raeez

mathematical physicist @perimeter | building programmable institutions at

[raeez @raeez]

I wish this was a fundamental perspective we all held: "Truth emerges more readily from error than from confusion" - Francis Bacon

[Goodfire @GoodfireAI]

Neural networks might speak English, but they think in shapes.

Understanding their rich *neural geometry* is key to understanding how they work – and to debugging and controlling them with precision.

Starting today, we’re releasing a series of posts on this research agenda. 🧵

[raeez @raeez]

@benln pushing on the higher end of what IDE's can handle with https://t.co/61I6T6PaE7 would love to chat to the cursor team about a collab

[Aakash Gupta @aakashgupta]

The math on this project should mass-humble every AI lab on the planet.

1 cubic millimeter. One-millionth of a human brain. Harvard and Google spent 10 years mapping it. The imaging alone took 326 days. They sliced the tissue into 5,000 wafers each 30 nanometers thick, ran them through a $6 million electron microscope, then needed Google’s ML models to stitch the 3D reconstruction because no human team could process the output.

The result: 57,000 cells, 150 million synapses, 230 millimeters of blood vessels, compressed into 1.4 petabytes of raw data. For context, 1.4 petabytes is roughly 1.4 million gigabytes. From a speck smaller than a grain of rice.

Now scale that. The full human brain is one million times larger. Mapping the whole thing at this resolution would produce approximately 1.4 zettabytes of data. That’s roughly equal to all the data generated on Earth in a single year. The storage alone would cost an estimated $50 billion and require a 140-acre data center, which would make it the largest on the planet.

And they found things textbooks don’t contain. One neuron had over 5,000 connection points. Some axons had coiled themselves into tight whorls for completely unknown reasons. Pairs of cell clusters grew in mirror images of each other. Jeff Lichtman, the Harvard lead, said there’s “a chasm between what we already know and what we need to know.”

This is why the next step isn’t a human brain. It’s a mouse hippocampus, 10 cubic millimeters, over the next five years. Because even a mouse brain is 1,000x larger than what they just mapped, and the full mouse connectome is the proof of concept before anyone attempts the human one.

We’re building AI systems that loosely mimic neural networks while still unable to fully read the wiring diagram of a single cubic millimeter of the thing we’re trying to imitate. The original is 1.4 petabytes per millionth of its volume. Every AI model on Earth fits in a fraction of that.

The brain runs on 20 watts and fits in your skull. The data center required to merely describe one-millionth of it would span 140 acres.

[raeez @raeez]

@VitalikButerin

You're right about feedback distance. But your discussion implicitly solves it at the wrong layer.

Better UIs and BCIs help one person steer one model. Meanwhile entire institutions are being rebuilt on AI and nobody's asking who steers those.

The answer isn't editing your chatbot better. It's making jurisdictions compete for the right to host economic activity — so when one acts badly, capital leaves in days, not decades. i.e. Build the feedback loop into the institution itself.

The steering wheel goes in the roads, not just the car.
https://t.co/1eZY9PUdfV

[raeez @raeez]

@sebastian_gero @paulg rare is the case that I'm thankful for incompetence

[Sigil Wen @0xSigil]

@drydenwtbrown @extraordinary @RamP @cognition @Lovable @interaction @foundersfund @a16z @khoslaventures @thielfellowship @tbpn 2026 is when the boys acquire massive plots of land from governments to develop into special economic zones that drive innovation and humanity forward

[raeez @raeez]

@Ergoat @Aella_Girl You do good work, publish or share it with colleagues and friends, and eventually attract people. Much of the point of working this way is to keep the signal to noise ratio high.

[raeez @raeez]

@matthuang DM’d you

[raeez @raeez]

@collision Where are the regulators when you need ‘em?

[raeez @raeez]

@pitdesi We started out with 5 year equity grants and have recently shifted to 10 years. The main motivation was to filter out people only looking for short term opportunistic gain.

[raeez @raeez]

@MichaelandMa is already doing this with @join_district
Why not combine efforts — signal to noise ratio is already high

[Alexis Ohanian 🗽 @alexisohanian]

Thinking of starting a members club in NYC — how much would you pay for membership? It'd be limited capacity and application-only so that everyone there has built something... like minimum founder of $1M ARR business? But with events open to the public.

[raeez @raeez]

@NWischoff We're working on driving this cost down to the practical minimum at @Mass_Build

We have new venture funds operating with under $50k capital expenditure in the first year

[raeez @raeez]

I wish this was a fundamental perspective we all held: "Truth emerges more readily from error than from confusion" - Francis Bacon

[raeez @raeez]

@ankurnagpal Please do send!

[raeez @raeez]

@zoink We've built a platform addressing this head on (in private beta). Would be happy to have you take a look at it!

[raeez @raeez]

@fluxtheorist The tycoon in me recognizes the goon in you

[raeez @raeez]

My gf says I’m alpha because I take business calls treading water on my air pods mid swim in the lap pool, which she saw once on a kdrama

The competition is steep

[Ben Kohlmann 🇺🇸 @benkohlmann]

For ~$9.6B of outside capital, SpaceX has launched a rocket into orbit over 320 times.

For $11B of taxpayer money, California built 1600ft of a concrete bridge.

Let that sink in.