One decision, two layouts

What changes, and what is measured

The good version consolidates the summary, uses compact scalar context, and brings the evidence, recommendation, and action together. The map keeps roughly half the evidence width so individual facilities and invented localities remain distinguishable. The bad version adds repeated metrics, oversized scalar tracks, empty bands, and weak marks.

These are illustrative project boundaries: text coverage and empty bands use region-density; declared metric identities use repeated-metric; text distance uses evidence-proximity; scalar height uses max-height; solid CSS mark colors use mark-contrast. No semantic density or universal design-quality score is inferred.

Human review still decides relevance, justified repetition, geographic detail, recommendation quality, and contrast over complex terrain or imagery. The labels “good” and “bad” describe this task; they are not live engine reports or human approval.

Machine-readable catalog · Executable rules · Capture configuration

Policy: DR-006, DR-007, DR-008, DR-016.

Sources: Tufte: comparison, density, and graphical economy; W3C: non-text contrast. Layout thresholds are Viewrule fixture choices, not source-prescribed limits.