[myridium]

Why is GR required? Already in standard QFT (Minkowski metric; flat spacetime; special relativity) the propagator 'violates causality'. Although that phrasing is misleading; it doesn't mean there's any logical inconsistency.

'The future affects the past' you might say. Yes, well, all information about the future is derivable from information about the present. So actually those 'future influences' are determined by the present. In the same way, 'locality' is almost a moot concept, because the analyticity of fields implies that the value of fields at remote locations is determined entirely by the field locally.

[bollu]

I don't follow the comment about analyticity. I thought fields were real analytic, which meant one needed to use a power series to expand them, which has a potentially infinite amount of information in the Taylor coefficients? So I don't see how the local data "effectively" captures non local information.

[naasking]

Maybe GR isn't required. It seems like the simplest path to understanding how this could work for those who aren't familiar with QM though.