Evidence

Evidence ledger

This page is generated from the same data the scoring engine reads: the nightly reliability estimates and the active norm table. Nothing on it is written by marketing, and a row can only show an observed figure when the observation exists.

Active norm versionprovisional_v1Norm statusProvisional (placeholder distribution)Reference sample nnot yet collected

Reliability figures will be published once 30 usable completions exist.

ConstructLayerObserved ωObserved αnBand width sourceLast estimated
ComposureLAnot yet estimatedNot available0provisional placeholderNot available
ConnectionLAnot yet estimatedNot available0provisional placeholderNot available
CuriosityLAnot yet estimatedNot available0provisional placeholderNot available
DisciplineLAnot yet estimatedNot available0provisional placeholderNot available
DriveLAnot yet estimatedNot available0provisional placeholderNot available
IntegrityLAnot yet estimatedNot available0provisional placeholderNot available
ExperimentationLDnot yet estimatedNot available0provisional placeholderNot available
Feedback SeekingLDnot yet estimatedNot available0provisional placeholderNot available
ReflectionLDnot yet estimatedNot available0provisional placeholderNot available

"Provisional placeholder" means the confidence bands for that construct are drawn from a documented assumed reliability, not an observed one; the nightly job replaces placeholders automatically once at least 100 valid protocols exist for the construct. The reference distribution itself is rebuilt only at n ≥ 300 under the norm-versioning rules on the science page. What the instrument cannot yet do is stated in full on the limitations page, and the definition behind each construct is on the construct glossary.

Movement

What has changed, and on what base

Usable cases (completed, quality check clean)3Last nightly reliability estimatenot yet run on live data
  1. 23 September 2026Usable-case definition adopted: completed sessions with no validity flags. Reported figures use this baseline.

Questions

Not yet. Equilibrium is in pilot: score bands use documented placeholder reliabilities until 100 valid responses per construct exist, and norms are rebuilt only at 300.