The real puzzle is why no-one has carefully read the best chapters in John Bell's book "Speakable and unspeakable in quantum mechanics". Chapter 16 even has a careful discussion of the statistical issues involved (randomized settings, sample averages, standard errors). Most important of all, Bell describes the experiment which needs to be done, which still has not been done.
Excerpts from "Speakable and unspeakable"
Chapter 13
Atomic-cascade photons and quantum-mechanical nonlocalityAvoiding internal details for the moment, consider just a long black box with three inputs and three outputs. The inputs are three on–off switches – a master switch in the middle and a switch at each end. The outputs are three corresponding printers. The one in the middle prints ‘yes’ or ‘no’ soon after the start of a run, and the others each print ‘yes’ or ‘no’ when it ends. While the switches are ‘off’ the box restores itself as far as possible to some given initial condition in preparation for a run. The master switch is then operated and left ‘on’ for a predetermined time T. At time (T – δ each of the other switches may or may not – depending, for example, on random signals from independent radioactive sources external to the black box – be thrown to ‘on’ for a time δ. The length L of the box is such that L/c >> δ, where c is the velocity of light. So the operation of a switch at one end would not, according to Einstein, be relevant to the output at the other end.
We will consider only runs certified by a ‘yes’ from the middle printer, and not mention it any more. It just guarantees, as will be seen, that the internal process gets off to a good start. Let A (with values ± 1) denote the yes/no response of the left printer, and B( ± 1) likewise for the right printer. Let a ( = 1,2) denote whether or not the left switch is operated during the run, and b( = 1,2) likewise for the right switch. With sufficient statistics we can test hypotheses about the joint probability of A and B given a and b: ρ(A, B | a, b).
Going into the black box, we could find what is sketched (at the ‘Gedanken’ level) in Fig. 1. Only the centre and one end are drawn. The other end is the mirror image of the first. An oven provides a beam of suitable atoms in their (j, P) = (0, +) ground states. A pulse of laser photons γ00 is activated (after a predetermined delay during which remote equipment is alerted) by the master switch. This excites some atoms to a certain (1, –) level (Fig. 2). Most of these decay straight back to the ground state, but some cascade back with emission of photons γ0, γ1, γ2. Some such cases are identified by a γ0 counter C0 with a suitable filter. And, for some of these, photons γ1 and γ2 go towards detecting equipment at the two ends of the box. Filters F1 and F2 pass only the correct photons γ1 and γ2, and signal when they absorb wrong ones (i.e., they are a little more articulate than filters commercially available). Veto counters V identify events in which photons go off in other unwanted directions. Only the operation of counter C0 and the nonoperation of the vetos V and F1,2 authorize the middle printer to issue a ‘yes’ certificate for the event. Photons γx and γ2 then go towards distant detectors, Cx and C2, preceded by linear polarizers. These latter are set to pass polarizations at angles to the vertical controlled by the corresponding switches.
The firing or nonfiring of counters C1 and C2 authorizes the corresponding printers to print ‘yes’ or ‘no’.
The heart of the matter is a strong correlation of polarization between photons γ1 and γ2, dictated by the spins and parities of levels A and C in Fig. 2. Because the atom has initially and finally no angular momentum, the photons can carry none away. For back-to-back the photons can carry none away. For back-to-back photons this means a perfect circular polarization correlation – left-handed polarization for γ1 implies left-handed γ2, and right-handed γ1 implies right-handed γ2. Allowing also for parity conservation this translates into an equally strong linear polarization correlation: a given linear polarization on one side implies the same polarization on the other. In detail, in the ideal case of small opening angles and fully efficient counters, the probabilities of the various responses of C1 and C2 according to quantum mechanics are ... [including an image of the atomic decay sequence envisaged by Bell] ... 2 sqrt 2.
Fig. 1. Centre and left-hand-side of Gedanken set-up.
Fig. 2. Suitable atomic-level sequence.
Chapter 16
Bertlmann’s socks and the nature of realityConsider the general experimental set-up of Fig. 7. To avoid inessential details it is represented just as a long box of unspecified equipment, with three inputs and three outputs. The outputs, above in the figure, can be three pieces of paper, each with either ‘yes’ or ‘no’ printed on it. The central input is just a ‘go’ signal which sets the experiment off at time tx. Shortly after that the central output says ‘yes’ or ‘no’. We are only interested in the ‘yes’s, which confirm that everything has got off to a good start (e.g., there are no ‘particles’ going in the wrong directions, and so on). At time t1 + T the other outputs appear, each with ‘yes’ or ‘no’ (depending for example on whether or not a signal has appeared on the ‘up’ side of a detecting screen behind a local Stern–Gerlach magnet). The apparatus then rests and recovers internally in preparation for a subsequent repetition of the experiment. But just before time t1 + T, say at time t1 + T – δ, signals a and b are injected at the two ends. (They might for example dictate that Stern–Gerlach magnets be rotated by angles a and b away from some standard position). We can arrange that cδ << L, where c is the velocity of light and L the length of the box; we would not then expect the signal at one end to have any influence on the output at the other, for lack of time, whatever hidden connections there might be between the two ends.
Fig. 7. General EPR set-up, with three inputs below and three outputs above.