It's helpful to go a bit further for these corrections. What's the reason not to use "uncorrelated" here?

In ordinary English, "uncorrelated" is indeed used for this (and a host of other things, because ordinary English is very vague). The problem is that it means something else in probability theory, namely the much weaker statement E(a-E(a)) E(b-E(b)) = E((a-E(a)(b-E(b)), which is implied by independence (p(a,b) = p(a)p(b)), but not does not imply independence. If we want to speak to those who know some probability theory, this clash of meaning is a problem. If we want to educate those who don't know probability theory to understand the literature and be able to talk with those who do know probability theory, this is also a problem.

(Note too that uncorrelatedness is only invariant under affine remappings (X and Y chosen as the coordinates of a random point on the unit circle are uncorrelated. X^2 and Y^2 are perfectly correlated. Nor does correlated directly make any sense for non-numerical variables (though you could probably lift to the simplex and use homogeneous coordinates to get a reasonable meaning).)

My sense is not that Eliezer is insulting, so much as mainstream (and neurotypical) discourse treats far too much as insulting.  Might be a distinction without a difference in practice, but "insulting" is relative to viewpoint and context, not an objective matter.

For a trivial example, see any discussion about utility monsters.

Yet here we have a high-quality and very well received post about politics, and of the ~70 comments only one appears to have been mindkilled.

Being about political mechanisms rather than object-level disagreements helps a lot.  Even though political mechanisms is an object-level disagreement for monarchists vs republicans.

This is one of those "The future is already here — it's just not evenly distributed" situations.  Training is hard and expensive.  Using is not.  Whether you need to retrain for a given target is an architectural decision -- it does make it harder to train (but sublinearly in targets).

A spelling mistake like the wrongly placed comma "please ,I was" is also unlikely for any attack that's sophisticated enough to be a deepfake attempt.

I agree with the other points, but not this:  sophistication is not a scalar.  It's quite possible to have access to sophisticated tools (which replicate and scale easily), but be sloppy or bad at English orthography (and not realize it).  

I don't think this is useful evidence for deep-fake scam over video-replay scam or vice-versa, but it could easily be evidence for either scam over actual help attempt.  It depends entirely on how out of character such a misplaced comma would be for this particular friend.

sociopaths by the clinical definition make about 1-4% of population.

smart sociopaths make maybe 0.1% of the population

Are you asserting that "smart" is top decile to 2.5%, or that sociopathy is correlated to intelligence?

I'd consider a sigma away from the mean to be smart, so 0.3-1.3%.

A video of Daniel Dennett giving an excellent talk on free will at the Santa Fe Institute: https://www.youtube.com/watch?v=wGPIzSe5cAU It largely follows the general Less Wrong consensus, but dives into how this construction is useful in the punishment and moral agent contexts more than I've seen developed here.

That does look right.

Altering the structure of divorce alters the payoff-matrix for behaviors inside the marriage itself.