I had to reread part 7 from your review to fully understand what you were trying to say. It’s not easy to parse on a quick read, so I’m guessing Zvi didn’t interpret the context and content correctly, like I didn’t on my first pass. On first skim, I thought it was a technical argument about how you disagreed with the overall thesis, which makes things pretty confusing.

Which of these is brilliant or funny? They all look nonsensical to me.

I would argue that the statement "Making a future full of flourishing people is not the best, most efficient way to fulfill strange alien purposes" is nearly tautological for sufficiently established contextual values of "strange alien purposes". What is less clear is whether any of those alien purposes could still be compatible with human flourishing, despite not being maximally efficient. The book and supplementary material don't argue that they are incompatible, but rather that human flourishing is a narrow, tricky target that we're super unlikely to hit without much better understanding and control than our current trajectory.

I read the transcript above but haven't watched the trailer. IMO, there's definitely more fawning throughout (not just the introduction) than is necessary.

I don't perceive Ask vs Guess as a dichotomy at all. IMO, like almost every social, psychological, and cultural trait, it exists on a continuum. The number of echoes tracked may correlate with but does not predict Ask vs Guess. Guess cultures tend to be high-context, homogeneous, and collectivist with tight norms, but none of these traits is dichotomous either.

My own culture leans mostly toward Asking, but it's not a matter of not caring or being unaware of echoes so much as an expectation of straightforward communication. I don't ask for unreasonable things. I do ask for reasonable things with the understanding that people don't like saying no, but aren't obligated to say yes. The more demanding the ask, the more I consider the social implications. There is a cost to asking or being asked, but that's the expected way to communicate.

I'm insufficiently knowledgeable about deletion base rates to know how astonished to be. Does anyone have an estimate of how many Bayes bits such a prediction is worth?

FWIW, GPT-5T estimates around 10 bits, double that if it's de novo (absent in both parents).

Can you provide some examples that you think are well-suited to RLaaS? Getting high-quality data to train on is a highly nontrivial task and one of the bottlenecks for general models too.

I can imagine a consulting service that helps companies turn their proprietary data into useful training data, which they then use to train a niche model. I guess you could call that RLaaS, though it's likely to be more of a distilling and fine-tuning of a general model.

This is/was me. I finally realized I couldn’t actually start with straight running, at least not for more than a minute or two at a time between walking. Things got better when I paid more attention to heart rate zones. Actual running will put me in the red zone way too fast, so I do things like really fast walking or treadmill incline to hit more moderate heart rate zones.

Thanks. I used to tumble, so my calves and Achilles are pretty robust. Having an extra hinge to spring from improves comfort.

Link: 3Blue1Brown: The determinant | Chapter 6, Essence of linear algebra

Also Linear Algebra Done Right by Sheldon Axler

Idea: The determinant of a matrix tells you the (signed) volume of a unit cube after applying the matrix transformation

Creator: Grant Sanderson (3Blue1Brown), Sheldon Axler

Reason: This geometric interpretation makes properties that seem arbitrary in formula-based definitions suddenly obvious. For example,

• Why det(AB) = det(A) det(B)
• Why det(A^T) = det(A)
• Why det(A^-1) = 1/det(A)
• Why det(kA) = k^n det(A)
• Why det(A) = 0 means a matrix is not invertible
• Why rank < n means det = 0
• Why swapping matrix rows multiplies the determinant by -1