As you can easily see when you visit LessWrong.com, the site transfer is complete, and we are now on the new URL. The data import will run for the next few hours, and by tomorrow morning everything should be complete.
Let us know if you run into any problems, find any broken links, or anything of that sort!
Alright!
Profiles are still kind of broken in that Markdown doesn't work, so my profile has lost functionality compared on on old LW.
Q: Is stuff that was posted on old LW after the import to LesserWrong being merged in as well?
Individual comment links still don't work great, due to relying on anchors; my browser "misses" a lot and I have to go back to the URL bar and hit enter again. A proper system of individual pages for comments like Reddit/old LW would be better, I think.
Yep, we imported all content that was posted to the old LW (we created a new database dump at 6PM yesterday). Let us know if you notice anything missing.
Re anchors: I think we should improve the way we do comment links. I am really not a huge fan of how Reddit does it, since I never feel like I have enough context, and always feel a bit lost on the comment page, but am also not happy with our current solution. Happy about more suggestions for this.
Yep, we imported all content that was posted to the old LW (we created a new database dump at 6PM yesterday). Let us know if you notice anything missing.
Great, thanks!
Re anchors: I think we should improve the way we do comment links. I am really not a huge fan of how Reddit does it, since I never feel like I have enough context, and always feel a bit lost on the comment page, but am also not happy with our current solution. Happy about more suggestions for this.
I mean, that's what the context option is for. Old LW had the built-in add-one-context link, too, which reddit doesn't, although adding one context at a time is pretty slow (Reddit of course has its context links that immediately provide 3 context, but doesn't have these on the comment pages themselves). Personally I think "link to the comment, let the reader handle context-fetching" is fine. If you really think people are confused you can make the "add context" or "parent" links more prominent.
Edit: Sorry, this is unrelated but I just noticed it editing this comment. Apparently when I post a comment, any included Markdown is silently converted to equivalent-but-possibly-different Markdown. (E.g., I used astersisks to emphasize something, and they were converted to underscores.) Can I ask that this not be done? That those of us using the Markdown editor have our actual text unaltered when we go to edit it later? Thank you!
Edit: Sorry, this is unrelated but I just noticed it editing this comment. Apparently when I post a comment, any included Markdown is silently converted to equivalent-but-possibly-different Markdown. (E.g., I used astersisks to emphasize something, and they were converted to underscores.) Can I ask that this not be done? That those of us using the Markdown editor have our actual text unaltered when we go to edit it later? Thank you!
FYI, GreaterWrong does exactly what you ask (i.e., GW caches your actual entered raw text, and retrieves that cached text when you later edit your comment/post—regardless of how that entered text has been transformed for display by Less Wrong).
Apparently when I post a comment, any included Markdown is silently converted to equivalent-but-possibly-different Markdown. (E.g., I used astersisks to emphasize something, and they were converted to underscores.) Can I ask that this not be done?
I'm assuming this happens when using the pure-markdown editor? If not, consider switching to the pure markdown editor if you haven't already (click 'edit profile'). If so, I agree this is suboptimal but we probably won't get to it for awhile (it involves digging into the gritty details of the markdown/html/draftjs converters)
(Among other things, simply making it so it's _possible_ to use markdown on most text fields such as the User Profile is a fair amount of work that seems more urgent)
I'm assuming this happens when using the pure-markdown editor?
Yes, sorry, forgot to state that explicitly, that's what I'm doing and it does indeed still occur.
Hit a broken link: https://www.lesswrong.com/user/Eliezer_Yudkowsky/submitted/?count=117&before=t3_17j. This is from a page of chronological post history, clicking on the "Prev" link at http://lesswrong.com/user/Eliezer_Yudkowsky/submitted/?count=128&before=t3_14r.
It also appears that there is no way on the new site to browse chronological post history by a specific author. If LessWrong no longer supports this, is there any alternate site for accessing Eliezer's historical posts, or is everything that is not part of a sequence lost? This is an unfortunate step, as Eliezer's writings were a valuable resource for helping people learn rationality, and limiting access to them seems like a poor idea. Please restore some way to browse post history chronologically, preferably in a way where specific time frames can be loaded.
Did you try https://www.greaterwrong.com/users/eliezer_yudkowsky?show=posts ? I mean, I'm not sure how reliable the whole "import" deal is, but the bulk of LW1 content was actually preserved quite nicely.
Yep, greaterwrong a pretty decent solution. You can also just go on Eliezer's profile and click Load More, which should give you everything in chronological order.
The above link is one of those highly specialized links I am unsure we can make work in the exact way it used to, but we should definitely make the /submitted/ link work so you are forwarded to a list of all of the author's posts (or at least their profile),
Thank you totallybogus, the greaterwrong site is exactly what I was looking for. It might be a good idea to have this site linked somewhere from profile pages so users can see how to reach these archives.
Habryka, the load more option does not provide any practical way of getting to stuff older than a few dozen posts, and does not have any way to get a link to a given range of posts to allow you to return without again trying to load the entire history a few posts at a time through the load more option. It would be a very nice addition to either provide a way to search by date range, or a link to the greaterwrong site that does provide easy access to older posts.
Personally, I think the greaterwrong site (or rather, something functionally equivalent to it) should be the main or only way LW2 is accessed via the web. The technical solutions found in Lesser Wrong (including Meteor and Vulcan) may be appropriate in an actual PWA (Progressive Web App, basically a bunch of JS and 'style/formatting templates' that get installed on your device/computer as a proper, well-defined application and then interact with the remote site via a web-based API of some sort) but they have no advantages to speak of in a traditional website.
We are pretty close to being a proper Progressive Web App, and I definitely see this as the long-term goal of our current design.
There is no reason to reload the whole LW application on every visit, but with Application Storage being phased out as a web-technology, and Service Workers not yet being fully available on Safari and Edge, there isn't yet a great way to give everyone a good experience with a progressive web app. But I expect this will change within the coming year, with safari and iOS support being in the late stages of the beta. So there is a reason behind the design choices we are making.
Two post-migration issues:
Login for legacy LW1 users now seems to be totally broken: a user error “LW1 user detected, legacy salt attached” shows up in the JS console (and appears immediately on GreaterWrong) but nothing further happens.
The third party site GreaterWrong doesn’t seem to be showing updated karma scores or recent (newly-imported) LW1 comments in its user profiles, although these do appear on LesserWrong.
1. Interesting. We didn't touch any of the passwords of accounts that we imported last time (yet). It does seem that we made the error messages in the LoginForm worse by replacing "Unknown Error" with nothing at all. Were you able to login with a password before, but are now no longer able to?
2. That's interesting. My guess is GreaterWrong is doing some kind of caching here and that hasn't updated yet? Happy to help in any way I can if the error is on our side.
Were you able to login with a password before, but are now no longer able to?
Yup. (Obviously I'm now on a different user account) FWIW, it looks like the issue might be on the client side. It's hard to tell though. (The user error showing up on GW is as expected, by the way; they never implemented the LW1 login 'hack' in the first place.)
My guess is GreaterWrong is doing some kind of cashing here and that hasn’t updated yet?
The thing is that 'LW2' comments always seem to show up right away. So that's weird.
Hmm, I can definitely still log in with my two legacy accounts. I would appreciate it if you could ping us on Intercom and help us figure out what might be going wrong.
Could it be browser-dependent? (I'm on up-to-date FF52esr right now.)
Edit: Also, I have now verified that the error object “LW1 user detected, legacy salt attached” appears twice in the JS console (even for a single login submission), with different JS source references (although the data in the error object itself seem to be identical). So it may be that a LW1-style login is being attempted but is failing, either because some password data got corrupted on the LW2 side, or due to some client-side issue.
I've fixed the problem with imported comments not appearing on user pages (for some reason they appear when requesting "allRecentComments" but not "postCommentsNew"), but the user karma totals still don't seem to be correct, even on LW2.
Many of the old posts have unthreaded comments. So maybe it would be good to allow an option to sort comments "oldest first", since this is the most natural way of reading unthreaded comments? In fact, maybe it should be the default comment sorting for such old posts? I seem to recall that this was the case on the old LW site even though newer posts there had comments sorted by karma.
Also, it is a little awkward that the karma on old comments is (old upvote/downvote count, with each vote worth 1 point) + (new upvote/downvote count, with vote weight determined by user karma). It means that when I upvote an old comment it suddenly jumps e.g. from 1 to 4 karma, higher than anything around it. Though I don't know if there is anything reasonable to do about this.
Ah, you're right. I think it's still decently readable with the "most recent" sorting algorithm, but I do think we should also provide an "oldest first" sort.
Now I am noticing a phenomenon with upvoting where I accidentally double-click on an upvote, and it downvotes the comment instead. When I press the upvote button more times then the comment karma seems to behave randomly but eventually goes far into the negatives. I guess this must be designed to guard against spam upvotes or DoS attacks or something like that? Anyway, I am frustrated that when I wanted to upvote a comment I seem to have ended up downvoting it instead, and there doesn't seem to be any way to change it.
By the way, is there a better place to report issues like this than just writing comments in threads like these?
I also noticed this today for the first time. This is not intentional, and I am not really sure what's causing it. We will have to look into that.
Github is usually the best place to report issues: https://github.com/Discordius/Lesswrong2/tree/devel
Some old links to non-post content also seem to be not redirected: https://www.lesswrong.com/about/
Two issues:
1. Yeah, I am not sure exactly what we should do with the wiki, and how we should relate to it. In the long-run I want to properly revive it again and redesign it, but that's quite a bit away. Open to suggestions from people.
2. Thanks for noticing that! I will fix that right now. [Update: Will take a bit longer than expected, but should get around to it today]
Well, if you don't want to link directly to it yet, you could have a page dedicated to the broader Rationalsphere which links to everything outside of the main site, such as the Wiki, Slack, IRC, Rationalist Tumblr, ect. That way people would at least know that these things exist.
Hey there, I don't know if it happened due to this change, but some LW 2.0 content is no longer accessible to me. Specifically, Kaj linked to this post on LW 2.0 (called "Idea Inoculation + Inferential Distance"):
https://www.lesserwrong.com/posts/aYX6s8SYuTNaM2jh3/idea-inoculation-inferential-distance
-> When I follow the link, I get a "Sorry, we couldn't find what you were looking for."
But when I search LW 2.0 for "inoculation", I still see the post and the comments; I just cannot click on the provided links because I get the same error message.
Looks like the post author is going Roko and wiping everything he/she ever wrote from the site - I have no other explanation of what's happening. (To be clear, that's of course within their rights as someone participating in an informal social group! It certainly isn't that much of an inconvenience to the broader userbase.)
Oh, I expect that he'll be fine, but he's clearly decided that he doesn't want to be a part of LW2 after all. It is striking how similar this is to the Roko case though - what people don't really understand today is how silly that whole episode in LW1 history had made him look, and how much of a perceived loss of 'face' that must have been for him.
I think that we're dealing with something very similar here (if perhaps not on the same scale) and I do have to wonder what this says about our broader attitude to newcomers, mild dissidents and the like. How can we ensure that people as far as practicable will feel genuinely welcomed here, and not like they're constantly being judged for the correctness and properness of their 'rationality', or placed under an uncomfortable 'spotlight'?
(I do have some simple ideas along these lines that might be especially appropriate for LW2, and I've even hinted at them before, but I expect them to be a bit contentious so I'd rather be sure that people are actually interested...)
I'm not sure if this is known to you, but I realize it may not be obvious to other people (and it's not a secret, but not widely advertised either): "Conor" was someone's alt account, not actually a newcomer. (His main account has also been wiped.) I don't think it would be fair to call him a 'dissident' either. 'Placed under an uncomfortable spotlight' does seem closer, although I can't quite fathom the reaction unless there were comments I didn't see (perhaps ones that GW didn't have time to mirror, between their creation and their final removal.)
FWIW, I didn't necessarily intend the term "dissident" in an especially negative sense, or even with any real negative connotation. I literally mean "someone who disagrees or dissents, one who separates themself from the established religion; a dissenter." It was also meant to highlight the fact that there are clearly a lot of people like that, as a necessary consequence of LW's overall nature as a remarkably developed 'memeplex' (just like most real-world religions and perhaps political ideologies).
I'm no expert on reading people, and have never met Conor or interacted with him much. But if you'd asked me in advance whether Conor would delete his entire posting history, upon deciding LW2 wasn't for him, I would have said that seemed very unlikely.
And combined with other recent events which I'm not sure how public they are, I worry that something more is happening.
(I do have a minor good sign that the virtue of silence compels me not to reveal.)
But if you'd asked me in advance whether Conor would delete his entire posting history, upon deciding LW2 wasn't for him, I would have said that seemed very unlikely.
Conor said, several times, that this was the strategy he would take if he decided that LW2 wasn't for him. (I would link you to those comments, but...)
Huh. Okay. Score one predictive failure.
I'm glad he's presumably okay. Also, saddened to have lost him and his writings.
I would also have said that it seemed quite unlikely, but my perception is that he's had ample opportunity to clarify and has not done so, which leaves me feeling that 'gone Roko' is the most likely conclusion available to me. (I have definitely seen confirmation that he is still around and communicating with others outside of this site, as of today, and is not radiating obvious distress in so doing.)
I will go ahead and confirm that the author has removed his LW posts. I haven't had a chance to talk with them about it in detail yet and so don't want to speculate further.
Not really on topic, but there's some blatantly obvious spam in the "All posts" view right now. How should we report this sort of stuff in the future?
I guess with better Google ranking comes more spam. We will probably implement some obvious spam classification system, and until then just downvoting should be enough. We might make it so that authors above a certain karma level can mark content as spam and have it be automatically hidden.
Please make the "mark as spam" action reversible.
I am saying this because on SSC there are often comments like "oops, didn't want to report this comment, clicked the wrong button by accident, no way to undo it".
Major problem alert! Attempts to access http://lesswrong.com/ result in a timeout. If you go to http://www.lesswrong.com/ instead, it works fine. However, I totally thought the site had just been down these past few days. This seems to warrant immediate fixing.
This is a DNS cache issue, and I don't think there is super much we can do about that. Either your local machine, or whatever your DNS server is, has the IP address of the old lesswrong.com server cached under the lesswrong.com name, and so when you enter the old URL it tries to access the old server. This will automatically fix itself as the DNS cache of people's machines expires. We changed the DNS about 5 days ago (before that Trike's server manually forwarded to our URL), so this should be fixed soon.
Hold on, I'm confused here -- www.lesswrong.com is a subdomain of lesswrong.com, so how would that occur here? Shouldn't those initially point to the same place to be distinguished later?
Bare domains are messy, and do some weird stuff. You can't actually easily forward bare domains, and doing so requires some hacky stuff (which in our case is dealt with by SimpleDNS, I think, though this is all still Trike Apps' territory, so I don't know the details that well).
We changed the DNS of both lesswrong.com and www.lesswrong.com, but they might still have had different cache-expiration dates on your machine, so they don't necessarily need to change at the same time.
Broken link alert! Was trying to find a particular old post (could we get a search page that allows to see more at once, btw, like the old search page?) and the following links it turned up did not work:
https://www.lesswrong.com/posts/kQsHD9nt3pvmqPJAh/the-comedy-of-behaviorism https://www.lesswrong.com/posts/jnjJywmtY6MpTmE4g/the-comedy-of-behaviorism
(Not sure why I got two links for what appear to be the same post, but that's another matter.)
Ah, that's just our search integration being broken. The data import seems to have messed up our search, and I am working on fixing that.
Another issue: I am still logged on on lesserwrong.com, but when I use my saved password from lesswrong.com to login to the site, nothing happens - I neither get logged in, nor do I get an error message.
(Actually, after pressing the "Sign in" button for several times in rapid succession, I get an "Unknown error" message, but not before.)
EDIT: In the original migration announcement from a few days back, the OP linked to the github issues page here: https://github.com/Discordius/Lesswrong2/issues - is it enough to report issues in comments on this post, or do you need reports there?
Hmm, this could be the bug that's seemingly resulting in a number of LW1 ("legacy") accounts being locked out. Did you have a recovery email set on LW1 before the first database import? If you didn't, but perhaps set it later, then make sure that you can request a "reset password" email from https://www.lesswrong.com with that address (don't worry, this won't actually reset your password or anything unless you click the link in the resulting email message). If you see an error because your recovery address can't be found, visit the "edit account" page in your logged in session and resubmit your data there (your email should be there, but resubmit anyway), then retry the "forgot password" feature on lesswrong.com and make sure that it works for you. This should hopefully ensure that you can always upgrade your account, even if you get locked out from LW1 logins.
totallybogus' recommendation of resetting my password worked. But what I'm still confused about is that there is no website feedback for submitting a wrong password. If you try to login as a user that doesn't exist, you get an immediate (though tiny) error message saying "User not found". In contrast, if the user does exist and you enter an arbitrary (and wrong) password, you get no feedback whatsoever.
In fact, that may have been all that happened in my case - maybe my account had been successfully migrated already, so submitting my old LW1.0 password didn't match my new LW2.0 password, and therefore the login failed; but because I got no website feedback, I had no way of telling the difference.
All that said, I'm aware that allowing unlimited login attempts with arbitrary passwords would constitute a serious security risk (one you've undoubtedly already taken into account), but the current situation of getting no feedback whatsoever for wrong login attempts is also suboptimal.
Strongly agree. I messed up some of the login error messages myself, and we should fix that very soon.
Some old links (e.g. from googling for things) are broken: https://www.lesswrong.com/lw/o5f/wireheading_done_right_stay_positive_without/
Given that this is a major way people find things (coming from google, including trying to find old things they remember), I'd try to fix this when possible. (Though ignore if this is just related to other people's bugs)
Will lesserwrong.com continue to work or should I expect it to become unavailable at some point in favor of lesswrong.com?
It's my understanding, thought Oliver can of course correct me if I'm wrong, that the canonical domain will be lesswrong.com, and all lesserwrong.com/* links will redirect to lesswrong.com/*, to ensure that any links on the web to lesserwrong.com continue to work.
It's working for me using the password that was created after migrating to LW 2.0, but it looks like it's HTTP-only right now. Will SSL/TLS be set up at some point?
Yep, I just activated HTTPS/SSL! Should be active within the next 5 minutes or so (we needed to request a new certificate, so that took a bit).
Ah, alas. It broke some of our cross-site-scrypting protection. Deactivating it for another 10 minutes or so, but should be up very soon. [Edit edit: Fixed now]
Broken link:
http://lesswrong.com/lw/2l0/should_i_believe_what_the_siai_claims/2f14
Expected behavior: You can see the comment, a la archive.org:
https://web.archive.org/web/20170424155218/http://lesswrong.com/lw/2l0/should_i_believe_what_the_siai_claims/2f14
(do make sure you hit the '+')
Actual behavior: You can't see the comment on the page unless you click "show more" comments. Click "show more," and the page reloads, and it zooms down to see his comment. Given that lesswrong.com/{...}/2f14 is a direct link to that comment, it should show that comment.
Were you able to login with a password before, but are now no longer able to?
Yup. For what it's worth, ISTM that this is likely a client issue, although it's hard to tell.
My guess is GreaterWrong is doing some kind of cashing here and that hasn’t updated yet? Happy to help in any way I can if the error is on our side.
The thing is that new comments of the "LW2" sort always seem to show up immediately, so that's a bit weird.
Could it be browser-dependent? (I'm on up-to-date FF52esr right now.)
Edit: Also, I have now verified that the error object “LW1 user detected, legacy salt attached” appears twice in the JS console (even for a single login submission), with different JS source references (although the data in the error object itself seem to be identical). So it may be that a LW1-style login is being attempted but is failing, either because some password data got corrupted on the LW2 side, or due to some client-side issue.