://meta.stackoverflow.com/questions/326569/under-what-c...). It's not there to help individuals solve personal problems with their code at all.
It's there to answer questions that can be useful to everyone.
Therefore, before asking a question, you must have that purpose in mind. Otherwise it gets closed (https://meta.stackoverflow.com/questions/417476/question-clo...), and for the site to work properly this must happen swiftly (https://meta.stackoverflow.com/questions/260263/how-long-sho...).
> Questions are marked as duplicate if they are vaguely similar to an earlier question, even if they're not the same or if the old question is outdated.
In practice, the similarity is much more than vague; I have seen countless complaints over the years about duplicate closure that were completely ridiculous, along with many people rejecting the idea entirely without even trying to understand the purpose.
The standard for a duplicate is well established and clearly laid out:
https://meta.stackoverflow.com/questions/254697/when-can-a-q...
https://meta.stackoverflow.com/questions/417476/question-clo...
https://meta.stackoverflow.com/questions/384711
https://meta.stackoverflow.com/questions/357021
https://meta.stackoverflow.com/questions/339261
> And nobody thought it worthwhile to rectify this problem despite loud complaints from the majority of its users.
Because it is not a problem and the complaints come from people who do not understand the site's purpose or design, who are fundamentally trying to use the site as something that it is explicitly designed not to be.
> LLMs and code copilots don't fully satisfy what SO can do. SO answers often contain deep insights and wisdom.
Yes.
And in order for SO to work that way, it must filter out the "questions" (requests for help) that aren't suited for a searchable Q&A database, which are better addressed by an LLM or a discussion forum.
Deep insights have value when they're attached to searchable, well-written questions that properly frame the information provided in the answer.
Answers are searchable when the answers to fundamentally the same question are in the same place.
> SO is a good product lost to hubris.
What are you talking about? The curators these days feel the exact opposite of hubris; the good content is buried under a mountain of crap and the prospect of fixing this is despair-inducing. The only silver lining is that the declining question rate means less time spent on turning away even more crap.
A useful reference that is focused on programming doesn't need to have more than three times as many questions as Wikipedia (which is about literally anything) has articles. But SO has that many. The fact that people have stopped wanting to ask questions is pretty close to the opposite of "a problem", honestly.