Methodology / v0.1
Archived — superseded by v0.2
Methodology v0.1
Assessments generated under v0.1 are stamped with this version. The current methodology is v0.2.
Core framework
v0.1 established the foundational assessment model: seven dimensions, four flag valences, a weighted scoring algorithm, and the scoring constraints that remain in force in v0.2.
- Seven assessment dimensions, each with named flag categories.
- Four flag valences: Strength (+2), Caution (−1), Concern (−3), Context (0).
- Priority weights: dimension ranked first × 7, last × 1.
- Recommendation threshold: minimum +10 point improvement.
- Source attribution: every published flag must have at least one verifiable source.
What changed in v0.2
- Flag version history (version column, flag group permalinks at /flags/[id]).
- Alternative-priority recommendations — when no qualifying alternative exists under stated priorities, the engine probes alternative weightings.
- Descriptive score-delta labels (e.g. “Significantly better aligned”) replacing raw point values.
- Six-branch assessment summary sentence based on the pattern of results.
- Source rendering on the result page, including verified dates.
- Catalogue request flow for services not in the catalogue.
- Per-flag verification overdue indicator.
- Catalogue changelog and public data export.
- Review cadences formalised: AI assistants monthly, browsers quarterly, others six-monthly to annually.