Level III · a third kind of revision
Not every failure is a failure of the answer.
Change the weights: a parameter update. Change the explanation: a model revision. Change what the options were: a category revision.
The third one is rare, expensive, and never forced by the evidence.
Kind A
Responds to a single probe at one location. Never produces a pattern across the detector.
Kind B
Produces a spread pattern across the detector. Never registers at one location only.
Sort the specimens
Specimen 1
- Registers at one location
- No spread pattern
Specimen 2
- Spread pattern across the detector
- Never a single location
Specimen 3
- Registers at one location
- Repeatable to within noise
Specimen 4
- Spread pattern
- Pattern spacing varies with the aperture
Specimen 5
- Registers at one location
- Location varies run to run
Specimen 6
- Spread pattern
- Disappears if the aperture is closed
This happened.
For most of the nineteenth century light was a wave and matter was particles. Two categories, each with its own mathematics, each with its own half of the world.
Then light started depositing energy in lumps sized by frequency, and X-rays started recoiling off electrons like billiard balls. In 1924 de Broglie proposed the reverse: a particle of momentum p has a wavelength λ = h/p. Not as an image — as a number. Davisson and Germer fired electrons at nickel three years later and the wavelength came out where he said it would.
The anomaly was not in the electron. It was in a pair of words that had been treated as exhausting the options.
Interactive
Not a switch. A trade.
Fringe visibility
V = 1.00
The constraint
V² + D² = 1.00
Full interference. No information anywhere about which path was taken.
Schematic, not a simulation. The relation V² + D² ≤ 1 is real — Greenberger and Yasin formulated it in 1988 and Dürr, Nonn and Rempe tested it in 1998 — and it is the honest version of “wave or particle”: the two descriptions are not alternatives you pick between, they are quantities whose squares are bounded by one.
What the toy above is not
Seven invented specimens do not reproduce the history of quantum mechanics, and nothing you did on this page is evidence about physics. The sorting task is an illustration of a move; the physics is a case where the move was actually made, at enormous cost, over decades, against real resistance.
The relation under the graph is the part that has no analogue anywhere else on this site: V² + D² ≤ 1 is derived, numerically precise, and experimentally tested. It says exactly how much of each description you can have. Nothing in reading has a theorem.
And the thing this page must not say
Quantum mechanics does not show that reality is subjective, that all interpretations are equally valid, or that observation creates the world. Anyone telling you it does is selling something, and the interesting fact is strange enough without help.
The claim here is narrower and historical: sometimes an anomaly is best treated as a defect in the available vocabulary rather than in the object. That has happened. Whether it is the same move this project keeps finding elsewhere is exactly what is in doubt.