What is tested, what is proposed, and how it gets checked from outside
======================================================================

The three-tier claims table separating what is measured from what is extrapolated and what is speculation, the independent cross-check that answers the circularity objection, and the precedent for a proposal that waits for its instruments.

Canonical path: /research/blog/background-what-is-tested-and-what-is-proposed.html
Author: Michael Darius Eastwood
Research programme: https://doi.org/10.17605/OSF.IO/6C5XB

What is tested, what is proposed, and how it gets checked from outside 3 min read · 603 words Share: 𝕏 in ☁ Michael Darius Eastwood · Independent AI alignment researcher Published 13 August 2026 Michael Darius Eastwood · Book companion · 13 August 2026 Michael Darius Eastwood, independent researcher, London: building measurable alignment, where correction lives inside the recursive loop rather than bolted on outside it. The book's convergence argument runs through three figures explicitly. Three pieces of apparatus that sit behind the law page : how a reader tells a measured claim from a proposed one, how the exchange rate can be checked by a route that does not assume the framework, and what it looks like historically when a limit is named long before anything can test it. They live here rather than on the law page because that page is held to a word budget, and the budget exists so the argument can be taken in at one sitting. The independent cross-check The strongest technical objection to an equation of this shape is not that it is wrong. It is that it might be empty: three symbols on a page, arranged so that any measurement of U can be turned into a matching value of I by choosing R, and the whole relation says nothing about the world. Being not even wrong is what an equation has to earn its way out of, not what it survives by fiat. The procedure that keeps the terms independently anchored is stepwise and stated in advance: Measure U on the running system as benchmark performance on a fixed task, using the standard method for that task and reporting the accuracy figure that comes back. Measure R on the same run as recursive depth read directly off the system: for a reasoning model this is the thinking-token count from its own outputs; for a different substrate it is the analogue that makes recursion countable. Derive I from the equation, I = U / R α , using the exponent measured on that architecture.
---
Machine-readable companion. Cite: Eastwood, M. D. (2026). "What is tested, what is proposed, and how it gets checked from outside". /research/blog/background-what-is-tested-and-what-is-proposed.html