LessWrong 2022 Review: Top 5 Posts
24
414
4.2K
resolved Feb 3
Review posted to LW
Dec 5
Scores become visible
Dec 18
Resolved
YES
AGI Ruin: A List of Lethalities
Resolved
YES
Where I agree and disagree with Eliezer
Resolved
YES
Let’s think about slowing down AI
Resolved
YES
Simulators
Resolved
YES
MIRI announces new "Death With Dignity" strategy
Resolved
NO
Six Dimensions of Operational Adequacy in AGI Projects
Resolved
NO
Counterarguments to the basic AI x-risk case
Resolved
NO
Epistemic Legibility
Resolved
NO
Reward is not the optimization target
Resolved
NO
It Looks Like You're Trying To Take Over The World
Resolved
NO
What an actually pessimistic containment strategy looks like
Resolved
NO
DeepMind alignment team opinions on AGI ruin arguments
Resolved
NO
Staring Into the Abyss As a Core Life Skill
Resolved
NO
Luck based medicine: my resentful story of becoming a medical miracle
Resolved
NO
The Redaction Machine
Resolved
NO
(My understanding of) What Everyone in Technical Alignment is Doing and Why
Resolved
NO
chinchilla's wild implications
Resolved
NO
A Mechanistic Interpretability Analysis of Grokking
Resolved
NO
Lies Told To Children
Resolved
NO
Accounting For College Costs

Edit: the review is now open here.

Every year LessWrong does a review of one of the previous years where people write reviews of old posts and vote on which ones stood the test of time. Starting sometime in December 2023, there should be a review of posts from 2022. In that review, which five posts will be voted the best (by 1000+ karma users*)?

If the format is the same as last year, then after final voting ends, I will check the review voting page, sort by "Final Vote Total (1000+ karma users)", and choose the first five posts listed. (this link does that for last year's review.) I will resolve the options for those posts to YES and all other options to NO.

If the format is sufficiently different that the above method doesn't work, I will try to find another reasonable method of determining the top 5. If the review doesn't have voting, or if the review doesn't happen, every option resolves N/A. Edit: the format appears to be the same as last year.

This page shows posts eligible for a 2022 review. I've started this market with the top 20 posts by karma, but feel free to add eligible posts that you think might win. Last year one of the top 5 posts in the review was only #43 in the karma ranking.

*I'm limiting it to the 1000+ karma user vote to limit the potential for bettors on this market to mess with the results of the review. If you are a 1000+ karma user, pretty please don't change your votes just to win mana.

Get Ṁ200 play money

🏅 Top traders

#NameTotal profit
1Ṁ765
2Ṁ217
3Ṁ96
4Ṁ51
5Ṁ49
Sort by:

Voting has ended, here's my screenshot of the results:

bought Ṁ20 of Where I agree and di... YES

The 1000+ karma user sort is now available on the dashboard, as we've moved into the review phase: https://www.lesswrong.com/reviewVoting/2022?sort=reviewVoteScoreHighKarma .

It currently seems to be sorting by your own vote, though.

sold Ṁ20 of Simulators NO

There is mana to be made in making Sum(prob) = 5 btw, right now it is a bit higher than 5

bought Ṁ10 of Answer #cf51b181cd37 NO

Pro tip: on this type of markets, it's better to lean towards betting on NO (at first). Before I started betting, the sum of probabilities on the markets was over 6.5 or something, even though only five resolve YES. This is a result of creating many markets with initial probabilities 50%.

(I've done some arbitrage so that the probabilities are more meaningful, but there's still a bit of mana on the table.)

Another, extreme example of this phenomenon: this market still has a sum of probabilities ~2.5, even though only one option resolves YES. This is after I (and some others) have done a bunch of arbitrage.

Oh boy assassination market time.

Edit: nvm, I will follow the plea not to manipulate.

More related questions