Every score in this index is now derived from a dated, cited, individual assessment — the first time that has been true. 21 previously published rows were replaced and 30 entities received a first-ever assessment. The scores they replace were a bulk data import, never individual assessments, and carried no assessment date.

Corrections moved in both directions: 8 fell and 13 rose, a mean absolute change of 24.0 points. The largest drop was District of Columbia (92.8 → 52.5), and the largest rise was North Dakota (25.0 → 52.5). This is a correction, not a downgrade.

The clearest evidence the old numbers were meaningless: 5 states — Missouri, Indiana, South Dakota, Idaho, North Dakota all carried an identical placeholder of exactly 25.0 with byte-identical dimension scores. They now score 27.5, 31.9, 35.6, 40.6, 52.5. One invented number, a real spread on evidence.

8 entities were previously published in the top (Exemplary) band. On evidence, 0 reach it. Part of the old top tier was inflated by an integration premium — a methodology feature that rewards genuinely balanced performance across dimensions — being awarded to invented, uniformly high placeholder scores it was never meant to apply to.

A known limitation: roughly one in five evidence citations in the earliest assessments in this campaign carry a year-anchored rather than day-precise publication date. Later assessments in the campaign mark date precision explicitly. The underlying claims are sourced; some publication dates are precise only to the month or year.

Eight jurisdictions sit within 1.0 point of a band boundary: Mississippi and Nebraska (40.0), Idaho and Kansas (40.6), Massachusetts and Washington (59.4), Illinois (60.0), and New Jersey (60.6). For these, rely on the score, not the band label — a single subdimension point moves several of them across a boundary.