Entanglement is correlation without communication — the spookiest thing in physics, and the closest nature gets to a distributed system primitive.
Consider a pair of gloves packed into separate boxes and shipped to opposite sides of the planet. Opening one box reveals a left glove. Instantly, it is known that the other box contains a right glove.
Nothing spooky yet — the gloves were assigned at packing time. This is a hidden variable explanation: the outcomes were predetermined.
Quantum entanglement is stranger: the gloves don’t have handedness until one box is opened. And Bell’s theorem proves this isn’t just ignorance — no assignment at packing time can reproduce the observed correlations.
Einstein, Podolsky, and Rosen (1935) argued that quantum mechanics must be incomplete. Their reasoning: if measuring particle A instantly determines particle B’s state (even at great distance), either:
Since (1) seemed impossible, they concluded (2) — quantum mechanics must be missing something.
They were wrong. Bell showed a third option exists.
Classical correlations (the gloves) obey Bell inequalities — mathematical limits on how correlated predetermined outcomes can be.
Quantum entangled particles violate Bell inequalities. Their correlations are stronger than any hidden-variable theory allows. This has been confirmed experimentally with increasing rigor since the 1970s, culminating in loophole-free tests in 2015.
The correlations are real, they’re non-local, and they’re not explained by pre-assigned values.
Bell’s theorem (1964) proves: no local hidden-variable theory can reproduce all predictions of quantum mechanics.
“Local” means information doesn’t travel faster than light. “Hidden variable” means outcomes are predetermined. Bell showed these two assumptions together are incompatible with quantum predictions — and experiments confirm quantum mechanics.
Something has to give: either locality or hidden variables (or both). Most physicists abandon hidden variables.
Here’s the catch: you can’t use entanglement to send information faster than light.
When you measure one entangled particle, the other’s state is correlated — but the individual measurement results look completely random. Only by comparing results (which requires classical communication) can you detect the correlation.
Entanglement is a resource, not a channel. It enables protocols (teleportation, superdense coding) but always requires a classical side-channel.
Suppose Alice and Bob share entangled particles. Alice measures hers. Bob’s particle is now correlated — but Bob’s measurement results, taken alone, look uniformly random. He can’t tell whether Alice has measured yet, what she measured, or what result she got.
The correlation only becomes visible when Alice and Bob compare notes — over a classical (light-speed-limited) channel.
This is not a loophole. It’s a theorem: the reduced density matrix of Bob’s particle is the same regardless of what Alice does. No signal passes.
CRDTs (Conflict-free Replicated Data Types) achieve non-local correlation without communication. Two replicas, operating independently, converge to the same state — not because they exchange messages, but because they share mathematical structure.
Entanglement works the same way. Two particles exhibit correlated outcomes not because a signal passes between them, but because they share a quantum state. The correlation is structural, not causal.
CRDTs converge because their merge function is commutative, associative, and idempotent. The mechanism is fully traceable.
Entanglement correlations have no known mechanism. Bell’s theorem rules out the obvious ones. The correlation just is — a brute fact about quantum states.
This is what Einstein found unacceptable. It remains philosophically uncomfortable, even as it’s experimentally certain.
This lesson establishes:
Next: The Measurement Problem