https://www.umsu.de/blog/ Musings in Analytic Philosophy https://www.umsu.de/blog/2026/837#c2484 https://www.umsu.de/blog/2026/837#c2484 Mon, 14 Sep 2026 09:33:04 +0000
You ask whether we *need* to deny CEM to make sense of QM (on any live interpretation of QM). I suspect the answer is no. But holding onto CEM comes at a cost. The details depend on how the CEM-friendly semantics is spelled out, but the basic problem is that every selection function effectively associates a definite outcome with every possible measurement, so relative to every selection function, there are hidden variables of the kind Bell's theorem seem to refute. That is, relative to every selection function, one can derive contradictions with QM if one accepts certain plausible further assumptions X. Given that QM is determinately true, it follows that X is false relative to every selection function; so X is determinately false. That's a cost. But it's only a cost. One can bite the bullet and accept, for example, the non-local counterfactual dependencies that I called "peculiar".

But the blog post is about a different question. It's about whether counterfactual conditionals in English have a semantics that renders CEM valid (e.g. Stalnaker's) or one that renders CEM invalid (e.g. Lewis's or Kratzer's). This is an empirical question that's ultimately decided by observations about how competent speakers use counterfactual conditionals. My claim is that when people think about counterfactuals in EPR-type situations, with Bell's theorem in mind, many competent speakers make judgements that are incompatible with CEM. What's the evidence? 1. This is the most natural interpretation of what they say: the common slogan "unperformed experiments have no results" certainly sounds like "for any unperformed measurement and result x: ¬∃x(M>x)". 2. Many commentators explicitly reject the cost that would come with accepting CEM (see above). 3. Some commentators explicitly endorse the anti-CEM verdict: Vaidman, Griffiths, but also Joyce, Lewis, Jeffrey, myself.

I'm somewhat sympathetic to a third possibility: that the conventions of English don't settle how we should talk about EPR-type situations. Perhaps this is a matter of open texture. In debates about CEM, it often seems to me that there are two dialects of English: some seem to speak a version of English in which CEM is valid, some a version in which it is invalid.]]> https://www.umsu.de/blog/2026/837#c2483 https://www.umsu.de/blog/2026/837#c2483 Sun, 13 Sep 2026 18:56:36 +0000
On the point about physicists being willing to reject instances of CEM: It's not clear to me that the physicists who deny counterfactual definiteness are really denying instances of CEM. They deny that it's *definite* what would have happened, had Bob's measurement been different. But that's something we defenders of CEM agree with. We will agree that it's definite that there's a fact of the matter about what would have happened with Alice, had Bob measured a different axis. But we'll deny that there's a definite fact of the matter about what would have happened, had Bob measured differently. I'd be surprised if the physicists are distinguishing between "Definitely: ((A>C) or (A>~C))" and "(Definitely: (A>C)) or (Definitely: (A>~C))". And insofar as they need only deny the latter in order to square counterfactual locality with QM, I think CEM does a reasonable job capturing what's going on with the debate about counterfactual definiteness. Is there some reason that they need to deny the former?]]> https://www.umsu.de/blog/2026/837#c2482 https://www.umsu.de/blog/2026/837#c2482 Sun, 13 Sep 2026 11:33:38 +0000
I don't follow your derivation of (3). QM doesn't say that if Alice had measured Z-spin while Bob and Carol measure X-spin, then there must be an even number of 'up's. In your setup, X_A*X_B*X_C = +1 and X_A*Z_B*Z_C = Z_A*X_B*Z_C = Z_A*Z_B*X_C = -1. We don't have Z_A*X_B*X_C = -1. In fact, Z_A*X_B*X_C can't have a definite value given that X_A*X_B*X_C has one. So "there's no fact of the matter" about what Alice would have measured.

We could try to fix this problem by changing *two* measurement choices. What if Alice and Bob had both decided to measure Z-spin, while Carol still measures X-spin and gets -1? Since Z_A*Z_B*X_C = -1, we can infer (by COUNTERFACTUAL LOCALITY and NOMIC STABILITY) that Alice and Bob would get the same result. But we can't say which result they would get. There's still no fact of the matter about what Alice would have measured.

Also, about the dialectic. I don't agree that if we reject COUNTERFACTUAL LOCALITY then there's nothing left of my argument. For one thing, my argument draws on the fact that CEM seems to imply *especially peculiar* violations of COUNTERFACTUAL LOCALITY. More importantly, the argument is that these violations are peculiar enough that many competent people who have thought hard about these situations have come to conclusions which (arguably) imply that the relevant instances of CEM are false. The point is not that these instances *are* false. The point is that there are (strange) cases in which a large number of competent speakers make reflective, sincere judgements that clash with the validity of CEM.
]]> https://www.umsu.de/blog/2026/837#c2481 https://www.umsu.de/blog/2026/837#c2481 Sun, 13 Sep 2026 10:15:21 +0000
Suppose that, in fact, they all measure X spin, and that Alice gets 'up' while Bob and Carol gets 'down'.

Assume the following two principles are not just true but definitely true:

COUNTERFACTUAL LOCALITY: If either Alice, Bob, or Carol had measured Z spin instead, that wouldn't have changed the others' measurements.

NOMIC STABILITY: If either Alice, Bob, or Carol had measured Z spin instead, they wouldn't have proven QM wrong.

Together, these assumptions imply that the following counterfactuals are definitely true:

(3) If Alice had measured Z spin, she'd have gotten 'down'
(4) If Bob had measured Z spin, he's have gotten 'up'
(5) If Carol had measured Z spin, she'd have gotten 'up'

On (3): COUNTERFACTUAL LOCALITY tells us that Bob's and Carol's X-spin down measurements wouldn't have changed. And QM says that, if Alice measures Z-spin while Bob and Carol measure X-spin, then there must be an even number of 'up's. So if they're not going to prove QM wrong, then Alice has to get `down'. So NOMIC STABILITY implies (3). Similar reasoning gets us (4) and (5).

So if we accept both COUNTERFACTUAL LOCALITY and NOMIC STABILITY, then we have to say that each of (3), (4), and (5) are definitely true. Even though measuring Z spin is a paradigmatic chancy process, we would say that there's a definite fact of the matter about how the measurement would have gone, had any of them performed it.

If we reject COUNTERFACTUAL LOCALITY---as I think we should---then there's nothing left of your argument against CEM. If reject NOMIC STABILITY, then the argument against CEM similarly falls apart, since the defender of CEM could likewise endorse COUNTERFACTUAL LOCALITY by insisting that Bell's inequality *would* be violated, were Alice to change her measurement.

So I think that, if you're going to make this argument against CEM, then you're going to have to accept quite a lot of counterfactual definiteness yourself. As a defender of CEM, on the other hand, I get to say that there's some true counterfact about what Alice would have seen, had she measured Z spin, but there's no *definite* counterfact about what she'd have seen. It might have been 'up', and it might have been 'down'.
]]> https://www.umsu.de/blog/2026/830#c2477 https://www.umsu.de/blog/2026/830#c2477 Tue, 09 Jun 2026 00:26:45 +0000 https://www.umsu.de/blog/2025/815#c2476 https://www.umsu.de/blog/2025/815#c2476 Sun, 31 May 2026 06:01:16 +0000 arxiv.org/abs/1310.0017


There's a lot of quantum stuff you can ignore. The "dF without symmetry" theorem is self contained.]]> https://www.umsu.de/blog/2026/829#c2475 https://www.umsu.de/blog/2026/829#c2475 Mon, 11 May 2026 14:54:54 +0000 https://www.umsu.de/blog/2026/829#c2474 https://www.umsu.de/blog/2026/829#c2474 Sun, 10 May 2026 18:28:21 +0000