
up at night. in a good way.

someone has to care and
it might as well be me
going to fix itself


society
and pretending any of these
are separate hobbies

deep in Claude — writing prompts so long they could qualify as novellas, gently trying to convince it I'm not jailbreaking, I'm just curious.
separately obsessed with the question nobody in tech wants to answer honestly: if AI writes most of the code, what exactly are we hiring engineers for? (taste? vibes? the haunted look of someone who has mass-refactored a monolith and lived?)
helping friends figure out how to hire when “implement an LRUCache that supports O(1) average-time” is a party trick any LLM does in its sleep. the whole interview process needs a funeral and a rebirth. working on it.
· · ·also losing sleep over this: can you bet on the future of blue chip art prices? not hypothetically — on Polymarket, on Kalshi, with real money on the line. “will a Basquiat sell above $50M at auction in 2027?” “will a Leonora Carrington outperform a Yoshitomo Nara over five years?”
the markets exist. the data pipeline to feed them accurately doesn't. auction results are buried in Christie's PDFs. gallery prices move through dealer whisper networks. secondary market signals are scattered across a dozen platforms that don't talk to each other.