A scientist writes the hypothesis, specifies the analysis, and lodges the document with a timestamp — before collecting the data. The registration is public and immutable. When the results come in, the question that was asked is fixed; it cannot drift toward the answer that arrived.
The failure mode it guards against has an acronym: HARKing — Hypothesizing After Results are Known. The researcher, feeling all the while that she is merely clarifying, edits the question to match the data she got. The loop closes quietly: the reference that was meant to be external (the prior hypothesis, the thing the evidence was supposed to test) becomes internal, authored by the very data it was supposed to constrain. A mirror, not a window.
The Canon chapter calls pre-registration the cleanest modern case of the canon-operation: the sealed text as a lock against the most insidious drift there is, the drift that feels like understanding. The timestamp is the colophon; the immutable registry is the append-only ledger that “adds nothing and takes nothing away.” The Deuteronomic prohibition — you shall not add to the word, nor take from it — and the pre-registration platform solve the same problem with the same tool, three thousand years and one vocabulary apart. The framework’s own kills-ledger is the same discipline turned on itself.
Sources. Study pre-registration (the OSF registry; clinical-trial registration); the failure mode HARKing — “Hypothesizing After the Results are Known” (Kerr, 1998). Search: preregistration HARKing Kerr 1998 hypothesizing after results known.
Appears in: The Canon — the word they would not let drift