2026-07-09

#insomnia research
slept for like 2.5 hours, felt absolutely doomed waking up.
thought the day would be a waste but just forced myself (like i was lifting a weight) to think about interesting problems for some time. and i felt completely fine after that. and this lasted the whole day. i even ran better than yesterday when i slept more.
just pretend like you don't need sleep. ignore everyone's advice. even if you are physiologically doomed or something, which i doubt, ignoring it is empirically way better. heros don't wear oura rings.

neurocognitive poetics
Quantifying the Beauty of Words: A Neurocognitive Poetics Perspective

• they've trained a classifier that can predict if human subjects will call a word beautiful are ugly.
• what are their features?
  • The eight features used included two sublexical features (number of syllables, sonority score) and six lexical features (word length, surprisal, orthographic neighborhood density, semantic word similarity, valence, and aesthetic potential).
    • all were computed algorithmically, need to read the appendix

language rabbit hole - "why does piss sound ruder than pee?"

#psychotherapy
Freud: Buddhism, new age, non-verbal, primal screaming, sexual dearmoring / bio-energetics

Lacan: German Idealism, reflection, talk therapy, logos therapy

The split subject (severed from the whole) constructs a scenario in relation to a lost object (in the deepest level (imo): a holy grail, divine beatitude).

Reminds of the myth of Brahman turning into a stag to chase his own creation.

sinthome: unique, idiosyncratic way a person organizes their jouissance (an intense, paradoxical blend of pleasure and suffering)

• for R. Crumb, it was his sexual comics
• those on "my strange addiction" have it worst

Does Lacan offer a cure?

Jung:
"There is no linear evolution; there is only a circumambulation of the self. Uniform development exists, at most, only at the beginning; later, everything points toward the centre."

• what was his methodology for this? transcendental reflection? saw it in a dream? on a walk? how do these guys make such assertions?

#economics
if there's a high ai token tax, i think it would end up pushed onto the heavily commodified llm wrapper software middle-man rather than consumers or ai labs.

whoever is most dependent on their neighbors in the supply chain will shoulder the tax burden. this may also motivate smaller labs targeting efficient small param models to sell to lower margin, lower compute demand, software wrappers. most people don't need Fable 10 anyway, and this will reduce energy consumption.

it makes sense to study job re-training in depth. from a cursory glance, it seems that federal initiatives have been historically ineffective,

cato's libertarian view is that the private sector will naturally invest in training employees if they have a need for trained employees. so there is no need for a job training program.

but this thinking breaks down when:

• you need long-term training for complex tasks. businesses rely on the state to shoulder the educational burden for many fields
• entrepreneurial blue collar work, like plumbers, electricians, building inspectors. you can maybe have an apprenticeship model here, but it seems state or ngo-sponsored vocational training is the primary path into these professions.

cato's conclusion seems ridiculous. a more reasonable take is: understand the widest skill gaps in the economy. train people for these real and valuable jobs.

worth looking into effective job transition programs in other countrys.

"Becker, Human Capital (1964), ch. 3: seed is alienable and can be collateralized; human capital cannot. Firms will finance firm-specific training (they capture the return) but not general training (the worker quits and captures it)."

^ this is also a ridiculous thing to say and is obviously not empirically true. economists are unserious people who treat "firms" as machiavellian min-maxers.