It is judged by the same method it uses to judge electoral systems.

There are two numbers, published together. The authors' self-assessment is about 88/100 (a band of roughly 84–89 under different weightings). An independent scoring of the same text — seven isolated reviewers: six expert panels and a simulated citizens' jury, none shown the self-score — came back at about 74/100 (band 72–75). Every single criterion scored lower under independent review. That gap is the honest headline, and it is published rather than smoothed away.

The independent panel's sharpest finding was on simplicity. The model's own rules say any criterion scoring below 5 caps the whole result, and the panel first put simplicity at 4.9 — which would have capped the design near 49/100. A comprehension layer was built in response: a one-page map of the whole model, a counted sixteen-concept citizen budget, and an institutional minimality ledger. The three panels that scored lowest re-checked it at 5.6 — off the cap, not solved. No re-check went above 7.

The independent score is itself labelled a proxy: the real rotating expert panel and the real convened citizens' jury the rules call for have not yet sat. Until they do, no authoritative figure exists — a self-grade is what the authors believe, and the proxy is a rigorous rehearsal of what independent eyes will say. Outcome quality scores lowest of the substantive criteria on both counts for the same reason: the strongest mechanism on paper is a hypothesis until the system runs.

Once it is running, the design scores itself continuously and in public, so any decline shows up as a falling number on a dashboard. That is the early-warning system against the slow erosion that kills other democracies before anyone names it.