ZapperZ said:
... I would also like to say something else here. In all of the so-called arguments I've seen with people who are disputing the "photon" picture, here's the ONE thing that I have never seen - the prediction on where the photon picture fail but the wave picture prevails. All I have seen are arguments that say to the effect that "well, wave picture can explain that too".
As far as I can tell, the photon picture has never satisfactorily explained the most fundamental properties of light, in particular, interference effects and the association of a
frequency with a particle-like photon. As Bohr said at the 1921 Solvay meeting:
“[The hypothesis of light quanta] presents insuperable difficulties when applied to the explanation of the phenomena of interference ... [it] excludes in principle the possibility of a rational definition of the conception of a frequency ...” Hendry, John, “The Creation of Quantum Mechanics and the Bohr-Pauli Dialogue”, D Reidel Publishing Company 1984, page 28
Some more recent facts that QM cannot explain satisfacorily are the results of the Bell test experiments. QM attempts of explain them simply by assuming a formula that fits selected observations, but it has never given any causal explanation for that formula and it seems to me no better than a myth. The photon model is causing people to ignore alternative explanations for the same results. The most realistic alternatives assume a wave model of light that allows the intensity of each "photon" (assumed here to be simply a pulse of classical light) to be split at a polariser. The raw data of some of the experiments can be fitted directly by such a model, simply assuming the classical version of Malus' law and then assuming "perfect" detectors that produce counts in exact proportion to input intensities. For other experiments we need to assume slight deviations from these assumptions, but still within the general picture of a wave model and an intensity reduction of
every pulse at a polariser, as opposed to the QM idea of reduction in probability of passage.
Clauser and Shimony in their 1978 report described the basic idea, though, as far as I can remember, they did not take the next step of considering the possibility that real apparatus might not convert input intensities
exactly into probabilities. It seems to me that the conversion will never be exact, due to the existence of "dark counts" at the low end of the scale and of saturation at the high end. I have not been able to find any discussion of this point, all the papers assuming the ideal case.
... At some point, they HAVE to diverge. The which-way-type experiments have clearly started to show this divergence.
Yes, but I think too many of the papers on the subject have been by theorists who have not stopped to think about the actual mechanisms involved in each bit of apparatus! The idea I had yesterday about the choice between two outputs of a beamsplitter perhaps being due to tiny variations in frequency was probably, on second thoughts, wrong, but there are other possibilities. The experiments do seem to show that the intensity is not split 50-50 but they do not show convincingly that it is the all-or-nothing effect given by the photon picture. I suggest that we simply have not tried hard enough to update the wave model to allow for recent observations.
Now people who are trying to push the wave picture needs to show and design an experiment that can support their arguments.
Suggested experiments:
1. Repeat a few Bell test experiments but this time investigating a range of different beam intensities, varying the beam intensities by at least two different methods. I predict that a wave model will fit the results better than the QM one, in which beam intensity plays no part.
2. Try splitting an unpolarised beam then splitting again a few times and see if the resulting detections really do follow the expected pattern. If we start with N photons per sec, QM would predict N/2 after one split, or slightly less by a factor e, say, so call it eN/2. The prediction after k splits would presumably be (eN/2)^k. I predict that the counts will fall off faster than this, since the intensity per pulse will reduce at each stage and will reach a point where it is not distinguishable from the dark count. [I must admit that I wish I had access to a lab so that I could find out the pitfalls in the above before committing myself!]
Cat
