Hypotheses

Working theories, unfinished thinking, and the experiments that test them. This is the candid corner of the blog: claims are published before they are proven, alongside the evidence that would change our minds.

1 active hypothesis

Open questions we are testing

Hypothesis October 21, 2026 7 min read

Project Ganglion: can biology make AI more adaptive?

Project Ganglion asks whether selected principles of biological neural organization — local processing, recurrent dynamics, adaptive regulation, selective plasticity, and distributed coordination — can be expressed as computational mechanisms and make artificial systems more robust under change. The work is early, and no results are claimed yet.

Read the hypothesis

How we publish hypotheses

State the claim. Define the test. Report the result.

Every hypothesis here is written to be challenged. We publish the reasoning before the results, including when the evidence points against us.

Challenge a hypothesis