>>11409204It only demands nonlocality if you presume that the outcomes of measurements are determined before a measurement is made.
It's basically a proof by contradiction. You start out by assuming that the outcomes of measurement are predefined. Then, you take different combinations of measurements, and show that the number of occurrences of different combinations just don't add up right when you try to trace them back to the predetermined states.
Pic is from this video:
https://www.youtube.com/watch?v=sAXxSKifgtUImagine you're measuring the polarizations of entangled photons, where the orientation of each detector is changed each time. In the image, imagine 'a' is measuring polarization at 0 degrees, 'b' is at 45 degrees, and 'c' is at 90 degrees. Each of these measurements has a binary, hit-or-miss outcome. If you presume the outcomes of measurements are predetermined, then each entangled pair should correspond to one of the eight binary combinations of a, b, and c. You can't measure exactly which one it is, but you don't need to. All that matters for the following argument is that you presume each entangled pair to be in exactly one of these combinations.
You might argue that each photon in the pair could have a different combination of a, b, and c. However, you can experimentally verify that, whenever the two detectors are at the same angle, the outcomes are always either both hit or both miss. So, assuming predetermined outcomes, both photons must be in the same combination.
Now comes Bell's argument. For each entangled pair, the detectors are separately reoriented. You record the number of occurrences of every pair of outcomes (e.g. a and b both hit, b hits and c misses, etc.). After collecting data, you trace each pair of outcomes back to the predetermined states that would cause them (there are always two such states for each pair of outcomes, corresponding to the two outcomes of the orientation you didn't measure) (cont.)
