April 2, 2000
Revised May 16, 2002
In 1935 EPR (Einstein, Podolsky and Rosen) wrote a famous paper challenging a part of quantum theory. Some of the history behind this is included in my paper on The Tangled Methods of Quantum Entanglement. My very first paper on the "Bell tests"-- tests designed to check the EPR challenge -- is published as The Chaotic Ball (see updated version, October 2002). It is concerned mainly with the best-known "loophole" in these tests -- the "fair sampling", "detection" or "efficiency loophole" -- and was intended to be for beginners. In fact, all my papers are intended to be read with no formal training. All that is needed is a little time and concentration.
On the other hand, it must be admitted that you do need that concentration. Can I recommend readable book, just to start you off? The Mystery of the Quantum World, by Euan Squires (Adam Hilger Ltd, 1986). I'm sorry you can't get this off the Web.
There is a lot on the Web, but beware, as it is mostly wrong! You will find misleading ideas about how good the experiments were, and find "Bell tests" that are useless (Squires is no better in this respect). The ones used in real experiments are, of course, the only ones that matter, yet these are not even mentioned! I am sorry I have not yet produced a really simple paper just on these. The best proof I know for the test actually used in two out of Alain Aspect's three experiments is included as an appendix to my paper on the subtraction of accidentals, quant-ph/9903066. Aspect, by the way, is the person who did those experiments in 1981-2 that led to the current belief in the impossible.
The most important thing to know about Bell tests is that the majority of them are invalidated by the "detection loophole", also known as the "fair sampling assumption". In real experiments, it is necessary to allow for the fact that the detectors do not register every "particle", and to make any test possible auxiliary assumptions are needed (for a fairly comprehensive list, see Appendix A in my Tangled Methods paper, as well as quant-ph/9903066). The most popular tests depend on the assumption that the ensemble of pairs detected is a fair sample of those emitted. I should be surprised if any realist who has examined the facts thinks this is reasonable. In realist models, you only get a fair sample in very special cases, and these cases are most extremely unlikely to occur in the actual experiments. In the optical ones, an important factor is how the detectors respond as you change the input intensity. This is something that the people concerned carefully avoid investigating! I have not had reports back from the experimenters who did at one point rashly offer to check ... (See Challenges to Quantum Optics )
I repeat: the Bell tests used are not the perfect ones that Bell himself considered! These perfect ones are discussed in popular books and articles, but they have never been tested. The tests had to be modified because in all real experiments it is known that detectors have low "efficiencies". (In a realist wave model, incidentally, detectors are recognised as being a little too complicated to describe by just their efficiency. Efficiency is a concept that applies much better to photons than to waves. How can you talk of the probability of detection of a photon when photons don't exist? For more on the latter see my essay on the Nature of Light.)
Next time you read that realist models are as bizarre as quantum theory ones, I hope you will know better. A realist model that agreed with the quantum theory prediction and worked with perfect detectors would indeed be strange, but there is no need for this: detectors are not perfect. I confidently predict that if ever a perfect Bell test were performed it would not be violated, as the real world is local.
Return to front page