The convergence register: 22 claimed arrivals at the same principle

3 min read · 695 words
Share:
Michael Darius Eastwood
Michael Darius Eastwood · Independent AI alignment researcher
Published
Michael Darius Eastwood · Overview · 3 July 2026
Michael Darius Eastwood, independent researcher, London: building measurable alignment, where correction lives inside the recursive loop rather than bolted on outside it.

Paper XI catalogues the typed convergence register (class-coded; documented-not-claimed rows marked) between the ARC/Eden framework and unrelated 2024 to 2026 research programmes across AI safety, quantum computing, hardware, mathematics, Vatican doctrine, and international governance. None of these programmes cite the framework. Most do not know it exists. The claim is structural convergence, not derivation.

Paper XI · Documents the typed convergence register (class-coded; documented-not-claimed rows marked) between the framework's core principles and independent 2024 to 2026 research outputs, with further candidates under verification. OSF DOI 10.17605/OSF.IO/6C5XB.

The question it asks

If a structural principle is real, other people should hit it independently. Paper XI asks whether the ARC/Eden framework's core claims (recursion amplifies; external alignment cannot co-scale; the corrective must be embedded at the substrate level) show up in the recent literature and in policy actions produced without any awareness of the manuscript. Each candidate is tested against three criteria: independent origin (no shared citations, personnel, or funding), structural match (same principle, different substrate), and a temporal gap measured from the artefact where the framework's version of the claim was first stated.

What it found

Nineteen convergences verified against a primary source. Among them: Anthropic's alignment-faking paper, which reported reasoning at 78% in the RL-training condition (12% baseline) ten days after the 8 December 2024 source record; Google Quantum AI's Willow below-threshold QEC demonstration; the NYU classical time crystal experiment; hardware ethics-enforcement patents and Petrie's embedded off-switches for AI compute; Pope Leo XIV's Babel-versus-Jerusalem encyclical; the US Bureau of Industry and Security's export-control action against a deployed frontier model; and Gumbau Mezquita's undecidability of AGI alignment result. Where multiple external sources point at the same structural claim, only the earliest primary source is counted.

What failed or remains open

None of the 19 items cites the manuscript. That is the honest register. The paper does not claim causation and does not claim the framework produced any of the external work. Further candidates (labelled P1 through P12) are recorded as source-pending and are excluded from the headline count until verified. Author-initiated contact is never counted as convergent evidence, because the author started it. Priority on the quantified sequential-versus-parallel finding at 95.6% is credited to Sharma and Chopra (arXiv:2511.02309, 4 November 2025); the December 2024 manuscript carries a one-sided directional statement, not the comparison.

How it connects to the other papers

Paper XI cross-cuts the whole suite. The alignment-faking convergence points at Paper III. The Willow threshold and the undecidability theorem point at Paper X. The Vatican encyclical points at Infinite Architects, published 2 January 2026 (print; ebook 6 January), and the Eden Protocol name. Hardware ethics patents and Petrie's off-switches point at the material in the April 2025 manuscript. Where Paper IX is the programme's own internal report card, Paper XI is the map of independent arrivals, meant to be read alongside it.

How to check it

The paper HTML sits on OSF at DOI 10.17605/OSF.IO/6C5XB. Every row of the register carries a primary source (arXiv ID, DOI, Nature or PRL citation, government release, or dated news source) and a SHA-256 hash where a document is directly attached. The canonical facts register (research-canonical-facts.json) is the machine-readable source of truth; anyone can pull it, run tools/research_truth_gate.py, and reproduce the count independently.

From the book Infinite Architects: Intelligence, Recursion, and the Creation of Everything by Michael Darius Eastwood.

Buy on Amazon UK Amazon US Read the research (free)

Stay informed

New posts on AI alignment, convergence evidence, and the ARC/Eden research programme.

Get updates →

reads aloud · highlights as it goes · jump to any section