Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

LW 2.0 Open Beta Live

15 Vaniver 21 September 2017 01:15AM

The LW 2.0 Open Beta is now live; this means you can create an account, start reading and posting, and tell us what you think.

Four points:

1) In case you're just tuning in, I took up the mantle of revitalizing LW through improving its codebase some time ago, and only made small amounts of progress until Oliver Habryka joined the project and put full-time engineering effort into it. He deserves the credit for the new design, and you can read about his strategic approach here.

2) If you want to use your current LW account on LW2.0, we didn't import the old passwords, and so you'll have to use the reset password functionality. If your LW account isn't tied to a current email, send a PM to habryka on lesswrong and he'll update the user account details on lesserwrong. He's also working on improving the site and sleeping and things like that, so don't expect an immediate response.

3) During the open beta there will be a green message in the bottom right hand corner of the screen. This is called Intercom, and is how you can tell us about issues with the site and ask other questions.

4) The open beta will end with a vote of users with over a thousand karma on whether we should switch the lesswrong.com URL to point to the new code and database. If this succeeds, all the activity from the open beta and the live site will be merged together. If the vote fails, we expect to archive LW until another team comes along to revive it. We currently don't have a date set, but this will be announced a week in advance.

Comment author: IlyaShpitser 17 September 2017 02:05:36AM *  0 points [-]

Vaniver, I sympathize with the desire to automate figuring out who experts are via point systems, but consider that even in academia (with a built-in citation pagerank), people still rely on names. That's evidence about pagerank systems not being great on their own. People game the hell out of citations.


Probably should weigh my opinion of rationality stuff quite low, I am neither a practitioner nor a historian of rationality. I have gotten gradually more pessimistic about the whole project.

Comment author: Vaniver 19 September 2017 06:34:40PM 0 points [-]

Vaniver, I sympathize with the desire to automate figuring out who experts are via point systems

To be clear, in this scheme whether or not someone had access to the expert votes would be set by hand.

Comment author: ozymandias 16 September 2017 01:25:38AM 7 points [-]

Sorry, this might not be clear from the comment, but as a prospective writer I was primarily thinking about the comments on my posts. Even if I avoid culture war stuff in my posts, the comment section might go off on a tangent. (This is particularly a concern for me because of course my social-justice writing is the most well-known, so people might be primed to bring it up.) On my own blog, I tend to ban people who make me feel scared and defensive; if I don't have this capability and people insist on talking about culture-war stuff in the comments of my posts anyway, being on LW 2.0 will probably be unpleasant and aversive enough that I won't want to do it. Of course, I'm just one person and it doesn't make sense to set policy based on luring me in specific; however, I suspect this preference is common enough across political ideologies that having a way to accommodate it would attract more writers.

Comment author: Vaniver 16 September 2017 01:50:37AM 3 points [-]

Got it; I expect the comments to have basically the same rules as the posts, and for you to be able to respond in some low-effort fashion to people derailing posts with culture war (by, say, just flagging a post and then the Sunshine Regiment doing something about it).

Comment author: IlyaShpitser 15 September 2017 08:00:32PM 1 point [-]

Yeah I agree that people need to weigh experts highly. LW pays lipservice to this, but only that -- basically as soon as people have a strong opinion experts get discarded. Started with EY.

Comment author: Vaniver 16 September 2017 01:42:41AM 2 points [-]

My impression of how to do this is to give experts an "as an expert, I..." vote. So you could see that a post has 5 upvotes and a beaker downvote, and say "hmm, the scientist thinks this is bad and other people think it's good."

Multiple flavors lets you separate out different parts of the comment in a way that's meaningfully distinct from the Slashdot-style "everyone can pick a descriptor;" you don't want everyone to be able to say "that's funny," just the comedians.

This works somewhat better than simple vote weighting because it lets people say whether they're doing this as just another reader or 'in their professional capacity;' I want Ilya's votes on stats comments to be very highly weighted and I want his votes on, say, rationality quotes to be weighted roughly like anyone else's.

Of course, this sketch has many problems of its own. As written, I lumped many different forms of expertise into "scientist," and you're trusting the user to vote in the right contexts.

Comment author: Vaniver 16 September 2017 01:41:52AM 2 points [-]

Note that the person who is running the LW survey this year (same as last year) has kindly added some questions about LW 2.0, but isn't directly involved with the project. You can take the survey here.

Comment author: IlyaShpitser 15 September 2017 09:18:40PM 1 point [-]

What's the benefit? Also, what's the harm? (to you)

Comment author: Vaniver 15 September 2017 11:17:13PM 8 points [-]

Main benefits to karma are feedback for writers (both informative and hedonic) and sorting for attention conservation. Main costs are supporting the underlying tech, transparency / explaining the system, and dealing with efforts to game it.

(For example, if we just clicked a radio button and we had eigenkarma, I would be much more optimistic about it. As is, there are other features I would much rather have.)

Comment author: ozymandias 15 September 2017 03:55:59PM *  13 points [-]

Thank you for making this website! It looks really good and like someplace I might want to crosspost to.

If I may make two suggestions:

(1) It doesn't seem clear whether Less Wrong 2.0 will also have a "no politics" norm, but if it doesn't I would really appreciate a "no culture war" tag which alerts the moderators to nuke discussion of race, gender, free speech on college campuses, the latest outrageous thing [insert politician here] did, etc. I think that culture war stuff is salacious enough that people love discussing it in spite of its obvious unimportance, and it would be good to have a way to dissuade that. Personally, I've tended to avoid online rationalist spaces where I can't block people who annoy me, because culture war stuff keeps coming up and when interacting with certain people I get defensive and upset and not in a good frame for discussion at all.

(2) Some inconspicuous way of putting in assorted metadata (content warnings, epistemic statuses, that sort of thing) so that interested people can look at them but they are not taking up the first 500 words of the post.

Comment author: Vaniver 15 September 2017 09:22:40PM 8 points [-]

I expect the norm to be "no culture war" and "no politics" but there to be some flexibility. I don't want to end up with a LW where, say, this SSC post would be banned, and banning discussions of the rationality community that might get uncomfortable seems bad, and so on, but also I don't want to end up with a LW that puts other epistemic standards in front of rationality ones. (One policy we joked about was "no politics, unless you're Scott," and something like allowing people to put it on their personal page but basically never promoting it accomplishes roughly the same thing.)

Comment author: IlyaShpitser 15 September 2017 02:28:41PM *  4 points [-]

(a) Thanks for making the effort!

(b)

"I am currently experimenting with a karma system based on the concept of eigendemocracy by Scott Aaronson, which you can read about here, but which basically boils down to applying Google’s PageRank algorithm to karma allocation."

This won't work, for the same reason PageRank did not work, you can game it by collusion. Communities are excellent at collusion. I think the important thing to do is making toxic people (defined in a socially constructed way as people you don't want around) go away. Ranking posts from best to worst in folks who remain I don't think is that helpful. People will know quality without numbers.

Comment author: Vaniver 15 September 2017 09:15:54PM 5 points [-]

Oli and I disagree somewhat on voting systems. I think you get a huge benefit from doing voting at all, a small benefit from doing simple weighted voting (including not allowing people below ~10 karma to vote), and then there's not left from complicated vote weighting schemes (like eigenkarma or so on). Part of this is because more complicated systems don't necessarily have more complicated gaming mechanics.

There are empirical questions involved; we haven't looked at, for example, the graph of what karma converges to if you use my simplistic vote weighting scheme vs. an eigenkarma scheme, but my expectation is a very high correlation. (I'd be very surprised if it were less than .8, and pretty surprised if it were less than .95.)

I expect the counterfactual questions--"how would Manfred have voted if we were using eigenkarma instead of simple aggregation?"--to not make a huge difference in practice, altho they may make a difference for problem users.

Comment author: Habryka 15 September 2017 11:52:19AM 22 points [-]

All old links will continue working. I've put quite a bit of effort into that, and this was one of the basic design requirements we built the site around.

Comment author: Vaniver 15 September 2017 09:11:38PM 10 points [-]

"Basic design requirements" seems like it's underselling it a bit; this was Rule 0 that would instantly torpedo any plan where it wasn't possible.

It's also worth pointing out that we've already done one DB import (lesserwrong.com has all the old posts/comments/etc. as of May of this year) and will do another DB import of everything that's happened on LW since then, so that LW moving forward will have everything from the main site and the beta branch.

Comment author: gbear605 15 September 2017 03:38:06AM *  4 points [-]

If it doesn’t, we’ll likely turn LW into an archive.

To clarify, does this mean that once the open beta conclude either A) we will switch permanently to the LW 2.0 or B) LW as a whole will close down to new posts/comments?

EDIT: From http://lesswrong.com/r/discussion/lw/pes/lesswrong_20_strategic_overview/ it is clear that this is in fact the case.

Comment author: Vaniver 15 September 2017 05:39:39AM 2 points [-]

Correct.

View more: Next