Your claim in the first link seems consistent with this recent experiment: https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.121.090503 https://journals

Your claim in the first link seems consistent with this recent experiment: https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.121.090503 https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.12... Indeed, it's wrong to think of "a" photon emitted at a particular time, and it takes either path. The emission is a quantum event with a probability distribution over time, capable of interfering with earlier or later versions of itself. I think if you use a fast optical switch to allow photons into the experiment for only a short window of time, and that window of time is less than the path difference, you won't see interference. That's different from a low-flux laser, because each photon's probability envelope is truncated in time by the shutter.