4 ms·

Stack Exchange Moderator Strike

7 points by jlericson 3y ago

IEnjoyTapNinja 3y ago
The main part of the policy seems to be the following: "We've reminded moderators that suspensions (and typically mod messages as well) are for real, verifiable malfeasance only, and should not be enacted on the basis of hunches, guesses, intuition, or unverified heuristics." While I understand that StackOverflow does not want false positives, the only kind of AI generated text that is truly verifiable, is the one that still contains "As an AI language model..."
johnny99k 3y ago
At some point (and it seems we are getting close with chatGPT), we aren't going to be able to reliably tell the difference between AI and non-AI generated content.
- IEnjoyTapNinja 3y ago
  I would argue that we have already passed that point in several areas: https://news.ycombinator.com/item?id=36139349 https://news.ycombinator.com/item?id=36139349 https://news.ycombinator.com/item?id=36003595 https://news.ycombinator.com/item?id=36003595 https://news.ycombinator.com/item?id=34045748 https://news.ycombinator.com/item?id=34045748
donmcronald 3y ago
I think it'll be interesting to see how Stack Exchange holds up against all the AI stuff. On one hand, I go there to find info about things I don't understand, so I'm going to have a hard time spotting hallucinated junk. On the other hand, topic experts normally point out mistakes and differences in opinion and I find the competing viewpoints to be extremely helpful at times. So, I would say the debate and iteration on "the best" solution for problems is the thing I like most about Stack Exchange. The thing I fear the most is having that debate and iteration polluted by AI to the point where the signal to noise ratio becomes too high to be useful.
shagie 3y ago
My crystal ball of unfounded speculation on this says... It's that in the GPT hunting/banning back in December and January they probably hit a lot of false positives. SE lacks a good way for people to complain about moderation (the ‘send feedback’ is a rather poor tool to do so) and things got escalated to where lawyers started getting contacted. SE as a global company can’t attempt to hide behind a “we’re only in the US” and so is now swinging the moderation rules to the other extreme (its always one extreme or the other) since they don’t have sufficient CMs or lawyers to properly address the issues. So now, six months later, the EU digital services act is kicking in hard with the unanswered/unresolved moderation challenges that should have been done in a timely manner.
- jlericson 3y ago
  Yeah... that's not it. I used to handle escalation tickets on SO and we were handling tickets from the EU years before I joined. Also, that's not how any of this works.
  - shagie 3y ago
    It would have then helped for SE corporate to expand a bit more on what constituted misdiagnosing ChatGPT content, how that was determined and how it was determined that certain countries were prone to false positives. The disconnect between regular users capable of flagging (though the "what is an acceptable flag" has likewise gotten more and more restricted too - see also "not an answer" is not the right flag if there was any attempt to answer, even if it was the wrong answer to the wrong question), moderators, and corporate makes trying to have a community run site difficult - especially as the tools that one would use to manage, moderate, and curate it get more and more restricted. The curious part of the "is this it?" with the digital services act is that it was signed into law 6 months ago (November 2022) and has provisions that have a 6 month window and concerns moderation of users and improperly removing their content. Anyways, that was my unfounded speculation. I am still disappointed that incorrect information written by a human isn't more severely moderated or curated on SO and SE sites. If I'm going to get wrong information, might as well get it from a source I need to double check anyways and is in a more consumable format. The value of SO was that "good answers are voted up and rise to the top" which is much less commonly the case anymore.

7 points by jlericson 3y ago

The main part of the policy seems to be the following: "We've reminded moderators that suspensions (and typically mod messages as well) are for real, verifiable malfeasance only, and should not be enacted on the basis of hunches, guesses, intuition, or unverified heuristics." While I understand that StackOverflow does not want false positives, the only kind of AI generated text that is truly verifiable, is the one that still contains "As an AI language model..."

johnny99k 3y ago

At some point (and it seems we are getting close with chatGPT), we aren't going to be able to reliably tell the difference between AI and non-AI generated content.

IEnjoyTapNinja 3y ago
I would argue that we have already passed that point in several areas: https://news.ycombinator.com/item?id=36139349 https://news.ycombinator.com/item?id=36139349 https://news.ycombinator.com/item?id=36003595 https://news.ycombinator.com/item?id=36003595 https://news.ycombinator.com/item?id=34045748 https://news.ycombinator.com/item?id=34045748

donmcronald 3y ago

I think it'll be interesting to see how Stack Exchange holds up against all the AI stuff. On one hand, I go there to find info about things I don't understand, so I'm going to have a hard time spotting hallucinated junk. On the other hand, topic experts normally point out mistakes and differences in opinion and I find the competing viewpoints to be extremely helpful at times. So, I would say the debate and iteration on "the best" solution for problems is the thing I like most about Stack Exchange. The thing I fear the most is having that debate and iteration polluted by AI to the point where the signal to noise ratio becomes too high to be useful.

shagie 3y ago

My crystal ball of unfounded speculation on this says... It's that in the GPT hunting/banning back in December and January they probably hit a lot of false positives. SE lacks a good way for people to complain about moderation (the ‘send feedback’ is a rather poor tool to do so) and things got escalated to where lawyers started getting contacted. SE as a global company can’t attempt to hide behind a “we’re only in the US” and so is now swinging the moderation rules to the other extreme (its always one extreme or the other) since they don’t have sufficient CMs or lawyers to properly address the issues. So now, six months later, the EU digital services act is kicking in hard with the unanswered/unresolved moderation challenges that should have been done in a timely manner.

jlericson 3y ago
Yeah... that's not it. I used to handle escalation tickets on SO and we were handling tickets from the EU years before I joined. Also, that's not how any of this works.
- shagie 3y ago
  It would have then helped for SE corporate to expand a bit more on what constituted misdiagnosing ChatGPT content, how that was determined and how it was determined that certain countries were prone to false positives. The disconnect between regular users capable of flagging (though the "what is an acceptable flag" has likewise gotten more and more restricted too - see also "not an answer" is not the right flag if there was any attempt to answer, even if it was the wrong answer to the wrong question), moderators, and corporate makes trying to have a community run site difficult - especially as the tools that one would use to manage, moderate, and curate it get more and more restricted. The curious part of the "is this it?" with the digital services act is that it was signed into law 6 months ago (November 2022) and has provisions that have a 6 month window and concerns moderation of users and improperly removing their content. Anyways, that was my unfounded speculation. I am still disappointed that incorrect information written by a human isn't more severely moderated or curated on SO and SE sites. If I'm going to get wrong information, might as well get it from a source I need to double check anyways and is in a more consumable format. The value of SO was that "good answers are voted up and rise to the top" which is much less commonly the case anymore.