Well, I have reason to believe that I am the real-life inspiration for the story (though don't ask me how Jack's system worked, as it is the EPR setup itself I can solve, not his imaginary system). I thought on first reading that it was just a coincidence, but I've now checked dates and places and there could be a causal connection. But that is another story. The point is that I have not (yet) been murdered by the editor of a famous journal. In fact, I have been welcomed warmly by quite a few experts - invited to give a seminar in Italy, give talks at conferences in Durham and Oxford. Exceedingly few would go so far as to say that they can see that my arguments prove Quantum Theory predictions actually wrong, but no-one yet has found a fault.
So how has this come about? How is it that experiments support QT so closely, when it is in fact wrong?
First let me tell you the basic secret. All we need is to forget that Einstein ever had the unfortunate idea (which he later had many causes to regret) that light was a particle. It is not even good enough to go along with Bohm-de Broglie theory and imagine light as a particle embedded in a wave. Light is a pure wave, an oscillation in a Maxwell-type field (which itself may need to be thought of in terms of a lower-level protofield, but this need not concern us).
Let's concentrate on atomic cascade ``paired-photon'' experiments, as these were the kind Aspect did in his famous experiments it the early 1980s in Orsay. (Some of my arguments do not apply to experiments involving actual indivisible particles, but no plausible experiments have yet been done using anything other than light.) Many experiments can be explained simply by realising that what a polariser does is not "allow the passage of a certain proportion of the incoming photons" but simply, as with macroscopic light, reduce the intensity. There is no need, from the point of view of EPR experiments, to think that the law governing this reduction is anything other than Malus' cosine-squared law, first established early in the last century. The crucial trick is to realise that it is not really possible to devise a detector that responds linearly to the intensity. In other words, the probability of detection cannot be even approximately proportional to the intensity except, perhaps, for a narrow range of values. This is really an age-old story. Whenever you get something that only happens with a given probability, there must be random inputs in addition to the "signal". Go back to Poincare in "Science and Method" (1909) and look at his descriptions of random processes. There has to be some way that the detector decides whether or not to detect the individual signal. An entirely reasonable way to model this, and one that is adequate for our purpose, is to assume that random electromagnetic noise is added then detection occurs if signal plus noise exceeds some threshold. The noise distribution is partly under the experimenters control, in that he can adjust the degree of screening and the temperature, but he cannot adjust it so as to achieve linearity. Thus it is a very natural extension of classical theory to say that we do not expect to find light at the "single-photon" level appearing to obey Malus' Law exactly.
If Malus' Law does not hold exactly for detection probabilities, then, for one thing, it becomes possible to reproduce the experimental results assuming the ordinary law that we multiply singles probabilities to get the coincidence probabilities - a "locality" rule that QT denies - and, for another, our real local model can now cause violations of the standard two-channel version of Bell's inequality.
"Aha", Aspect says (only he has not in practice replied to any of my letters), "but when we do a supplementary experiment (which we have to do in order to calibrate our polarisers) we find that we do get Malus' Law, to within 1.5% or so". I am sorry the story gets a bit involved here - you are welcome to skip it - but reality is very much more complicated in its detail (though simpler in its logic) than quantum theory allows. My story here is that it is entirely possible that the population of signals that you measure in the supplementary experiment is quite different from that involved in "coincidences". If we are thinking the physics out from scratch, why should we assume that the signals are emitted in pairs of equal intensity? This is the inevitable assumption of QT if it insists on particles. But waves could be of unequal intensities, and there could even be a general rule that high intensities for the A signal are always paired with below-average ones for the B signal. In this case, the high-intensity signals would contribute to the supplementary observations on the singles rates but would scarcely figure in the coincidences.
For more on two-channel experiments, see my paper in Foundations of Physics Letters (Vol 9, p357 (1996), "The Chaotic Ball: an Intuitive Analogy for EPR Experiments", which is also available as ref. 9611037 of the quant-ph archive at http://xxx.lanl.gov), and various others at http://www.aber.ac.uk/~cat.
Now other, "single-channel", experiments require more elaborate explanation. They may involve systematic failures of synchronisation, "enhancement" or, as I have discovered only recently, unjustifiable manipulation of the data (the subtraction of "accidentals" and other "corrections").
Perhaps that is enough about actual experiments to get you started.
Just what has been going on? Why has the scientific community insisted on accepting "evidence of non-locality", in view of the fact that most of the above ideas have been explored by others? How have the experiments managed to match QT predictions so closely if the actual formulae are wrong?
I shall attempt to answer the last question.
Quantum theory - at least, the relevant part, which is the part applying to "correlated particles" - had existed for decades without experimental test. It had become incorporated into University courses, the basis of many PhDs. Then, with Bell's help, in the mid-sixties it became possible to test it. But, as Colin Jack mentions in his Sherlock Holmes article, the minds of students (and, I fear, established scientists) have become fuzzy. Experimenters were given the brief of seeing if they could find any conditions under which Bell inequalities were violated. Nature was fairly willing to oblige. She provided a selection of possible artifacts, so that the experimenter had but to select the best combination. Experimenters have deluded themselves that they were producing nearer and nearer approximations to a perfect quantum state, whilst all the time they were in fact manipulating the many unspecified parameters (focussing, noise, voltages, beam strengths, polariser types, photomultiplier specifications etc. etc.) so as to get the "right" result. Nor can the theorists be held blameless: they have happily modified the theory when the experimenters suggested it was necessary. For after the first few successes it would have been a poor show to fail! No-one seems to have represented realism and poor old Einstein, though. Nobody seems to stopped to think whether some simple modification of classical theory might explain everything. Experimenters have been taught double-think, for many with whom I have corresponded have said that the do not believe anything non-local is actually happening.
Is it possible to "prove" I am right? I say "Yes". There are ``anomalies'' reported in Aspect's thesis that he cannot explain --- only small, but nevertheless I believe reproducible and hence not to be neglected. If the experimenters were to go back and look at all those preliminary runs in which different results were obtained, and to look carefully at the shape of that coincidence curve, I am totally convinced they would find other situations that Quantum Theory could not explain. If, though, as appears to be the case, established workers are unwilling to "waste their time" doing this, I have a host of experiments in mind that could be performed in any optics laboratory. All we need to do is play around with systems that are intentionally set "wrong", so that non-linearities and asymmetries are exaggerated. The logic will then be seen to follow real local rules, with the combination of polarisation direction and intensity acting as the primary "hidden variables", not any weird quantum "collapsing wave function."