There is an annoying bug/limitation on the new site, in that users who have posted more than 1000 comments (including yours truly, but this is affecting other prolific commenters to a far greater extent of course) cannot access their full commenting history, albeit they can here on lesswrong.com. (Tested on greaterwrong.com, but I assume that the same limit would apply on lesserwrong.) The comments do exist on the site, attached to discussions (albeit a similar issue may exist, affecting discussions with more than 1000 comments or so). This is annoying because I do want my commenting history to be easily accessible in full, and the same is likely true of many other users.
Oh, huh. I didn’t realize this as a bug. Thanks for pointing it out!
We have plans to allow people better filtering options for comments and posts in general, somewhat similar to what greaterwrong has to show posts sorted by month and year. So that would fix this problem. I am hesitant to allow the serve to return more than 1000 comments on a single graphQL request though, simply because of server-load reasons. So a proper pagination approach would help with this, which would come with the better filtering and sorting I am imagining.
In general, I think it’s very important to make the old content on the site discoverable and findable, and I definitely want to make sure we fix the kinds of bugs you brought up here.
Are you saying that Greater Wrong is currently requesting the whole 1000 comments history when you go to a user page and browse the user history? If so, I think you should get in contact with the Greater Wrong dev(s) and work on a solution that can work with the current pagination on that site. In practice, making sure that full comment history works on Greater Wrong is probably the easiest and quickest way to avert the perception of a regression from what LW1 makes available. Having a “proper” user history with monthly listings, etc. is a nice-to-have of course, but it does not strike me as critical.
I do this because there’s no way to request posts and comments sorted together chronologically with GraphQL. However, if you click the posts or comments tab, the pagination will work correctly for any number of pages.
Indeed, it’s working properly with the show=posts and show=comments URL parameters, and no content seems to be lost. Great news, but that was definitely non-obvious—thanks! (I’d naïvely assumed that if the individual chronological listings were available, that the combined listing would be built by searching for offset_posts and offset_comments such that offset_posts + offset_comments = offset, and the timestamps for the post at offset_posts and comment at offset_comments are as close as possible. Shouldn’t require more than log(N) reqs in the worst case—far less than that typically. But perhaps there’s some snag that makes this approach unworkable!)
Perhaps a notice should be added to the “combined” user pages to the effect that the ‘Posts’ and ‘Comments’ options may be preferable for some uses.
ETA: There seems to be some remaining edge cases around comments on deleted posts, or something like that. In LW1, you get a permlink to the comment from the user page, and can then browse the individual thread. Greater Wrong does not have a notion of viewing a single thread or anything similar, so it tries to get data about the post as a whole, fails, and clicking on the link to the post returns an error page. I have not investigated what LW2 does. This is a very minor issue overall of course. I mention it mostly because I’m wondering how it impacts preservation of e.g. the original discussions about the LW basilisk, which were on a now-deleted page. (Yes, some comments—including e.g. Eliezer’s initial reaction—were totally deleted, but many others were not! And yes, to be quite explicit about it, there are many popular misconceptions about the basilisk, and having these comments preserved is perhaps the one really effective way of addressing them. There was a very clear perception—from people who had actually read the original post! - of how silly it was for Roko to even come up with such an unlikely and contrived scenario, and then bring it up as something that might happen. Roko subsequently wiped his whole presence off the site—posts, comments and all; that he was such a major contributor in the early days is part of why this issue of accessing “comments to something that was deleted” comes up more often than people might expect otherwise!)
I’m not sure that I understand your question. It is obviously an annoyance, and something that high-karma LW1 users may specifically want to be aware of, since they’re far more likely to be affected by it. There are users on the site with far more impressive commenting histories and/or karma scores than mine, and I think the assessment of whether this bug is a ‘blocking’ issue should be left to these users.
Ah, I see. No longer important, but the reason for my question was that I was confused about how your comment related to the topic of the post (retiring LW1.0, calling LW2.0 “lesswrong”, with no “beta” tag).
There is an annoying bug/limitation on the new site, in that users who have posted more than 1000 comments (including yours truly, but this is affecting other prolific commenters to a far greater extent of course) cannot access their full commenting history, albeit they can here on lesswrong.com. (Tested on greaterwrong.com, but I assume that the same limit would apply on lesserwrong.) The comments do exist on the site, attached to discussions (albeit a similar issue may exist, affecting discussions with more than 1000 comments or so). This is annoying because I do want my commenting history to be easily accessible in full, and the same is likely true of many other users.
Oh, huh. I didn’t realize this as a bug. Thanks for pointing it out!
We have plans to allow people better filtering options for comments and posts in general, somewhat similar to what greaterwrong has to show posts sorted by month and year. So that would fix this problem. I am hesitant to allow the serve to return more than 1000 comments on a single graphQL request though, simply because of server-load reasons. So a proper pagination approach would help with this, which would come with the better filtering and sorting I am imagining.
In general, I think it’s very important to make the old content on the site discoverable and findable, and I definitely want to make sure we fix the kinds of bugs you brought up here.
Are you saying that Greater Wrong is currently requesting the whole 1000 comments history when you go to a user page and browse the user history? If so, I think you should get in contact with the Greater Wrong dev(s) and work on a solution that can work with the current pagination on that site. In practice, making sure that full comment history works on Greater Wrong is probably the easiest and quickest way to avert the perception of a regression from what LW1 makes available. Having a “proper” user history with monthly listings, etc. is a nice-to-have of course, but it does not strike me as critical.
I do this because there’s no way to request posts and comments sorted together chronologically with GraphQL. However, if you click the posts or comments tab, the pagination will work correctly for any number of pages.
Indeed, it’s working properly with the
show=postsandshow=commentsURL parameters, and no content seems to be lost. Great news, but that was definitely non-obvious—thanks! (I’d naïvely assumed that if the individual chronological listings were available, that the combined listing would be built by searching for offset_posts and offset_comments such that offset_posts + offset_comments = offset, and the timestamps for the post at offset_posts and comment at offset_comments are as close as possible. Shouldn’t require more than log(N) reqs in the worst case—far less than that typically. But perhaps there’s some snag that makes this approach unworkable!)Perhaps a notice should be added to the “combined” user pages to the effect that the ‘Posts’ and ‘Comments’ options may be preferable for some uses.
ETA: There seems to be some remaining edge cases around comments on deleted posts, or something like that. In LW1, you get a permlink to the comment from the user page, and can then browse the individual thread. Greater Wrong does not have a notion of viewing a single thread or anything similar, so it tries to get data about the post as a whole, fails, and clicking on the link to the post returns an error page. I have not investigated what LW2 does. This is a very minor issue overall of course. I mention it mostly because I’m wondering how it impacts preservation of e.g. the original discussions about the LW basilisk, which were on a now-deleted page.
(Yes, some comments—including e.g. Eliezer’s initial reaction—were totally deleted, but many others were not! And yes, to be quite explicit about it, there are many popular misconceptions about the basilisk, and having these comments preserved is perhaps the one really effective way of addressing them. There was a very clear perception—from people who had actually read the original post! - of how silly it was for Roko to even come up with such an unlikely and contrived scenario, and then bring it up as something that might happen. Roko subsequently wiped his whole presence off the site—posts, comments and all; that he was such a major contributor in the early days is part of why this issue of accessing “comments to something that was deleted” comes up more often than people might expect otherwise!)
GreaterWrong will now attempt to load the comments even when the post fails to load; the comments you mentioned should now be visible here.
And yeah, you’re right that the user page loading could be handled better.
Is that bug a blocker in some way? Are you saying you’re using the old site and want to keep it around until this bug is fixed?
I’m not sure that I understand your question. It is obviously an annoyance, and something that high-karma LW1 users may specifically want to be aware of, since they’re far more likely to be affected by it. There are users on the site with far more impressive commenting histories and/or karma scores than mine, and I think the assessment of whether this bug is a ‘blocking’ issue should be left to these users.
Ah, I see. No longer important, but the reason for my question was that I was confused about how your comment related to the topic of the post (retiring LW1.0, calling LW2.0 “lesswrong”, with no “beta” tag).