Controls
Each pair is two particles in the singlet state. Alice and Bob each measure their own particle along their own angle and get up or down. The angle between the two settings is θ. Bob's and Alice's bars keep counting across every setting you try; the comparison panel restarts when a setting changes.
Things to notice
Bob's bar never moves
Slide Alice's angle anywhere, run a few batches, then slide it again and run more. Bob's up and down bars stay at half and half, within the error bar, across every setting you have tried.
Why it cannot moveAlice tries to send a bit
Switch the toggle on and press play. Alice now jumps between two settings 90° apart to spell out a message. The readout splits Bob's record by which bit she was sending: the two halves agree.
What Bob would needThe correlation is real
Put Bob 60° from Alice and run. The measured point lands on the cosine curve at minus one half, below the straight line the simplest classical model gives. Bell showed no local model reaches the curve.
Bell's pointIt shows up only when the records meet
The right-hand panel is not something Bob can see. It is drawn from both lists laid side by side, and getting Alice's list to Bob takes an ordinary message at light speed or slower.
The missing stepEvery quantum ansible is this diagram with a step skipped
Le Guin's ansible names no mechanism. Card's philotic ansible pairs particles so the far end changes as the near end does. Both are the right-hand panel turning up at Bob's station on its own.
The two dossiersThe one way out is to change quantum mechanics
There is no slider for it here, because it is not a setting. Make the theory nonlinear and Bob's bar would move. Gisin and Polchinski showed that, and it is why nonlinear quantum mechanics is ruled out.
Register CAU-5What is going on
Entanglement is real and the correlations are strong. In the singlet state, if Alice and Bob measure along the same angle they always get opposite answers, and at other angles the chance of opposite answers is cos² of half the angle between them. That gives a correlation of minus cos θ, and Bell showed in 1964 that no model where each particle carries its own private instructions can produce that curve. The natural next thought is the entanglement radio: if Alice's choice changes Bob's particle at once, she can send him a message with no delay.
It does not work, and the reason is short. Everything Bob can measure on his own comes from his half of the state alone, which for the singlet is a coin toss whatever Alice does: measure, do not measure, pick any angle, the toss is the same. The partial trace proof was given by Eberhard in 1978 and by Ghirardi, Rimini and Weber in 1980, and Peres and Terno restate it in their 2004 review. The correlation only appears when the two lists are compared, and the lists travel by post. That is the no-communication theorem, CAU-5 in the register, and it covers cloning, amplifying and interfering on Bob's side too, which is what closed Herbert's FLASH proposal in 1982 and gave the world the no-cloning theorem along the way.
The full dossier, with citations, is entanglement signalling and the no-communication theorem. The two fictional devices it shadows are Le Guin's ansible and Ender's philotic ansible. The only escape hatch is nonlinear quantum mechanics, and Simon, Bužek and Gisin showed in 2001 that no signalling plus the usual structure of quantum states forces the theory to be linear, so the hatch is nailed shut from the other side.