Bringing reality into the lab
Bell gave physics a measurable verdict; now someone had to actually run the Bell test. The recipe sounds easy and is fiendishly hard: make many entangled pairs, send them to two distant detectors, let each detector independently and randomly pick a measurement angle, record the outcomes, and tally the score S. If S climbs past 2, local hidden variables are dead. The challenge is doing this so cleanly that no boring loophole — a stray correlation, a slow detector, a leaked signal — could fake the result.
The earliest serious attempts came in the early 1970s, mostly using pairs of photons of light, which are easy to make entangled and easy to send flying apart. Results leaned toward quantum mechanics, but the apparatus was crude and the doubters had room to grumble. The breakthrough that captured the world's attention came at the start of the 1980s.
Aspect's elegant 1982 result
In 1982 in Paris, Alain Aspect and his team ran the first really convincing Bell test, now remembered as the Aspect experiment. Their cleverest stroke addressed an obvious worry: what if the two detectors somehow influence each other, one quietly telling the other how it was set? Aspect's group switched the measurement angles while the photons were already in flight, too late for any signal traveling at light speed to coordinate the two ends. The choice of question on Alice's side could not possibly reach Bob's side in time. And still the score sailed past 2, just where quantum mechanics said it would.
Plugging every leak: the loophole-free tests
Aspect closed one loophole, but determined skeptics could point to others. Two stood out. The detection loophole: if your detectors miss most particles, maybe the ones you happen to catch are an unrepresentative, conspiring sample. The freedom-of-choice loophole: maybe the random angle choices were themselves subtly steered by some shared past cause. For decades, every experiment closed some loopholes while leaving a crack for another. A stubborn local-realist could always retreat to the gap left open.
Then came 2015. Several groups, most famously a team in Delft led by Ronald Hanson, pulled off the first loophole-free Bell test — closing the major loopholes all at once, in a single experiment. The Delft team used entangled electrons in diamonds set 1.3 kilometres apart, with detectors efficient enough and separation wide enough to slam every door simultaneously. Other groups confirmed it the same year using photons. The verdict was unambiguous: the score beats 2. Local realism is finished.
- Locality loophole — closed by choosing angles fast and far apart, so no light-speed signal can sync the two ends (Aspect's idea, perfected later).
- Detection loophole — closed by catching a high enough fraction of particles that the sample can't be a rigged subset.
- Freedom-of-choice loophole — addressed by drawing the angle settings from sources too independent (even distant starlight) to share a common cause.
- 2015: all the major loopholes closed together in one go — the loophole-free Bell tests.
What we are forced to conclude
The experiments are in, repeated worldwide, and they all agree: Bell's inequality is violated, exactly as quantum mechanics predicts. So at least one of EPR's two common-sense assumptions is simply false. Either the world is nonlocal — distant events can be correlated in ways no local mechanism can explain — or particles do not carry definite pre-existing values, or both. The shorthand physicists use is that nature exhibits nonlocality, or more carefully, that local realism is false. Einstein's comforting picture of sealed-envelope reality is not how our universe works.
Sit with how remarkable this is. A question that sounded like pure metaphysics — does reality have definite, locally-carried properties? — was answered by stopwatches, lasers, and diamonds. The 2022 Nobel Prize in Physics honored Aspect, John Clauser, and Anton Zeilinger for exactly this: turning a philosophical wager into settled experimental fact. The quantum correlations are real, stronger than any classical story allows. Which leaves one last, urgent worry — if a measurement here instantly shapes outcomes there, surely we can send signals faster than light? That is the question the final guide lays to rest.