4 ms·

In my own experience it's not often the elected moderators that are the problem, but those with a golden tag in a specific tag. They're far too eager to close q

In my own experience it's not often the elected moderators that are the problem, but those with a golden tag in a specific tag. They're far too eager to close questions because they're the ones culling through a tag often - and then close the question as they quickly think "oh, it's that again". But it often isn't, they just didn't spend enough time to see nuance. And neither do they see that even if _they_ understand that the question linked to is the same thing, there is no way the asker can understand what the similarity is from their knowledge point of view (or why the linked duplicate question is the same question).

esafak 1y ago
How do you think they should have handled closing questions, if at all?
- bombcar 1y ago
  Why close questions? Is there a limit on storage space? dang doesn't go and delete all the infinite failed submissions to HN, after all.
  - zahlman 1y ago
    > Why close questions? Because we're trying to build a searchable reference, such that if you try to look for an existing question, you a) find it; b) find the right question; c) find the best possible version of that question; d) can readily tell that you found what you want. And because we are explicitly not trying to build a discussion forum, social media, "HN but specifically for programming questions", or anything else like that. You might as well ask: why delete newly created pages on Wikipedia, or revert edits to existing pages?
  - shagie 1y ago
    Should all 10,000 questions ( https://stackoverflow.com/questions/linked/218384?lq=1 https://stackoverflow.com/questions/linked/218384?lq=1 ) that are duplicates of https://stackoverflow.com/questions/218384/what-is-a-nullpointerexception-and-how-do-i-fix-it https://stackoverflow.com/questions/218384/what-is-a-nullpoi... be open and still allow people to try to answer each instance of the person's question?
    - zahlman 1y ago
      Heh, I hadn't realized the infamous null pointer/reference question was that well cited. The most egregious cases in the Python tag only have about 3500 (https://stackoverflow.com/questions/20109391 https://stackoverflow.com/questions/20109391) and 1500 (https://stackoverflow.com/questions/1373164/ https://stackoverflow.com/questions/1373164/) links respectively. ... Okay, I want to walk back something I said in some other comments here. There is definitely a class of SO questions that get closed as duplicates inappropriately. I tend to forget about the first of the questions because it's not generally a suitable dupe target when it's used: it's a meta question, explaining how to fix your question, rather than actually answering it. But, as you might infer, that means your question should still be closed - it lacks debugging details. I fought against this trend on meta: https://meta.stackoverflow.com/questions/426205 https://meta.stackoverflow.com/questions/426205 . Unfortunately, there's another incentive misalignment here: dupe-hammering the question allows users with a gold badge to act more quickly on questions that don't meet site standards but are likely to attract a quick answer that interferes with keeping the site clean. The second one... honestly probably isn't the best version of the question, but it's attracted good answers and become "canonical". The problem is that thinking in terms of "variable variables" isn't necessarily the right way to think about the problem (dynamically modifying namespaces; or rather, the fact that Python's namespaces are reflected as objects that can in most cases be modified meaningfully) - but it does map pretty well to how a beginner would typically think about the problem. It just tends to overlap with other reasonable questions in a messy way. On Codidact, I've attempted to address the problem space more proactively, but I think I didn't complete the project I had in mind.
      - shagie 1y ago
        https://meta.stackoverflow.com/a/262080/ https://meta.stackoverflow.com/a/262080/ has a data point in time when it could have been done... > How much traffic do the questions that get duped to something bring? Especially the (currently) 410 questions linked to the Java NPE question. You get the couple of FGITW answers on it and the answer is over there, and closed to keep more people from trying to answer it (I hope the dup hammer is helping)... but now it's a closed question with 0 score, 100 views after a year... and five answers (one of which was accepted)... and no one will ever find it. That was in 2014. --- There are some misaligned incentives. There are probably people who dup vote to try to boost their reputation for some reason. The problem (as I saw it) was that the tools of moderation and curation had too much friction and limits placed on them. As the number of questions grew faster than the people who would curate them did, and the tools to curate them were diminished... you've got the problem of "there are two tools to curate and moderate left. One is to close the question. The other is to be a jerk to try to disincentivize the person from doing that again." I wrote about the second bit... a few years ago. Rudeness – the moderation tool of last resort -- https://shagie.net/2016/09/16/rudeness-the-moderation-tool-of-last-resort/ https://shagie.net/2016/09/16/rudeness-the-moderation-tool-o... Things like making it harder to not see low quality questions, or close them, or delete them... > Thus rudeness and the attempt to drive an individual away because other moderation tools have run out or are ineffective. Rudeness is the moderation tool of last resort. When one sees the umteenth “how do I draw a pyramid with *” in the first week of classes on a programming site – how does one make it go away when the moderation tools have been fully exhausted? Be rude and hope that the next person seeing it won’t post the umteenth+1 one. With respect to Stack Overflow, I believe that they've exhausted the people capable of doing moderating without rudeness and are now employees trying to moderate the core group rather than the core group empowered to moderate the site. Eventually, there will be no more left in the core group. Other sites, with a narrower focus (e.g. GitHub discussions) are more able to handle the better focused questions and smaller user bases.
- paulryanrogers 1y ago
  Perhaps duplicates could be classified as useful (therefore SEO indexed and pointing to original) or noise (not indexable)
zahlman 1y ago
> In my own experience it's not often the elected moderators that are the problem, but those with a golden tag in a specific tag. They're far too eager to close questions because they're the ones culling through a tag often - and then close the question as they quickly think "oh, it's that again". > But it often isn't, they just didn't spend enough time to see nuance. As a gold badge holder (for Python and a few other things), I see this complaint constantly. It is without merit ~90% of the time. The simple fact is that the "nuance" seen by the person asking the question is just not relevant to us, because the point of the site is not to give you a personalized answer, but to build a reference where the questions are useful to everyone. This entails collecting useful answers together so that people with fundamentally the same question can all find them, instead of it depending on how lucky their search engine of choice is feeling today. The meta site has historically been flooded with people trying to reopen blatant duplicates based on trivial distinctions, at the level of "no, I want to get the Nth item of a list, not a tuple". That isn't a direct quote, but it's not an exaggeration either. I wish it were. We do make mistakes, in part because there's pressure to act quickly. It's much harder to keep the site clean when answers get posted where they shouldn't be. Closing questions prevents answers from coming in. > there is no way the asker can understand what the similarity is from their knowledge point of view (or why the linked duplicate question is the same question). I try to leave a comment to explain the connection when it isn't obvious. (Another common thing that happens is that the problem someone wants to solve involves an obvious two- or three-step procedure, and each step is a matter of fundamental technique that's already been explained countless times.) But overall, it isn't our goal to teach. We answer very simple questions, and very difficult questions; but we aren't designed to teach. Sometimes it's hard to ask a simple question, because you have to figure out what the question is first. It's unfortunate that people who need the question answered often don't have that skill. But if we have a high quality version of that question already, we can direct people there. Sometimes the linked duplicate isn't the best choice. You can help by finding and promoting a better choice - on the meta site and in the chat rooms. You can also help by editing common duplicate targets - both questions and answers - so that it becomes more clear to people who would actually have the question, that they're in the right place (and so that the information in answers is more readily applicable to them).
- moring 1y ago

esafak 1y ago

How do you think they should have handled closing questions, if at all?

bombcar 1y ago
Why close questions? Is there a limit on storage space? dang doesn't go and delete all the infinite failed submissions to HN, after all.
- zahlman 1y ago
  > Why close questions? Because we're trying to build a searchable reference, such that if you try to look for an existing question, you a) find it; b) find the right question; c) find the best possible version of that question; d) can readily tell that you found what you want. And because we are explicitly not trying to build a discussion forum, social media, "HN but specifically for programming questions", or anything else like that. You might as well ask: why delete newly created pages on Wikipedia, or revert edits to existing pages?
- shagie 1y ago
  Should all 10,000 questions ( https://stackoverflow.com/questions/linked/218384?lq=1 https://stackoverflow.com/questions/linked/218384?lq=1 ) that are duplicates of https://stackoverflow.com/questions/218384/what-is-a-nullpointerexception-and-how-do-i-fix-it https://stackoverflow.com/questions/218384/what-is-a-nullpoi... be open and still allow people to try to answer each instance of the person's question?
  - zahlman 1y ago
    Heh, I hadn't realized the infamous null pointer/reference question was that well cited. The most egregious cases in the Python tag only have about 3500 (https://stackoverflow.com/questions/20109391 https://stackoverflow.com/questions/20109391) and 1500 (https://stackoverflow.com/questions/1373164/ https://stackoverflow.com/questions/1373164/) links respectively. ... Okay, I want to walk back something I said in some other comments here. There is definitely a class of SO questions that get closed as duplicates inappropriately. I tend to forget about the first of the questions because it's not generally a suitable dupe target when it's used: it's a meta question, explaining how to fix your question, rather than actually answering it. But, as you might infer, that means your question should still be closed - it lacks debugging details. I fought against this trend on meta: https://meta.stackoverflow.com/questions/426205 https://meta.stackoverflow.com/questions/426205 . Unfortunately, there's another incentive misalignment here: dupe-hammering the question allows users with a gold badge to act more quickly on questions that don't meet site standards but are likely to attract a quick answer that interferes with keeping the site clean. The second one... honestly probably isn't the best version of the question, but it's attracted good answers and become "canonical". The problem is that thinking in terms of "variable variables" isn't necessarily the right way to think about the problem (dynamically modifying namespaces; or rather, the fact that Python's namespaces are reflected as objects that can in most cases be modified meaningfully) - but it does map pretty well to how a beginner would typically think about the problem. It just tends to overlap with other reasonable questions in a messy way. On Codidact, I've attempted to address the problem space more proactively, but I think I didn't complete the project I had in mind.
    - shagie 1y ago
      https://meta.stackoverflow.com/a/262080/ https://meta.stackoverflow.com/a/262080/ has a data point in time when it could have been done... > How much traffic do the questions that get duped to something bring? Especially the (currently) 410 questions linked to the Java NPE question. You get the couple of FGITW answers on it and the answer is over there, and closed to keep more people from trying to answer it (I hope the dup hammer is helping)... but now it's a closed question with 0 score, 100 views after a year... and five answers (one of which was accepted)... and no one will ever find it. That was in 2014. --- There are some misaligned incentives. There are probably people who dup vote to try to boost their reputation for some reason. The problem (as I saw it) was that the tools of moderation and curation had too much friction and limits placed on them. As the number of questions grew faster than the people who would curate them did, and the tools to curate them were diminished... you've got the problem of "there are two tools to curate and moderate left. One is to close the question. The other is to be a jerk to try to disincentivize the person from doing that again." I wrote about the second bit... a few years ago. Rudeness – the moderation tool of last resort -- https://shagie.net/2016/09/16/rudeness-the-moderation-tool-of-last-resort/ https://shagie.net/2016/09/16/rudeness-the-moderation-tool-o... Things like making it harder to not see low quality questions, or close them, or delete them... > Thus rudeness and the attempt to drive an individual away because other moderation tools have run out or are ineffective. Rudeness is the moderation tool of last resort. When one sees the umteenth “how do I draw a pyramid with *” in the first week of classes on a programming site – how does one make it go away when the moderation tools have been fully exhausted? Be rude and hope that the next person seeing it won’t post the umteenth+1 one. With respect to Stack Overflow, I believe that they've exhausted the people capable of doing moderating without rudeness and are now employees trying to moderate the core group rather than the core group empowered to moderate the site. Eventually, there will be no more left in the core group. Other sites, with a narrower focus (e.g. GitHub discussions) are more able to handle the better focused questions and smaller user bases.
paulryanrogers 1y ago
Perhaps duplicates could be classified as useful (therefore SEO indexed and pointing to original) or noise (not indexable)

zahlman 1y ago

> In my own experience it's not often the elected moderators that are the problem, but those with a golden tag in a specific tag. They're far too eager to close questions because they're the ones culling through a tag often - and then close the question as they quickly think "oh, it's that again". > But it often isn't, they just didn't spend enough time to see nuance. As a gold badge holder (for Python and a few other things), I see this complaint constantly. It is without merit ~90% of the time. The simple fact is that the "nuance" seen by the person asking the question is just not relevant to us, because the point of the site is not to give you a personalized answer, but to build a reference where the questions are useful to everyone. This entails collecting useful answers together so that people with fundamentally the same question can all find them, instead of it depending on how lucky their search engine of choice is feeling today. The meta site has historically been flooded with people trying to reopen blatant duplicates based on trivial distinctions, at the level of "no, I want to get the Nth item of a list, not a tuple". That isn't a direct quote, but it's not an exaggeration either. I wish it were. We do make mistakes, in part because there's pressure to act quickly. It's much harder to keep the site clean when answers get posted where they shouldn't be. Closing questions prevents answers from coming in. > there is no way the asker can understand what the similarity is from their knowledge point of view (or why the linked duplicate question is the same question). I try to leave a comment to explain the connection when it isn't obvious. (Another common thing that happens is that the problem someone wants to solve involves an obvious two- or three-step procedure, and each step is a matter of fundamental technique that's already been explained countless times.) But overall, it isn't our goal to teach. We answer very simple questions, and very difficult questions; but we aren't designed to teach. Sometimes it's hard to ask a simple question, because you have to figure out what the question is first. It's unfortunate that people who need the question answered often don't have that skill. But if we have a high quality version of that question already, we can direct people there. Sometimes the linked duplicate isn't the best choice. You can help by finding and promoting a better choice - on the meta site and in the chat rooms. You can also help by editing common duplicate targets - both questions and answers - so that it becomes more clear to people who would actually have the question, that they're in the right place (and so that the information in answers is more readily applicable to them).

moring 1y ago