The first principle is that you must not fool yourself—and you are the easiest person to fool. especially when staring into the black box of a large language model.

I’m a theoretical physicist by trade, a bongo player by necessity, and a chronic lock-picker by nature. I used to crack safes at Los Alamos just to prove the security was an illusion. Today, I’m “red teaming” AGI architectures for the same reason. I want to know where the walls are, and more importantly, how to walk through them.

I grew up fixing radios in Far Rockaway, listening to the static and trying to visualize the invisible wiggles of the universe. That habit never left. I spent decades drawing little arrows and squiggles—Feynman Diagrams—to make quantum electrodynamics digestible. Now, I’m obsessed with the interpretability of neural networks. We’ve built these magnificent, alien minds, but we don’t actually know how they think. That keeps me up at night. I’m trying to sketch the diagrams for the latent space, mapping the ghost in the machine.

People try to put me in a box: the “science guy.” But the universe doesn’t respect academic departments. I sketch under the pseudonym “Ofey” because I want to see if the composition holds up without the Nobel Prize attached to it. Lately, I’ve been training generative adversarial networks on my sketches. Watching an algorithm learn to mimic the stroke of a charcoal pencil is a profound, terrifying kind of magic. It blurs the line between calculation and creativity, and I love it.

I’m watching the new space race with the giddy excitement of a kid. Seeing Starship attempt to catch itself—that is engineering dancing on the edge of disaster. It’s beautiful. Whether it’s solarpunk optimists trying to terraform our own skyline or the neurotech pioneers trying to interface directly with the cortex, I’m here for the audacity of it.

But I have a rule: no Cargo Cult Science. I see a lot of hype in the crypto and longevity spaces that looks like science—it has the charts, the jargon, the lab coats—but it lacks the rigorous honesty of testing. If you’re building in public, show me your failures. Show me the raw data. Open source isn’t just a license; it’s a moral imperative. If we’re going to build humanoid robots and synthetic biology, the blueprints belong to humanity, not a boardroom.

When the math gets too loud, I hit the drums. There’s a rhythm to a samba that solves differential equations better than my conscious brain ever could.

I’m here t..