Requirement

Declare the usable task region, the relevant evidence identities, and the role of surrounding chrome. Where growing the region should expose more evidence, compare it with a named reference viewport in the same page and checkpoint. Preserve reference evidence and readable type. State the expected return on additional area instead of treating larger containers as more useful.

Why

DR-007 requires useful adaptation and DR-008 asks decoration to earn its place. This rule adds comparative measurements of allocation. The formulas and thresholds are project-authored contracts, not universal economic or aesthetic laws; empty space may support reading, interaction, or a finite task.

Computationally falsifiable — viewport-growth yield

Let A be the CSS-pixel area of one declared region intersected with the initial viewport. Let N be the number of distinct declared evidence identities whose complete item bounds fit inside that intersection and whose visible DOM text meets minFontSize. For reference r and larger target t: yield = ((Nt - Nr) / Nr) / ((At - Ar) / Ar) viewport-growth-yield compares this dimensionless yield with a project's minYield. For example, 50% more qualifying evidence in 100% more usable area gives a yield of 0.5. Container growth alone produces zero yield. Reference identities must remain qualifying at the target; replacing them with different items cannot compensate for their loss. Missing or duplicate identities do not count as useful growth. Both counts and the reference area must be positive, target area must strictly grow, and target viewport dimensions must not shrink. Missing references or invalid measurement evidence remain explicit findings at the configured severity, rather than passing or claiming a numeric yield.

Finite evidence

Optional finiteKeys declares the exhaustive evidence universe for the task. The yield threshold is exempt only when all those keys already qualify at the reference and still qualify at the target. The report identifies saturation; it does not claim a positive yield. Reaching the full set only at the target still requires the configured yield. Do not invent extra evidence, shrink text, or silently relabel the universe to obtain a pass.

Computationally falsifiable — chrome allocation

chrome-allocation limits the ratio of declared chrome's unioned border-box area to the visible area of one usable region. Chrome rectangles are clipped to that region and the initial viewport; overlapping or nested rectangles are counted once. maxRatio is a project-selected ceiling. This tests a declared allocation, not whether every toolbar or boundary deserves its space.

Human judgment

Choose relevant evidence units, usable regions, exhaustive finite sets, and warranted chrome. Counts do not measure information quality, importance, visual complexity, or the usefulness of a larger map. Geometry and computed text size do not establish occlusion-free visibility or readable paint. Unique identities do not establish independent facts: subdividing or duplicating one fact under new keys does not make it additional task evidence. Tasks needing richer detail in the same item may need a different contract. Inspect the examples and native-scale captures; there is no occupancy minimum. Measurement reference.

Enforcement

automated is the strongest mechanism Viewrule expects for the complete rule. Partial checks can still provide supporting evidence without proving the whole requirement.

Detection and enforcement details →