pja

See also: https://acoup.blog/2022/03/11/collections-nuclear-deterrence-101/

Oh, look a Borg CVE has just landed in my Debian security inbox.

Scratch a libertarian, find a ...

I have currently settled on borg for backups personally, combined with some gruesome ssh hackery to let me do pull backups of external machines using ssh tunnelled sockets back to a backup server which is not reachable from the wider internet. These days I might just use tailscale to set up a VPN & pipe the backup straight over that without all the ssh shenanigans but the system I have works. You can also use borg to talk directly to rsync.net at special nerd rates: https://www.rsync.net/products/borg.html Bring your own support!

NB. Last time I looked at this, borg’s cryptography was somewhat suspect. Not actually broken, but definitely not using best practices. Restic is better, but at the time I was looking restic didn’t compress backups so it was a non-starter for me. These days restic does compression as well so is probably the right default choice. Borg2 has a rewritten encryption layer which supposedly fixes all the problems pointed out by cryptographers with Borg1, but it hasn’t hit a release version yet & is still in beta.

You have just demonstrated that you don’t understand Taleb’s critique. Admittedly, his basic critique is buried in tons of verbiage, but your response here is an irrelevance.

IQ measurements are next to useless on an individual level because a) IQ measurement is terrible & non-repeatable with very large variance between successive tests for any given individual & b) IQ doesn’t measure the thing you actually want, which is task-specific performance: it has terrible correlation with any given task-specific measure, barely rising above “vaguely related”.

At the population-level, IQ suffers from terrible statistical issues, including circularity affecting outcomes (SAT tests in the US are a particular problem), and inter-population differences that make comparisons extremely noisy. The field is also historically full of charlatans who literally made up data out of thin air, even before you start in with the problems with the actual data they drew upon & the stats they applied to it.

Ultimately, It doesn’t matter that you can measure some “factor” and show that there’s a weak correlation with lifetime wealth, or prison likelihood or whatever if that measurement is an otherwise useless one: Using IQ as a measure of an individual is wildly inappropriate. Using it as a population measure is next to useless because of widespread issues with both the input data & the statistical analysis done to torture some kind of correlation out of said data & call the job done.

Finally, when you’ve done all these population level stats on your so-called “g-factor” and squeezed some kind of vague relationship between various groups & your “g-factor” out of the data, what are you doing that /for/? What good do you expect to do in the world with that information? Because the only real-world use seems to be advocating for blocking the immigration or education of specific groups of people, despite the fact that, as has already been pointed out, you cannot use IQ on an individual level because it has extremely poor predictive value at the individual level. Sounds ... kind of racist don’t you think?

The seed oil / sunburn thing is something I saw going around online just this week. Is there /anything/ behind it at all or is it just another chunk of LessWrong science by anecdata?

High-end stats is kind of Taleb’s thing, so he gets to be as insufferable as he likes dunking on IQiots imo.

Yeah, he needs an editor. But the relentless dunking on IQiots is worth the verbiage imo.

This Twitter post from the guy who was synthesising the stuff in the US seems convincing: https://twitter.com/andrewmccalip/status/1689476909208600576

Not a superconductor - the hints of superconductor-like properties were due to a combination of iron contamination in the Pb forming iron fragments which (surprise!) show ferromagnetism & Copper Sulphide which shows a very similarly anomalous temp/resistivity profile (but is not a superconductor).

The most likely outcome (i.e., not a superconductor, lab error due to honest scientists being fooled by their own experiments) seems to be probably the true one.

Here’s another classic non sequitur: https://www.reddit.com/media?url=https%3A%2F%2Fi.redd.it%2Fwayy8oc30ygb1.png

Why does it tend to collapse into pseudo-religious babble when it goes off the rails? I guess it tends to be very repetitive, so maybe the training set has turned religiosity into some kind of attractive basin in the output space? Once in, you can’t get out again.

It’s as if it’s a markov chain with a bigger token space to work from. Oh wait.

Agreed, I read this as saying that the cop would otherwise have got away with it & that this was a /bad/ thing.

Also, the experimenting on babies thing is hilarious.