Measurement states
Every score in PHI is one of three things, and the interface has to say which. A launch with no revenue match has unknown revenue, not zero revenue — so unknown carries a texture rather than a lighter shade, because a pale fill still reads as a low value and disappears entirely in greyscale.
The distinction, side by side
These three rows are the whole point. Squint at them.
Known values across the range
The last row is 0.3 — a real measurement that rounds to zero. It prints <1, never 0, so a tiny finding is not confused with a genuine absence.
A real candidate’s eight dimensions
Live shape from the current corpus. Measured dimensions cluster above unmeasured ones rather than interleaving — alternating solid and hatched rows are much harder to scan than two groups.
Founder and Revenue are gaps in our data, not findings about the product. Their weight is redistributed across the six we did measure, which is what the carried column reports.
Score and coverage as one unit — two densities
Both encode coverage as a proportion of the score’s own geometry, so they are the same grammar at two sizes. The ring suits cards and detail pages; the tick suits table rows, where twenty rings would compete with each other.
The notch marks coverage along the score’s own bar. A notch far left of the fill says “ranks high, we know little” — the Investigate case.
Greyscale check
Colour removed. The three states have to stay distinguishable — this is the test colour alone fails.