Vote in the LessWrong review! (LW 2022 Review voting phase)

post by habryka (habryka4) · 2024-01-17T07:22:17.921Z · LW · GW · 9 comments

Contents

  Who Can Vote?
  How do I vote?
    Okay, but I am a nerdy LessWronger and I like to know how things work in detail. How does this work under the hood?
  Happy Voting
None
9 comments

Click here to begin voting. Click here [LW · GW] for a general overview of the 2022 review.

The final phase of the 2022 review is upon us! We’ve spent 2 weeks nominating posts, and a month reviewing them. And today, we begin the final vote!

The LessWrong Review has two major goals. 

Voting sends a signal about which posts were most important. This then feeds into various compilation and distillation efforts. By default we will create a sequence of top posts, but we are also looking into other curation efforts (like ebooks and audiobooks and maybe print-on-demand books).

Voting [? · GW] started yesterday (January 15th) and continues through the end of January. 

You can see the results of previous votes on the Best Of page: https://www.lesswrong.com/bestoflesswrong [? · GW

Who Can Vote?

All users registered before 2021 can vote. The LessWrong curation team will weight the votes of users with 1000+ karma more highly when assembling sequences, books, or other projects.

How do I vote?

Go to the Review Voting page [? · GW]. There, you'll see posts that you haven't yet voted on sorted first, with posts you've previously given a karma-vote to sorted to the very top (they have a green stripe along the left side. Strong upvotes have a darker green).

The posts will include the reviews that got written about that post. This is intended to help you make an informed vote. You can read reviews that look interesting or highly upvoted to get a better sense of how the post held up.

You vote by clicking buttons that assign a post a score. A score of 1 means roughly "this post was good." A score of 4 means "this post was quite important". A score of 9 means "this post was extremely important." A vote of 0 means "I don't have a strong opinion."

(A vote of "4" is 4x as strong as a vote of "1", a vote of "9" is 9x)

If you intuitively mark posts as "good/important/extremely-important", you'll probably do fine expressing your votes.

Okay, but I am a nerdy LessWronger and I like to know how things work in detail. How does this work under the hood?

The system is built on top of a quadratic voting system [LW · GW]. In quadratic voting, you have a limited number of points to spend. You can vote things more strongly, but the cost of a vote increases at a quadratic rate. So:

You have 500 points to spend. If you spend more than 500 points, your votes get proportionately downweighted[2] [LW(p) · GW(p)], which looks like this:

It's up to you whether to vote a lot of things at the strongest setting (and have the votes count less), or to limit yourself so that your strongest votes count for a full 9 points.

Happy Voting

I am excited to see how the vote shakes out. There were a lot of highly impactful posts written in 2022, and I think it will be good to have a bunch of people spend more serious time evaluating those posts with some benefit of hindsight and enough time for fads to die down. 

If you have any questions, bug reports, mild irritations or suggestions on UI improvements, comment or ping us on Intercom!

9 comments

Comments sorted by top scores.

comment by MondSemmel · 2024-01-17T10:40:03.585Z · LW(p) · GW(p)

Thanks for doing this event every year!

Since I didn't see it mentioned in the post, I'll reiterate my concern [LW(p) · GW(p)] that if AI stuff is not considered separately from non-AI stuff, then votes on AI stuff might swamp everything else:

For clarity, are the review & post results going to be separated into AI and non-AI stuff again, like they IIRC were in some previous year?

...

Overall my attitude is like, if the top 7 posts are all AI posts, that's not because they're necessarily better than the best non-AI posts, but rather because AI has been The Topic since 2022, plus the readership has dramatically shifted towards AI content. At which point we might as well declare LW to be a full-time AI site and consider all the rest to be mere hobbyist content =(. Such a ranking outcome would disincentivize authors from writing about the latter. Better to split the ranking into two top-25s or something.

Also, if the rankings are not split up, then if one only visits LW for AI or non-AI content, that gives an annoying strategic incentive to review-downvote all the other content. That doesn't occur if the rankings are separate.

To elaborate on that last point: Due to how quadratic voting works, if you want to maximize the effect of your votes on the relative rankings of a lot of posts (rather than for a few posts in particular), then the optimal voting pattern is to upvote all posts of the category you like +1, and downvote all posts of the disliked category by -1. (Net effect: a delta of +2 vote review karma on 250 posts of one's choosing.) Two natural candidates for this are AI vs. non-AI stuff. But if these categories are ranked separately, then the incentive for strategic voting disappears.

And besides the strategic voting stuff, another issue is that much of the AI stuff is very technical, so I often feel like I can't make an informed vote one way or another.

Replies from: habryka4, kave
comment by habryka (habryka4) · 2024-01-23T19:00:47.373Z · LW(p) · GW(p)

We'll probably continue separating AI stuff from non-AI stuff within the top 50 when we present the context to readers. If the majority of top 50 stuff is all AI, then we might also look further down the list for non-AI posts, though I want to avoid skewing the results of the votes too much (so very likely wouldn't go down more than to place 75 or so).

comment by kave · 2024-01-23T19:04:48.359Z · LW(p) · GW(p)

No firm decisions made yet, but I think it's fairly likely that we'll do an AI and non-AI separation. If, for example, 40 of the top posts were AI, I think we would very likely do a top-25 non-AI posts or something like that.

comment by Rafael Harth (sil-ver) · 2024-01-24T11:41:03.451Z · LW(p) · GW(p)

When I read the point thing without having read this post first, my first thought was "wait, voting costs karma?" and then "hm, that's an interesting system, I'll have to reconsider what I give +9 to!"

I can see a lot of reasons why such a system would not be good, like people having different amounts of karma, and even if we adjust somehow, people care differently about their karma, and also it may just not be wise to have voting be punitive. But I'm still intrigued by the idea of voting that has a real cost, and how that would change what people do, even if such a system probably wouldn't work.

comment by MondSemmel · 2024-01-31T14:26:08.084Z · LW(p) · GW(p)

Feedback: I wanted to adjust my LW Review votes on the final day (I had only distributed ~300-ish points so far), but the review is considered over at Jan 31st 15:21 CET (= 6:21 am PST), even though the review page said the review would last until Jan 31st. Is that intentional?

In any case, I've gotten frequently confused with these deadlines, so I request that they be posted incl. timestamps and timezones from now on.

Replies from: habryka4
comment by habryka (habryka4) · 2024-01-31T15:25:26.829Z · LW(p) · GW(p)

Oops, I think that's fair. I adjusted the period of the vote by 24 hours. Agree that the current deadline is confusing.

comment by Measure · 2024-01-18T18:58:36.914Z · LW(p) · GW(p)

"Sort by review vote" isn't working for me. There is no apparent rhyme or reason to the order, and it certainly isn't sorting by my review vote. (EDIT: Refreshing the page fixed this, but I would expect it to fully update when I change the sort mode or direction without needing a refresh.)

Additionally, "Sort by karma vote" seems to only care whether I voted on a post without caring about the strength or direction of the vote.

comment by romeostevensit · 2024-01-18T05:30:32.903Z · LW(p) · GW(p)

Review Voting page.

this links to the 2020 review

Replies from: habryka4
comment by habryka (habryka4) · 2024-01-18T05:31:06.999Z · LW(p) · GW(p)

Oops, fixed