Research register

Every study earns a stable record before collection. Plans, observations, findings, corrections and withdrawals stay visibly separate.

Registry state · one planned pilot · collection not started

01

Registered studies

A registration is a commitment to a method, not proof of data or a finding.

AF-001
Control-set longitudinal pilot
Protocol draft
Test whether a stable question set can produce lawful, comparable public observations across repeated windows.
Collection
Not started
Publication
No findings
Owner
Matthias Ramahi
Blocking gate
Protocol freeze, independent reviewer status, future collection cost and provider rights review remain open.

02

Study contract

Plan
Freeze the question set, public surface, locale, route, timing, exclusions and cost ceiling.
Observe
Record public outputs and visible sources with timestamps and missing-data reasons.
Compare
Break or annotate the series when a prompt, interface, model or policy change invalidates comparability.
Publish
Release only rights-safe evidence with denominators, method, limitations and corrections.

The next valid research action is to freeze the protocol—not to announce a result.

Inspect methodology v0.1