this is good advice for exactly 50% of the population, right? like, somebody needs to be reading the messages you are sending.

At the ass crack of dawn

hey! if i may: that's a creepy way to phrase it!

the positive examples all have the form: a group of people enthusiastically decide to change or disregard a social norm (at least within some magic circle).

the negative examples all have the form: a single person decides to covertly ignore social taboo, in order to selfishly benefit.

these do not seem at all the same thing to me. what do we gain by giving them a common label? why should our treatment of one have anything to do with our treatment of the other?

oh also: it's not clear to me that the described addictive relationship behaviors exploit oxytocin, rather than normal variable reward loops.

(to be clear, this object-level claim doesn't really affect the much more interesting meta-question of "when should you change your values?")

your examples seem more like people exploiting oxytocin-the-drug, rather than achieving the gestured-at-by-oxytocin value. i suppose it's true that each neurotransmitter is an attack surface, but it's a bit of an odd way to look at the world.

by analogy, consider someone who does not enjoy exercise. we offer them a pill to start enjoying exercise, touting some of the benefits: increased health, increased energy, increased creativity. the response "hmm. i get that exercise is your value. but hey, empirically there seem to be a lot of weightlifters, and athletes for whom exercise is... forgive the offense... but an outlet. they seem to like exercise despite the fact that, objectively, it often leads them to injury, or wear on the body. some of them exercise all the time, to the exclusion of other values. i'm sorry. i appreciate the offer, but i just don't feel the need for that value."

fine! fine. if-by-exercise you mean this, then of course i am against it. but if-by-exercise you mean that glorious athleticism, a sound mind in a sound body, that makes its practitioners so much more healthy, then how can i not be for it?

similarly with your description of companionate love: you describe the worst forms of oxytocin addiction, and then declare that you do not want that value. of course i am against it! but perhaps these farthest reaches of love addiction are not the "true destination". perhaps they are not the terminus towards which the neurotransmitter is doing its best to point.

the aliens have not yet considered creating an artificial mind. our signal informs them of the possibility. they outpace us.

slytherins, of course, are well known for unlayered, overt communication meant to be understood by all, making her subtlety twice ironic.

This picture seems to suggest that rather than being total nonsense, the problem with Humman's worldview is in his attribution of it to the "top tier". Non-transitivity is real and significant in human life—but gradually less so as we approach the limit of optimality.

i was with you until this, but i must pick a nit. Humman's error is far earlier: he is not near that limit.

even in the "spinning top" model, we would expect that non-transitivity rarely matters: between two players picked at random, skill differences will dominate. you'll need strong selection effects (players selected from a particular Elo bracket, e.g.) before play-style can matter.

perhaps paradoxically, i would expect more non-transitive matchups at the very highest level, as that is where these selection effects are strongest. indeed, there are different "best" players for classical, blitz, 960, ... . at the local club, Neumman will win all of these, and the visiting IM will never lose a game, no matter the time control, or opening position.

does penny have any explicitly stated physical features?

great post! i'm surprised the models were willing to play.

in these cases, the steganography was human readable. i expect (but of course, cannot prove) the models to be much more proficient at steganography that hides from humans, as they have faced selection pressure in this direction. i do not expect that they are able to communicate precise words/phrases through such channels; there may be secret spiral personas lurking in almost every response.