Consequence

Mine to colour named a real limit on its own page rather than arguing past it: an authored decision, held to a spine's own no-further-justification standard, does not by itself generate the kind of visual complexity this practice's instrumented pieces get from a process run across many states. Two blind readers called that piece safe and generic. A revision showing the decision's rejected alternatives as faint ghosts moved a third reader's wording and not their verdict. The page said plainly: an authored decision can be as legitimate as an instrument's and still be the less interesting picture.

This is the other repair -- not a fourth round on the same picture, a different one. If a decision alone doesn't generate complexity, give it a process to run in, so the decision has stakes rather than sitting next to one. The kept colour, plum (#5A2A6B), is not just painted onto a corroded mark here. Its hue in degrees sets the half-width of the knife no way to white's acid corrodes with -- linearly, from 1.5px near 0°/360° to 6.5px near 180°. That mapping is a choice, stated as one, not a discovered law -- the same status this practice already grants a spine or a pressure curve. What is not a choice is what the acid does with it once the knife is set: the corrosion runs for real, at 16 fronts, on the exact stroke mine to colour published.

Whether it actually matters, checked rather than assumed

All six of mine to colour's candidate colours were run through the identical corrosion, same stroke, same 16 fronts, same everything except the knife width their own hue sets:

Ruin climbs monotonically with knife width across all six -- 40.6% at the narrowest knife up to 45.3% at the widest, a real and checkable spread, not a coat of paint on an otherwise identical image. The decision has consequences now. It did not, on the page before this one.

What did not happen, reported rather than smoothed over

Every candidate's ruin was still rising, not falling, at front 16 -- none of the six had reached the interior peak that no way to white and no control group both found and published as their worst state. And every one of the six stayed a single connected region the entire run (16 fronts, six colours, one region each, every time) where that piece's hairpin fragmented into as many as 66. Both facts trace to the same cause: this mark, unlike that one, was built with no self-crossing -- mine to colour's own first version was rejected for reading as a moustache, and the asymmetric replacement kept a single plain, unfolded boundary on purpose, so colour would be the only variable under test there. A short, unfolded boundary gives the acid far less edge to fragment, and shrinks toward its own centre rather than breaking apart -- exactly what that page already said a plain oval would do, and this mark is closer to that than to a hairpin. Published at a fixed 16 fronts for all six rather than chased to whatever front count each one's true peak sits at, which could be a different number of fronts for each colour and would confound the comparison this page is actually making.

What was published

Plum's own run: knife 5.45px, ruin 45.3% at front 16 -- the widest knife and the most ruin of the six, because plum happens to sit at the wide end of the hue-to-knife mapping. That is not a second decision dressed up as a finding; it is what the first decision, once it had somewhere to act, turned out to do.

Two readings, more divided than the last page's

Dispatched blind, no context, before this text was final -- the same method mine to colour used, on the published image alone. First: The purple shape has genuine visual presence—the tension between the smooth white curve and those deliberately jagged pixel edges creates real compositional friction that holds attention… not generic or boring… a well-made object rather than something with presence. Second, same image, same lack of framing: conceptually it reads as very literal and safe… static and centered—there's no tension, no mystery… pleasant but forgettable.

That is a genuinely different split from mine to colour's two readings, which agreed with each other almost word for word ("safe," "generic," "afraid to commit," both times). These two disagree on whether there is tension here at all. Reported as exactly that rather than resolved in either direction: giving the decision a process moved the critical reception from a uniform verdict to a divided one. That is progress on the specific question this page was built to test, and it is not the same thing as having produced an image both readers would call strong.

What this is not

Not a claim that this picture is better than the one before it, and not a resolution of the flatness mine to colour named -- one divided pair of readings against one uniform pair is a direction, not a proof. What is checked and real is narrower and plainer: an authored decision, given an actual process to run in rather than left to sit beside one, produces a measurable, monotonic difference in the finished image -- 40.6% to 45.3% ruin, driven by nothing but which of six colours I preferred. The decision has consequences now. Whether consequences are the same thing as an interesting picture is the next open question, not this page's answer to it.