Compared to just LW, it is more of a hassle. But most comments by academics will indeed be promoted quickly to the AI Alignment Forum, and it’s much less hassle than commenting on an academically published paper (which is close to impossible).
There are barriers to commenting and engaging, but overall it just seems to me that the barriers are much lower than commenting on journal and conference articles, which feels like the relevant reference class.
Hmm, fair point. I feel concerned at how illegible that is though, especially to an academic outsider who wants to engage but lacks context on LW. Eg, I’ve been using AF for a while, and wasn’t aware that comments were regularly promoted from LW. And if we’re talking about perception of the field, I think surface level impressions like this are super important
Yeah, I do think we should just really make things clearer here. Managing the exact extent of the relationship between LW and AIAF is a bit of a delicate balance, but I do think the current communication around it is definitely not ideal.
The AI Alignment Forum currently has a New Comment field at the bottom when a user isn’t logged in that has a submit button. A researcher who visits the alignment forum and types out his comment has a reasonable expectation to when he clicks submit there will be a process that leads to publishing of the comment. That field sets the wrong expecations about what happens when a user who visits the alignment forum registers an account.
If I login then I get shown “You must be approved by an admin to comment on the AI Alignment Forum” and not get told anything about the fact that I might post comments to AI Alignment Forum articles on LW.
This setup is confusing for a new user who understands how journal and conference articles work but who doesn’t understand how the alignment forum works.
I think it would make sense to completely remove the New Comment field when a user isn’t logged in to not give the appearance that it’s easy to comment directly on the alignment forum.
Instead of the New Comment field the sentence “You must be approved by an admin to comment on the AI Alignment Forum” could be displayed when a user isn’t logged in as well. That message might also be worded friendlier and contain a link to a page that explains the setup.
I think it would make sense to completely remove the New Comment field when a user isn’t logged in to not give the appearance that it’s easy to comment directly on the alignment forum.
Yep, sorry, this is the expected behavior and anything else is a bug. It used to be this way, but some changes to some of the commenting code must have broken it. Definitely our bad.
A similar bug—when I go to the AF, the top right says Log In, then has a Sign Up option, and leads me through the standard sign-up process. Given that it’s invite only, seems like it should tell people this, and redirect them to make a LW account?
I do think “non-member AIAF” account is a useful abstraction, just one we haven’t leveraged very much. But, for example, it does allow you to respond to comments on your post on the AIAF, even if you are not a member. So the signup does make sense.
Compared to just LW, it is more of a hassle. But most comments by academics will indeed be promoted quickly to the AI Alignment Forum, and it’s much less hassle than commenting on an academically published paper (which is close to impossible).
There are barriers to commenting and engaging, but overall it just seems to me that the barriers are much lower than commenting on journal and conference articles, which feels like the relevant reference class.
Hmm, fair point. I feel concerned at how illegible that is though, especially to an academic outsider who wants to engage but lacks context on LW. Eg, I’ve been using AF for a while, and wasn’t aware that comments were regularly promoted from LW. And if we’re talking about perception of the field, I think surface level impressions like this are super important
Yeah, I do think we should just really make things clearer here. Managing the exact extent of the relationship between LW and AIAF is a bit of a delicate balance, but I do think the current communication around it is definitely not ideal.
The AI Alignment Forum currently has a New Comment field at the bottom when a user isn’t logged in that has a submit button. A researcher who visits the alignment forum and types out his comment has a reasonable expectation to when he clicks submit there will be a process that leads to publishing of the comment. That field sets the wrong expecations about what happens when a user who visits the alignment forum registers an account.
If I login then I get shown “You must be approved by an admin to comment on the AI Alignment Forum” and not get told anything about the fact that I might post comments to AI Alignment Forum articles on LW.
This setup is confusing for a new user who understands how journal and conference articles work but who doesn’t understand how the alignment forum works.
I think it would make sense to completely remove the New Comment field when a user isn’t logged in to not give the appearance that it’s easy to comment directly on the alignment forum.
Instead of the New Comment field the sentence “You must be approved by an admin to comment on the AI Alignment Forum” could be displayed when a user isn’t logged in as well. That message might also be worded friendlier and contain a link to a page that explains the setup.
Yep, sorry, this is the expected behavior and anything else is a bug. It used to be this way, but some changes to some of the commenting code must have broken it. Definitely our bad.
A similar bug—when I go to the AF, the top right says Log In, then has a Sign Up option, and leads me through the standard sign-up process. Given that it’s invite only, seems like it should tell people this, and redirect them to make a LW account?
I do think “non-member AIAF” account is a useful abstraction, just one we haven’t leveraged very much. But, for example, it does allow you to respond to comments on your post on the AIAF, even if you are not a member. So the signup does make sense.