CHAPTER 2The glove analogy

In which we try to be Einstein for a chapter, and very nearly succeed.

§2.1A pair of gloves in the post

Forget particles for a moment. Imagine I have a pair of gloves on my desk — one left, one right. I'm going to do something very ordinary. I take two identical-looking sealed boxes. Without looking, I drop one glove into each box, shake the boxes around so I lose track of which is which, then post one box to my friend Alice in London and the other to my friend Bob in Tokyo. Neither of them knows what's inside.

A few days later, the boxes arrive. Alice opens hers first. Inside, she finds — let's say — the left glove. The instant she sees it, she knows, without doing anything else, that Bob's box must contain the right one. If Bob then opens his, sure enough: right glove. Their results are perfectly correlated. Whichever one Alice gets, Bob gets the opposite.

FIG. 2.1 Nothing in this story violates physics. The correlation is "instant" only because the two gloves were already a left-right pair before they left my hands. The information was in the boxes the whole time; reading one tells you about the other because they share a common origin, not because they're communicating.

This is the kind of explanation Einstein wanted for the EPR experiment. Maybe each pair of particles is just like a pair of gloves: the source prepares them together, packs each one with a complete instruction sheet about what to do at every possible dial setting, and sends them out. When Alice's box flashes ▲, she's just opening her envelope and reading what was written there. Bob's particle, having the matching envelope, flashes the opposite. No signal crosses the gap. No spookiness.

Physicists have a name for this kind of theory: a local hidden-variable theory. "Local" because nothing crosses the space between Alice and Bob; "hidden variables" because there are extra properties (the envelope contents) that quantum mechanics doesn't bother to track but which, in Einstein's view, must really be there.

A clean, sensible, classical idea. Let's try to build one.

§2.2Applying the idea to the experiment

The challenge: Alice's dial can be set to any angle. So the particle's "envelope" can't just say "if dial is at 30° flash ▲." It needs to specify what to do at every possible angle. A complete instruction sheet, indexed by dial angle.

Here's a clean way to imagine it. Each particle leaves the source carrying one secret number — let's call it the particle's secret tag, an angle pointing somewhere on a circle. When the particle reaches a measurement box, the box compares its dial angle to the secret tag. If the dial is "near" the tag (within 90°), the light flashes ▲. If the dial is "far" from the tag (more than 90° away), the light flashes ▼.

Now suppose the source pairs each particle with a partner carrying the opposite secret tag — one particle's tag is at θ, the other's is at θ + 180°. When Alice and Bob both set their dials to the same angle, the dial is "near" one secret tag and "far" from the other. Their lights are guaranteed to flash opposite ways. Same-dial result: anti-correlated, every time. Just like the experiment.

FIG. 2.2 A particle's "envelope" is just one number — a hidden angle. The measurement box compares it to its dial setting and flashes ▲ if the angle is "near," ▼ if "far." Two particles in a pair carry opposite angles, which means same-dial measurements always give opposite flashes. This much, at least, the model gets right.

§2.3Trying to build it — the cross-dial test

So far so good. Same-dial trials: 0% match, exactly as the experiment shows. But the experiment also tells us about different-dial trials, and that's where this model gets tested.

Let's walk it through. Imagine the source spits out a particle pair. The secret tag λ is some random angle — we'll average over many particle pairs, with λ pointing in every direction equally often. Alice sets her dial to 0° (straight up). Bob sets his to some other angle θ. Each box, independently, divides the circle into two halves and asks "is λ on my ▲ side?"

For any particular λ, the two boxes' answers depend on whether λ falls in the same half-circle as Alice's dial, the same half as Bob's dial, both, or neither. Walk through the geometry — it's a bit fiddly but elementary — and you find a beautifully simple result. The fraction of trials where Alice and Bob's flashes match is just the fraction of the circle where their two ▲-halves disagree, which is θ/180°. A perfectly straight line.

Dial difference 0° → 0% match (both halves identical)
Dial difference 90° → 50% match (halves overlap halfway)
Dial difference 180° → 100% match (halves swapped completely)
In between, a perfectly straight line

§2.4A straight line versus a curve

Now compare the secret-tag model's prediction (a straight line) to the actual experimental result (a curve, from chapter 1).

FIG. 2.3 The straight line and the smooth curve cross at three points: the dials matched, the dials at 90°, the dials opposite. Everywhere else, they differ. The model gets the easy points right and the in-between angles wrong.

So the secret-tag model isn't quite right. You might object: "okay, sure, that particular hidden-variable theory doesn't reproduce the curve. But it was a pretty crude one. Maybe a cleverer construction would work? Let the secret tag be not just an angle but a whole list of things. Let the rule be more complicated. Let the source weight different secret tags differently. Surely some elaborate enough theory can produce that smooth blue curve?"

That is exactly the question Bell asked. And his great achievement — the entire content of the next chapter — is to prove that no matter how clever the construction, no matter how many hidden variables you sneak into the envelope, no matter how exotic your rule, any sealed-envelope theory that respects the locality assumption ("Alice's result doesn't depend on Bob's dial") must stay inside a specific numerical bound. And the experimental correlations go above that bound.

Why this is the interesting question

Most "Einstein had to be wrong" stories collapse the moment you ask "wait, is that the only hidden-variable theory anyone tried?" Bell's argument is so important precisely because it doesn't rely on the cleverness of any specific theory. He proved a universal bound — a wall that any sealed-envelope theory of any complexity, present or future, must respect. The universe, as we'll see, walks straight through that wall.

← Previous1 · The puzzle Next →3 · The clever experiment