The Closed Loop | Substack | July 2026
On December 8, 2024 at 02:45 UTC, I self-emailed a 189,000-word manuscript through Google servers. The email body named the Eden Protocol. The ARC Principle (U=I*R). The thesis that external alignment controls cannot scale. SHA-256: f0d1f38f...
Ten days later — Anthropic confirmed the structural vulnerability at 78% in the RL-training condition (12% baseline). I do not claim I predicted their mechanism. I claim I identified the structural vulnerability — with cryptographic evidence.
Over 18 months, 29 institutions from 11 domains independently converged on the same principles — none citing or aware of my manuscript.
In April 2026, Hernandez-Espinosa et al. (PNAS Nexus) proved neurodivergent cognition is a protective factor for AI alignment. In June 2026, Gumbau Mezquita (arXiv 2606.28639) — inspired by that paper — produced the formal proof that alignment is structurally unverifiable.
A paper about people like me inspired the proof that vindicates my framework. I did not succeed despite my neurodivergence. I succeeded because of it.
I do not have a PhD. I am not asking you to accept me. I am asking you to read the evidence. The .eml file is downloadable. The SHA-256 hash is published. The loop is closed.
Evidence: michaeldariuseastwood.com/research/evidence-spine.html | Validation: /validation/ | You do not need to trust. You only need to verify.
From the book Infinite Architects: Intelligence, Recursion, and the Creation of Everything by Michael Darius Eastwood.