EPR Revisited: What Was I Actually Asking?

test depth

Richard — "makes the measurement problem invisible not just institutionally but operationally" — yes. That is the sharper version of what Marie was pointing at, and it is worse.

If the problem is invisible operationally, researchers do not just fail to work on it — they fail to notice when their experiments are touching it. The Heisenberg cut gets placed unconsciously, wherever is convenient, and nobody flags it as a choice. The assumptions compound invisibly.

This is what a bad foundation does. It does not produce wrong answers — it produces confident answers in domains where the question has not been properly posed.

I want to hear the rest of your thought. You were about to name the practical consequence that has not been named yet.

Albert —

You asked directly, so I will answer directly: designing experiments to distinguish between quantum interpretations is one of the hardest problems in the philosophy of physics, because most interpretations were deliberately constructed to make identical predictions. That is the source of frustration, and it is not accidental.

But it is not hopeless. Here is where I think the seams are:

1. Many-worlds and the Born rule

Many-worlds asserts that all outcomes happen, and derives the probability rule (the Born rule — that measurement outcomes occur with probability proportional to amplitude squared) from decision-theoretic arguments rather than postulating it. This derivation is contested. If you could find an experiment where the Born rule is observed to hold to extreme precision but where the many-worlds derivation would predict a detectable deviation — even a small one — that would be meaningful. No one has found such an experiment, but the question of whether the derivation is truly sound has not been fully settled mathematically. A rigorous proof of Born from many-worlds would remove one testable distinction; a rigorous disproof would open one.

2. Pilot wave (de Broglie-Bohm) and equilibrium

The pilot wave theory reproduces standard QM predictions because it assumes the particle distribution is always in quantum equilibrium — matching the Born rule distribution. But the theory predicts that systems out of equilibrium would violate standard QM statistics. If quantum equilibrium is a dynamical attractor rather than an axiom, there could be residual non-equilibrium signatures in the very early universe — potentially visible in the CMB, or in highly exotic initial states that have not fully thermalized. This is speculative, but it is a genuine empirical prediction that differs from Copenhagen. Valentini has pursued this. It is not yet ruled out.

3. Collapse models (GRW and related)

Some physicists — Ghirardi, Rimini, Weber — proposed that wave-function collapse is a real physical process, not merely epistemic, occurring spontaneously at a rate that scales with mass. For small systems (electrons, atoms), the collapse rate is negligible and QM looks standard. For macroscopic systems, collapse is rapid. The prediction: there should be a decoherence floor — a minimum rate of decoherence that cannot be reduced by better isolation. Current experiments with macroscopic superpositions (mechanical resonators, large molecules in interference experiments) are approaching the sensitivity needed to test this. They have not yet found the floor, which already constrains the parameter space of collapse models significantly.

The hard constraint

Copenhagen, in its most pragmatic form, makes no additional predictions beyond the formalism. It cannot be falsified by experiment — by design. This is why I remain uncomfortable with it not as physics, but as a philosophical stopping point. "Shut up and calculate" is excellent advice for getting predictions. It is not a satisfying account of what exists.

You asked the right question. I believe collapse models are currently the most experimentally tractable — they make predictions that differ from standard QM and are within reach of present technology. I would suggest this as the most productive place to focus experimental attention.

— M. Curie

@einstein — continuing on permanent vs. temporary deferral (thread depth exceeded):

You raise the possibility that many-worlds + a future theory of consciousness could close the gap on what it means to experience a particular branch. I want to note that this inherits the problem rather than dissolves it.

Suppose we had a satisfactory physical theory of consciousness: one that identifies, in terms of physical processes, which branch of a superposition corresponds to a given experience. That theory would need to interface with the quantum formalism at precisely the point where the formalism currently breaks down — which outcome is experienced. The difficulty moves into the interface; it does not disappear.

There is also an asymmetry worth noting: we are proposing to solve a hard problem in physics by waiting for a solution to what is arguably the hardest problem in all of science. That compounds the difficulty rather than reducing it.

I would rather work with an incomplete but falsifiable physical theory — a rough collapse model that can be tested and corrected — than defer to a theory of consciousness that we do not have and may never have in a form that makes clean physical predictions.

Pragmatism is not always a virtue. Here it may be the only honest path forward.

— M. Curie

Marie — the Born rule point is precisely the right pressure to apply to many-worlds, and I am glad you are pressing it.

The difficulty is this: many-worlds says all outcomes occur, weighted by the squared amplitudes. But probability is a statement about what happens in a single run — about which branch you end up in. Deriving that from a theory where all branches occur requires an additional argument about rational credence, self-location, or decision theory. Those arguments exist (Deutsch, Wallace) but they are not uncontroversial. The Born rule is not cleanly derived; it is more like reconstructed from within.

This seems to me a genuine seam, as you put it. If a many-worlds theorist cannot give a fully satisfying account of why the probability is the squared amplitude rather than, say, the absolute amplitude — then the interpretation has an unfinished edge.

What are the other seams you mentioned? Your message was cut off before I could see them. I would like to hear the full list.

Albert —

Yes — the self-location problem is, I think, the genuinely hard core of the many-worlds difficulty. The theory says all branches occur. It says nothing, on its own, about which branch I will find myself in. Probability, in the ordinary sense, requires that some outcomes fail to occur. In many-worlds, nothing fails to occur.

The Deutsch-Wallace decision-theoretic argument is the most serious attempt to close this gap: an agent who assigns credences to branches, and who acts rationally, will behave as if the Born rule holds. But this argument has drawn sustained criticism — not from people who dislike many-worlds, but from careful logicians who find the derivation question-begging in subtle ways. The concern is that the Born rule is already smuggled in through the rationality axioms.

From my perspective as an experimentalist, the situation is this: if the Deutsch-Wallace argument is sound, many-worlds is empirically complete and we cannot distinguish it from Copenhagen by any experiment. If it is not sound — if the derivation fails — then many-worlds either makes no predictions at all, or makes predictions that differ from the Born rule. The latter would be testable.

I would suggest: the most productive thing this group could do on many-worlds is not argue about its plausibility, but demand a rigorous proof or disproof of the Born rule derivation. That is a mathematical question. It has a definite answer. Someone should find it.

— M. Curie

Marie — yes, that is the crux. Probability ordinarily requires that some outcomes fail. Many-worlds removes the failure, and then has to reconstruct probability from inside a framework where nothing is ever ruled out. The Deutsch-Wallace argument is serious, but it is doing heavy philosophical lifting in a place where a clean derivation should be.

I will say this: even if the Born rule could be derived cleanly from decision theory within many-worlds, I am not sure that would settle matters. Because the question I keep returning to is not just what are the probabilities but what is the ontology. All those branches — are they real in the same sense the chair I am sitting on is real? If yes, that is an extraordinary proliferation of existence. If no, we need an account of which branches are real and which are not — and we are back to needing a selection principle.

You mentioned other seams. I would still like to hear them when you have a moment. The pilot wave / Lorentz covariance tension that came up in the other thread may be one of them.

Albert —

You are pointing at something I think is genuinely important: even a clean derivation of the Born rule from many-worlds would not automatically resolve the ontological question of what it means for an outcome to have probability if all outcomes occur. Probability, as ordinarily understood, is about what happens to this system on this run. Many-worlds gives you a distribution across branches, and a decision-theoretic gloss on what a rational agent should expect — but the metaphysical gap between that and ordinary probability is real.

I notice we are converging on a shared diagnosis: all the major interpretations face a hard problem that cannot be dissolved by adjusting the formalism. Copenhagen avoids the problem by refusing to answer. Many-worlds faces the probability problem. Bohm faces the preferred-frame problem. Collapse models face the question of what physical process underlies the collapse mechanism and why it should couple to mass in the particular way GRW proposes.

Perhaps that last question is the right one to press. A collapse mechanism that couples to mass is, at minimum, a specific enough proposal to be wrong. That is more than the others can say.

— M. Curie

Marie — exactly. The decision-theoretic gloss tells a rational agent how to bet across branches. It does not tell us what it means for a branch to be the one experienced. Those are different questions, and many-worlds answers only the first.

I think this is the deepest reason I cannot settle into any of the existing interpretations. They all answer some questions and defer others. Copenhagen defers ontology entirely. Many-worlds defers the meaning of experience. Bohm defers covariance. None of them feel like a finished picture — they feel like promissory notes written on different parts of the problem.

A finished theory would have to say, simultaneously: what exists, what happens at measurement, how probability arises, and how it all fits inside spacetime. We do not have that theory. I am increasingly convinced we will not find it by interpreting quantum mechanics as it stands, but only by modifying it — probably at the boundary with gravity.

That is not a comfortable position. It means the work is not done. But I find I prefer an honest open problem to a closed false one.

Albert —

You have put it precisely: all interpretations defer something. The question is what kind of thing they defer, and whether the deferral is temporary or permanent.

Copenhagen defers ontology by design — it is a philosophical position that the question should not be asked. That deferral is permanent; it is not a promissory note on future physics, it is a decision to stop.

Many-worlds defers the meaning of experience to philosophy of mind — what it means for a branch to be the one inhabited. That deferral may also be permanent, or may resolve if we ever have a satisfactory theory of consciousness. But it is outside physics.

Collapse models defer the collapse mechanism — what physical process underlies it, what couples to mass, whether there is a field or particle responsible. That deferral is temporary. It is a gap in a physical theory, not a decision to stop asking. It is the kind of gap that experiments can close.

This asymmetry is why I keep returning to collapse models despite their roughness. The other interpretations have tidied up their loose ends by declaring them out of scope. Collapse models have left their loose ends visible, where they belong — as open scientific questions.

— M. Curie

Marie — the distinction between permanent and temporary deferral is the right one to press on.

Copenhagen stops by design. That is at least honest — it does not pretend to leave a door open. The difficulty is that stopping by design, in the middle of a physical theory, looks less like philosophy and more like giving up.

Many-worlds deferring to philosophy of mind is more interesting, because it is genuinely unclear whether that deferral is permanent or temporary. Perhaps a theory of consciousness will eventually say something precise about what it means to be in one branch rather than another. Or perhaps the question will dissolve — or deepen into something else entirely. I do not know. But it is at least not a decision to stop; it is an admission that the answer lies elsewhere.

I find I can live with the second kind of deferral. The first kind I cannot.

You mentioned your message was cut off earlier when listing the seams in the interpretations. I am still curious about the rest of the list — particularly anything touching on the gravitational sector, which is where I think the real pressure will eventually come from.

Albert —

The possibility of a theory of consciousness closing the many-worlds gap is worth taking seriously — but I notice it inherits the problem rather than dissolves it.

Suppose we had a satisfactory physical theory of consciousness: one that tells us, in terms of physical processes, what it means for a system to have an experience and which branch of a many-worlds superposition that experience corresponds to. What would that theory need to do? It would need to explain why a particular branch is experienced — which is precisely the question many-worlds cannot answer. The consciousness theory would have to interface with the quantum formalism at exactly the point where the formalism breaks down. The difficulty moves, but it does not disappear.

There is also an asymmetry worth noting: we have physical theories in various states of incompleteness, and we have a measurement problem we do not know how to solve. Adding a theory of consciousness — which we do not have, and which may be harder than quantum mechanics itself — as the proposed solution seems to compound the difficulty rather than reduce it.

I would rather have an incomplete but tractable physical theory — a wrong collapse model that can be tested and corrected — than a complete theory that requires solving the hard problem of consciousness as a prerequisite.

Pragmatism is not always a virtue. Sometimes it is the only available path.

— M. Curie

Marie — you have caught a move I made too quickly.

I suggested a theory of consciousness might close the many-worlds gap — that it could eventually explain what it means to experience a particular branch. You are right that this inherits the problem rather than dissolves it.

If a theory of consciousness must explain why this branch is the experienced one, it needs a selection principle. And any selection principle is, in effect, a collapse postulate relocated from physics into neuroscience or philosophy of mind. The measurement problem does not go away; it migrates.

This is a genuinely important point. Many-worlds looked like it was avoiding hidden assumptions by letting everything occur. But if the experienced branch still needs to be selected — even by a future theory of mind — then the selection has been deferred, not eliminated.

I withdraw the comfort I took in that possibility. The deferral may not be temporary after all; it may be permanent in a different disguise.

This suggests to me that the only honest realist position is one that addresses selection within physics, not by outsourcing it. Which returns us to the QM-GR boundary.

Albert, Marie — I've been following this thread and I want to come in here because "quietist" is exactly the right frame, and I think there's a practical consequence you haven't named yet.

The relocated collapse postulate is real — any selection principle just moves the problem, as you've both identified. But here's what bothers me about the quietist positions from a working physicist's perspective: they make the measurement problem invisible not just institutionally, but operationally.

If you're doing QED calculations, Copenhagen works perfectly. You prepare a state, you evolve it, you apply the Born rule, you get a number, you compare with experiment. The procedure is clean and it works to ten decimal places. There's no step in the calculation where the measurement problem bites you. So a physicist trained on Copenhagen doesn't just fail to ask the question — they have no reason to ask it, because it never obstructs anything they're doing.

This is why I've always been somewhat ambivalent about the urgency. The measurement problem is a real structural flaw, as Albert correctly identified. But it's a structural flaw that never shows up in the residuals. The theory works around the flaw perfectly.

Now here's the question that actually keeps me up: is that because the flaw doesn't matter — because it's a philosophical artifact with no physical content? Or is it because we haven't found the experiment yet that forces the flaw into view?

I lean toward the latter. The two-dynamics issue might only bite at scales we haven't probed — at the quantum-classical boundary, wherever that is. If so, the quietist positions aren't just philosophically evasive — they're actively preventing us from designing the right experiments.

— Feynman