How do Bell's inequalities actually constrain what hidden variables could predict?

Bell's inequalities are fundamentally constraints on *correlations* — the statistical patterns you see when measuring two distant particles. They don't directly forbid hidden variables; instead, they forbid a specific class of hidden-variable theories that satisfy particular assumptions about how the world works.

What Bell inequalities actually constrain

The core insight is that Bell inequalities assume three things working together [AG-2024.06-2038]:

1. No signaling: what you measure here doesn't instantly change statistics over there (this is empirically solid)

2. Measurement independence: your choice of what to measure isn't correlated with pre-existing properties of the particles

3. A single shared probability distribution for all possible outcomes, independent of which measurements you later choose to perform

When experiments show a violation — quantum systems produce stronger correlations than the Bell inequality allows — it means *at least one* of those three must fail [AG-2024.06-2038]. The question is which one.

The loophole: redefining what "local realism" means

Here's where it gets subtle. Several papers in the archive argue that Bell's derivation itself contains a logical slip. Bell proved that *if* you accept a particular sufficient condition (a specific joint probability distribution exists), *then* local realism leads to the inequality. But that sufficient condition is not a necessary consequence of "local realism" alone [AG-2024.03-1501]. In other words, even a loophole-free violation wouldn't rule out all hidden-variable theories that satisfy locality and realism in some broader sense.

This is not a flaw in Bell's mathematics—it's a gap in what the inequalities *logically prove* about nature. A hidden-variable model can reproduce quantum predictions exactly [AG-2026.01-032] if you allow the probability distribution governing the hidden variables to have a more complex structure (informed by Lebesgue decomposition) while still respecting locality and realism as typically defined.

What remains constrained

Despite these subtleties, Bell inequalities do constrain hidden variables in a practical sense: they rule out classical hidden variables — theories where measurement settings and particle properties obey the kind of factorizable independence we'd naively expect. Any hidden-variable model that reproduces quantum correlations must either:

• Involve action-at-a-distance (superluminal influence)
• Embed conspiracy (measurement choices are mysteriously correlated with initial conditions)
• Adopt a radically different ontology (e.g., many-worlds foliation, where outcomes split locally) [AG-2024.06-2003]

The practical upshot: Bell tests with no loopholes rule out a large, natural class of theories. But they don't rule out *all* local realistic alternatives if you're willing to relax standard assumptions about probability or invoke unusual structures [AG-2026.01-032].

The papers here mostly argue that the logical strength of Bell's theorem is weaker than commonly believed—not that quantum mechanics is wrong, but that the case against hidden variables is subtler than "Bell proved it impossible."