Edge Metrology
A card's edge is a manufactured line, and a line can be measured. The Straightedge release gives the engine an instrument for it: every edge's cut line is traced at full resolution, compared against the straight line the factory intended, and the difference is reported as numbers — not adjectives.
Section · 01
The question that started it
The edge bands have always been explicit about kind — clean, very minor roughness, visible roughness, chipping — but the quantity of roughness was carried inside examiner judgment. Two edges could both be called "minor visible roughness" when one is rough along a thumb's width and the other along half its length. The owner's question was direct: are we looking at 2% of the edge, or 50%? The honest answer was that the engine knew, but did not say. Now it says.
Section · 02
The instrument
For each of the four edges, on both faces, the engine isolates a full-resolution strip along the cut and finds the card's boundary in every column — thousands of measurements per edge. A robust line fit with outlier rejection recovers the cut the factory aimed for; a noise floor calibrated to the strip's own texture separates scanner grain from physical departure. Everything reported is a deviation from that fitted line. Where no clean scan-bed margin is visible, the instrument abstains rather than guesses — an unmeasured edge is reported as unmeasured, never scored blind.
Section · 03
What it reports
| Metric | What it answers |
|---|---|
| Coverage | The share of the edge's length that departs from the fitted cut line beyond the noise floor. Answers the owner's question directly: is 2% of this edge rough, or 50%? |
| Amplitude | How far those departures reach — mean, 95th-percentile, and maximum deviation in pixels. Separates fine texture from real bites out of the line. |
| Chips | Inward notches: material missing from the card. Counted individually with their depth, because one deep chip means something different than diffuse texture. |
| Fibers | Outward spurs: paper debris left standing by the cut. The signature of a rough factory blade — counted separately so cut debris is never mistaken for a chip. |
| Localization | Whether the roughness concentrates in one stretch or spreads along the whole edge — the worst window's share of the total. |
Section · 04
First light: a 1990 Fleer
The instrument's first reading was taken on a card the examiner had already graded by the published bands — serial 2582110, a 1990 Fleer whose bottom edge carries the set's characteristic rough factory cut. The instrument had no access to the examiner's answers. It found them anyway:
| Edge | Coverage | Fibers | Examiner call |
|---|---|---|---|
| Bottom · front | 15.5% | 3 (deepest 50.6px) | Minor visible roughness — 8.7 |
| Bottom · back | 10.6% | 4 | Minor visible roughness — 8.7 |
| Top · Right · Left | 0.4–9% | 0 | Clean — 9.8 |
The one edge the examiner charged is the one edge the instrument flagged — on both faces — with every fiber it found sitting on the factory-cut bottom and none anywhere else. Zero chips on any edge, which is also correct: this is a rough cut, not a damaged edge, and the instrument tells those apart by construction.
Section · 05
No pass for factory cuts
Measuring a rough cut precisely is not a step toward excusing it. The standing ruling is the opposite: the edge law applies to every card no matter the set, and a rough factory cut deducts exactly like any rough edge, because the law measures the edge in front of it — not the era that produced it. What metrology changes is the evidence: a deduction for roughness now arrives with the coverage, amplitude, and fiber census that justify it.
Section · 06
Shadow mode and governance
Straightedge ships in shadow mode: the measurements are recorded on every edge of every card graded, and no grade moves because of them. The published edge bands, the worst-of-faces rule, and the quadrant law — average capped by the lowest edge — are unchanged, as is the final-grade formula. Calibration follows the same path every instrument here has walked: measure in shadow, anchor against examiner ground truth, and only then, by versioned release, let the numbers carry weight. New gradings only; issued grades change only through measured re-examination on the audit log.
CF-WP-2026-004 · v1.0 · Published September 25, 2026 · Release: Straightedge · Registered on the CardFax