niplav

318Joined Jun 2020

Comments
68

Great investigation! Now I'm slightly less salty that your post is exclusively cited when it comes to the relation of range and accuracy (though I may still bask in the glory of second-hand citation :-p).

Few users updated their predictions, and updating was not associated with lower Brier scores overall, though there was not enough data to infer much here. Of 9230 updates, 3141 (34%) were performed by the most frequent individual predictor and 4710 (51%) were due to the top 3 most frequent updators.

Matches my experience, though I think Metaculus is slightly better in this regard. Should still give observers pause to think about how suboptimal those platforms are.

The models were:

<1y: 0.9400*Prediction - 0.0154

1-3y: 0.9122*Prediction - 0.1066

3-5y: 0.8927*Prediction - 0.0837

5+y: 0.8587*Prediction - 0.1089

This is super cool!

I wanted to look into whether forecasters appeared to get better over time. For this, I took those forecasters with >100 predictions, and compared their performance on their first 50 predictions to their last 50.

The answer appeared to be "maybe". There was no improvement in Brier scores or over confidence, but it is possible that they may have tried to predict more difficult questions in their later questions

I think that 100 predictions just isn't enough, especially if you're not doing deliberate practice. I think my predictions started getting okay after having experienced ~100 question resolutions, which would imply several hundred predictions. Surprised to hear the reviewer had the opposite opinion!

It should be possible to test this by performing a similar analysis, but looking at predictions made after a certain number of resolutions for that user and checking whether there is an improvement. I think resolutions should be the focus here: You can learn very little from predictions that you don't know the outcome of yet (though I've found it helpful to predict Metaculus with the community prediction hidden and then check against the community). I'm not sure it would be worth the effort to perform this analysis, but I'll put it on my todo list.

For the Metaculus data I could glean less information, as there were fewer questions, and no user level data available.

FWIW Metaculus now makes their user-level data available to researchers if you ask nicely.

Since we now know that 41% of things happen ;-), it'd be interesting to see whether things that are far off happen more rarely (or, in plain english, do questions with longer horizons resolve positively less often?). I don't think you looked into into this here, right?

As for data sources, I've started working on a collection of forecasting datasets, but my funding for that ran out and wasn't renewed :-/ Maybe I'll find a way to finish it.

At least on LessWrong you can move something to drafts and then publish it again, IIRC. Given the underlying infrastructure is the same this should also work on the EA forum?

Writing good Wiki articles is hard, and translating between worldviews even harder. If someone wants to do it, that's cool and I would respect them, but funding people to do it seems odd—"explain X to the ~10k EAs in the world". Surely those fields have texts that can explain themselves?

Small movements (like species with few members, I think[1]) die more quickly, as do younger movements.

Also EA seems to have a quite specific type of person it appeals to & a stronger dependence on current intellectual strands (it did not develop separately in China & the Anglosphere and continental Europe), which seems narrower than socialism/communism/reactionary thought.

I think it's good to worry about EA disappearing or failing in other ways (becoming a cargo-cult shell of its original form, mixing up instrumental and terminal goals, stagnating & disappearing like general semantics &c).


  1. I've tried to find a paper investigating this question, but haven't been successful—anyone got a link? ↩︎

If agents are usually suffering then negative sum games might be good via destructive warfare that removes negentropy, which is one consideration in cooperation-focused agendas for suffering-focused longtermism.

Answer by niplavNov 08, 202220

There is the server schelling.pt which inherits more from the postrat/tpot/tcot community on twitter, but which I've been the main poster on (not admin of) for the last year and a half. It's the schelling point!

I also lend significant credence to a similar view, which I'd summarize as "animals with fewer neurons aren't less sentient, their sensations are just less complex".

I used to think that babies had fewer neurons, but people would not believe babies to be lower on moral patienthood, but I was mistaken: apparently neurogenesis ends early (and adult neurogenesis is still in question):

neurogenesis in humans generally begins around gestational week (GW) 10 and ends around GW 25 with birth about GW 38-40.

However, perhaps we don't just want to go by neuron count: number of synaptic connections seems perhaps just as important.

There's an interesting tension showing itself with the potential areas you point to: Two out of the five are mostly not public goods, and EA has traditionally focused on supplying public goods.

Both life extension and cryonics should be very much in everybody's interest assuming that most people want to live longer and healthier—so if they aren't, they are either very irrational (possible) or it's just not in their interest.

For a successful cryopreservation, you need facilities for storage, liquid nitrogen and staff overseeing the operation. All that costs money. Plastination alleviates some of those costs, and economies of scale would also apply.

And cryonics is expensive: The cryoprotectants currently used alone are nothing to sneer at.

  1. The perfusate has a shelf life of several years when stored in an ordinary refrigerator. Alcor’s purchase price for the ingredients in all 10 2-liter bags of perfusate, including M22, is ~$1,500. The concentration of M22 increases by a factor of 1.67 between bags, except that the last 3 bags have the same terminal concentration. While 10 bags is sufficient to achieve the desired terminal jugular cryoprotectant concentration, 16 bags were prepared for the initial trial (the final 9 bags having the same terminal concentration) to ensure that enough bags were available to achieve terminal jugular cryoprotectant concentration.

A Big Hairy Audacious Goal for Cryonics, Ralph Merkle, 2014

Load More