I think that the double slit experiment shows that this isn't an issue... Since we are summing with all other noise in the universe, it doesn't matter exactly which photons came from my transmitter, if any at all, but the total count will still (probabilistically) reflect the the we're after...
On long distances you are talking about receiving single photons. And while we may view it as probabilistic, it will either interact with our equipment or it won't. If it doesn't, communication failed.
But yes, I wondered about that when writing my comment. Certainly photons are more than just "bullets" that are getting more and more sparse. So I got curious and it seems[1] that in context on our detector you can indeed put hard bounds on where the photon may be received. As far as I understand it, based mostly on the abstract, I do not really grasp it.
But ignoring the paper, I think I get your point. If photon has any chance of reaching the target (even lower than virtual particle spawning) and even if you don't know the future of the Universe (other photons hitting detector), you can use some complex encoding so that observing certain (allegedly otherwise unlikely) patterns allows you to say you received some message with some probability (which is what communication always is). If you have synced clocks. Simultaneity is relative though so there's also that.
I think that the double slit experiment shows that this isn't an issue... Since we are summing with all other noise in the universe, it doesn't matter exactly which photons came from my transmitter, if any at all, but the total count will still (probabilistically) reflect the the we're after...