The self-score, opened up (§XVII)

How the live score is built

The model scores itself against its own twelve weighted criteria — in public, forever. Each criterion starts on the design-stage self-assessment prior (≈88; an independent proxy panel scored the same text ≈74, and both are published) and only moves as real operating evidence accrues. Criteria with no evidence yet stay honestly on the prior.

Live self-score

89/100

Weighted across the twelve criteria.

Weighted mean

8.86/10

Still on the paper prior

11/12

Criteria awaiting real operating evidence before they can move.

Per criterion

CriterionPriorLiveEvidenceConfidenceSource
Legitimacy & consent 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Outcome quality 8.0 8.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Rights protection 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Accountability 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Capture & corruption resistance 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Transparency & verifiability 9.0 9.4 measured 40% audit ledger verified intact
Representation & proportionality 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Resilience 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Adaptability & self-correction 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Intergenerational fairness 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Simplicity & usability 7.0 7.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)
Inclusiveness 9.0 9.0 paper prior 0% design prior (no operating evidence yet, §XVII.3)

An honest 88 is the point: the model’s own §0.6.5 forbids declaring a paper 10/10. Outcome quality (Criterion 2) stays a hypothesis until real decisions are measured against their predictions.