Jarek wrote:To test Pr(A=B) + Pr(A=C) + Pr(B=C) >= 1 we need "three coins" and independent experiments consisting of measuring at least two of them.

Measuring all three, the above inequality has to be satisfied by statistics of such independent experiments.

(1) There are only two coins in any EPR-Bohm type experiment, not three. (2) Even if there were three coins, they cannot be measured in any EPR-Bohm type experiment, in any possible world. (3) Statistics and probabilities are for the conmen (like Gill) who want to deceive the world, and their use in the Einstein-Bohr debate obfuscate the requirements of local-realism.

Jarek wrote:The question is: what can happen while we measure only two so that this inequality is not satisfied? - what is possible e.g. in models with Born rule like QM or MERW.

Please share if you know other models allowing for that.

Models with Born rule like QM and MERW (whatever that is) are manifestly non-local models. But Nature is manifestly local. And all quantum mechanical correlations, probabilities, and statistics are reproduced strictly local-realistically in the two papers I have linked above, namely in (1)

https://arxiv.org/abs/1405.2355 and (2)

https://arxiv.org/abs/1806.02392.

minkwe wrote: ... you will see how naive Bell was in coming up with this charade.

Bell's argument is indeed misguided, if not naive, and the whole saga of Bell's so-called "theorem" is indeed a charade. The sooner it is exposed as such and put to rest the better.

***