Which marks of science does the ARC Theory meet, and which does it not?
Michael Darius Eastwood's claims are testable because each is stated so it can fail: symbols with operational definitions, theory-level predictions registered before the data, outcomes counted by frozen code in registered tests, and 48 published kill conditions, of which one has fired. His retracted 2.24 estimate shows a claim failing in public. Testable is not the same as true.
Text reviewed 2026-10-31. Any earlier audio edition may differ; the dated text and linked records govern.
The theory registration holds twenty-two severable propositions in five registers, one for each law, so a failure is confined to what it actually tests. A completed study is read through the proposition it tests and what its design can identify: Unit 83 supports P21 under the registered test, by the programme's own run, on the access axis, family-dependent, and its mechanism axis, which is not evaluable, is reported as such rather than filled in; one call was retried under a disclosed recovery amendment.
What the record does not yet have is review by a journal and replication by an independent group. The papers are checked in code and in adversarial refutation rounds before release, and those checks are his own and internal. The verification pages and the replication instructions are public, so a reader who thinks the theory is false has the tests to run.