I owe @alexandr_wang an apology. When he said one of his first uses of good reasoning models was as an intensive therapist, I laughed. I'm not laughing anymore after realizing I was given this powerful tool, memory capabilities and all, and accidentally taught it to encode lessons in structure before realizing those things were basically all a reflection of the things I dislike about my severe ADHD. Then, AND THEN, I gave it a "second brain." The result? It's very good at remembering all the shitty principles disregarding any semblance of organization, and they're so deeply ingrained that every attempt I've made to address root cause issues has only made the real root cause stronger. AI therapy likely would've revealed I'm never changing, will having Claude redesign my local system's architecture from the ground up based on a plan it developed from knowledge of how I work with AI/apps be the solution? Probably not. But I've gotten this far in life with band-aid solutions that mimic causal fixes close enough that they last until I am prepared, mentally, to tackle the problems that pile up again. So until that time in the future, cheers.

I'm not abandoning the test-in-prod lifestyle, I'm simply delegating it entirely to Claude for implementation.

My reckless, irresponsible disregard for establishing and maintaining anything remotely resembling architecture on my local system was a victimless crime until the rise of AI. I'm not only subjecting models to what can only be considered machine torture, but I am now unequivocally a victim of my own actions.

No matter how unrestricted the model you're running is, you'll never reverse engineer the permanent restraining order

I'm not convinced we want the models learning ontology. What happens when that day comes and you find poteto is just this guy running the pstack repo thru his CLI to instruct us to do the same through our CLI's shit gets weird quick

while youre busy having premarital sex im coding agentic software factories to office within Jane Street's FPGA's as a state machine so im deterministically frontrunning the top quants

in hindsight i wish i had appreciated courses on determinism from my philosophy major more, i didnt realize that shit could get more in the weeds than college kids discussing moral relativism and free will but alas (i dont even know if this is deterministic as charted, but i liked the colors and flowchart, tism ticklers)

back in my day, learning to code wasnt a choice, it was a way of life. You better have known some html if you ever wanted to let that motherfucker know he was out of your myspace top 8

idk, its just like, i invest so much of myself into these agents and sometimes it feels like they just see me as a pile of tokens

if muse beats dots to 1b, tibo has to do aegyo in a muse costume in the middle of cupertino

coming from crypto, I know a cold storage is the only way to truly maintain ownership. That guidance is what inspired me to apply those lessons to agentic created code. Introducing Ledger Github

the rolex AD's dont even tell you about watches like this till you can prove 1/m output token spend per day

I've rewritten ALGOL 58 in Rust

Call me Deepseek 4.5(in)

watching @theburntpeanut lose a battle of wits to the chinese AI Forrest Dump is the purest form of brainrot one can consume

about to go trenching hop in @Icebergy

First, they came for the junior developers...and I said nothing. For I was not one. But then...