I’m not sure what to make of it. Regardless of whether the variant does much in humans, I’m surprised that they found it had a big effect on one of the things they decided to measure (in fruitflies, and it isn’t exactly the same variant, but still). I think it’s only very weak evidence of the IQ thing being real, but it’s still surprising to me, so there might be something to be learned here.
kman
Karma: 1,677
I think it suggests that the blindness comes from a different, rarer variant (in the PROM1 gene). If anything that would update me towards thinking the IQ effect, if real, also comes from that rarer variant.
(Edit: the 6 carriers of the PROM1 mutation studied in this paper had an average verbal IQ of 104, which is pretty strong evidence against it giving a huge boost to verbal IQ as reported by the CORD7 paper).
More information: that “CORD7” variant is actually in an IQ GWAS, but showed ~0 effect. Note that I couldn’t rule out the PROM1 variant in the same way since it’s much rarer and not in the GWAS.
Perhaps more relevant to the quoted sentence is that it takes advantage of a very cleverly non-flat prior.
Any prior over an unbounded space needs to be non-flat in this way, or you’d never be able to learn anything. To put it more precisely: if you are assigning nonzero probability to every hypothesis in the space, there will exist some description length L(
Given a flat prior over a vast hypothesis space, even simple tasks require an enormous amount of data.
Solomonoff induction doesn’t require enormous amounts of data unless the thing it’s trying to learn is actually enormously complex.
Changing your preferences wouldn’t affect what you end up experiencing. And if I looked at the causal history of how you ended up with your preferences about what sorts of worlds/experiences you care about, I’d bet it would have a lot to do with what you experienced in the past.
In the short term, sure.
I’m not familiar with the history of people being worried about overpopulation, but I’d guess a lot of these mid 20th century people you refer to were worried about the relative short term based on projections which ended up being wrong? I guess one lesson you could learn from this is that timing can be hard to call even if the endpoint is predictable. Huh, that sounds familiar...
Do these expectations take selection effects into account? I’m also thinking longer term than 60 years. A Malthusian equilibrium is the natural state for a population of organisms to be in. We’re currently out of equilibrium, but the obvious expectation is that we will at some point settle back into a Malthusian equilibrium unless we somehow choose not to or otherwise go extinct.
Given the coming AI compsec apocalypse, now might be the time to switch to Mac from Windows/Linux (or convince your family and friends to do so). My reasoning: Apple’s vertical integration will make it more likely that you get software and firmware patches for vulnerabilities found by frontier compsec models, before the adversaries get their hands on models with sufficient capabilities to exploit those vulnerabilities.
If you are a Linux poweruser and on top of this stuff this might apply less strongly to you. I personally intend to continue using Linux (perhaps against my better judgement), but will likely suggest Mac to family and friends going forwards.
Note: I am a longtime Windows (previously) and Linux (currently) user and have never owned a Mac.
Seems like you’re leaning a lot on the benefit of hindsight, and also looking too much at short term trends? A return to a Malthusian equilibrium is the simple obvious advance prediction.
clearly in retrospect this was extremely incorrect
No? In a counterfactual world where AI wasn’t going to be a thing, this still seems like the default long term outcome if not prevented via coordination (natural selection doesn’t stop). If you mean that it was incorrect because AI ended up being a thing before it became a problem, then sure.
Thanks for the data point!
I’m curious whether 100mg caffeine in pill form would work for you.
A bit off topic, but I have a question about anthropic reasoning that’s been bugging me for a while: what do you make of us not finding ourselves in a glorious transhumanist future? (I know UDT says not to give up on this basis)
Did you feel anything from 200mg PX at all?
In an unpublished blind comparison between PX and CA I found 100mg CA in pill form surprisingly weak compared to what I expected from drinking coffee. I ended up increasing the dose to 200mg for both. I rated the peak effects of 200mg PX somewhat weaker on average, and napped more on the PX days.
In pill form?
How much caffeine had you been consuming prior to that?
What is your theory of how fast government can prevent extinction?
This read to me as a bit inevitabilist. Putting aside for a moment that it might be hard to avoid: do we actually want to live in a world in which anyone can easily find out anything about anyone else?
and I strongly suspect that concerns about terrorism, great power conflict, and small-scale bad actors will be serious enough that these highest priority uses will not be foregone
Can you say more about why you strongly suspect this?
We need smarter+wiser humans actually have a shot at getting ASI right. The current people need to be shut down.