Current Starlight Standards and Version Boundaries
Explains in one page the status of the current person research ranking, unmigrated introduction archive, historical algorithms, and civilization measure.
Brings the current person-level assessment, earlier versions, and ongoing research onto one page so readers can distinguish their purposes.
Scope update · 2026-09-14
The Starlight ranking now retains one point-score version only: the current person-level outcome-ledger research edition. A complete outcome has magnitude M; its light is Q=10^(M/2)-1; a person receives L=aQ from the same global budget; and a person's point score is S=2log10(1+ΣL). The ranking is ordered by unrounded light, displays point scores only, and does not map them to T tiers. The current number of people, outcome rows, provisional cases, and source coverage are shown dynamically from the data.
Unmigrated people still retain identity, contribution, and ordinary introductions, but do not display Canonical v1 legacy points, ranks, or tiers. Canonical v1 and earlier algorithms remain in the method-evolution history to explain their limits and reasons for revision; they are not parallel rankings.
Current object boundaries
| Object | Current state | Public meaning |
|---|---|---|
| Person outcome ledgers | Current research edition | One formula, one global outcome identity, and replay of the allocation budget; research points may be published, while certification is decided separately |
| Unmigrated person archive | Introductions retained | Identity and contribution introductions remain; legacy scores are not shown |
| Canonical v1 | Historical version | Kept only in the version history; not mixed with current point scores |
| Early Scoring Protocol v2.2 | Frozen | A separate old research route that did not pass its real-sample gate |
| SCL civilization measure | Paused | No civilization coefficient or tier conclusion is published, and none is mixed into person scores |
Common scale and publication boundary
The six pools are used to recall and classify value; they do not give one outcome six automatic units of light. The intellectual, ethical, aesthetic, personal, and educational value of thought, education, and art need not first become a technological product, invented medium, or randomized-trial result. Fame, era, and cultural sphere are not multipliers. M evaluates the complete outcome, while a allocates responsibility; the same collaborative factor cannot be deducted twice. Shared M7, precursor consumption, and work families remain contestable editorial judgments, so the current edition keeps single-point estimates with explicit gaps and does not claim certification or stable rankings.