I don't understand. The proposal is to use no classic channel?
The idea is to entangle all the photons of a star? All the photons in a narrow band?
From another star you can not magically distinguish the entangled photons. You can only make some measurements, and compare it with the measurements of the photons are entangled with, and get some correlations.
For example, you can create a pair of entangled photons in A, send one of them to B. In B the other person/alien can "transfer" the entanglement to a thermal photon from B and send it to A. Now in A you can measure the polarization of the photon you saved and the photon you received from B. Depending of how the other person/alien "transferred" the entanglement, you will get the same or opposite results when you measure the polarizations. So B can use "same" or "opposite" to transmit a bit of information to you.
The difficult part is to pick the correct photon in with a star making a gazillion of photons in the background.
You can create a gazillion of pairs of photons in A and send one half of each pair to B and get later the photon form B. But you must match exactly each received photon with the correct photon of the other half of the pair that you stored. Matching the wrong photon will produce only random noise.
IIUC the idea is that B will not entangle one photon but a gazillion of photons. It is possible to create in B an entangled state where the photon stored in A has the same or the opposite polarization of a gazillion of photons in B and send them. So now it would be easier to pick one of them.
The problem is that someone E in the middle can also look at the photons. Without the photon stored in A, the other person E can only measure random noise, it's impossible to know if B was transmitting "same" or "opposite". The problem is that the intermediate person E will see that in the polarizer all the photons pass or not pass together, never 10% of them pass and the other 90% don't pass.
The intermediate person E will see that half of the time all the photons pass and half of the time all the photons don't pass. This is a clear sign that the photons are entangled. So the secrecy of the quantum communication will be broken, in spite the content will never be discovered.
* Matching the wrong photon will produce only random noise.
Not really related to what you have commented, but I wanted to mention it:
Back in Earth, our communication protocols usually have lots of redundancy. Whatever redundancy you can get into the channel, it usually gets inserted at whatever cost it takes - slowing down things a bit here and there - but ensuring a proper message gets to arrive to the other side (at the receiver).
I'm far (FAR) away from trying to sketch here an interstellar quantum comms protocol, but I find certain that if there aliens out there deploying quantum networks, they almost certainly will have redundancy at the level required (to overcome entropy), to get proper messaging in the used channel.
I agree, but for comparison, let's look at the laser than the NASA uses to measure the distance to the Moon. It has a pulse of 75mJ in 10ps, i.e. 7,000,000 Watts for a very short time, and they get back only a few photons. Alpha Centauri is 100,000,000 times more far away, so the signal is 1e16 times dimmer. You need a bigger laser or a lot of time just to be lucky to get a photon there.
Also, in comparison, the Sun has 3e26 Watts. That's distributed in the full spectrum, and the laser has a very narrow bandwidth. I can't find the exact number now, but from https://en.wikipedia.org/wiki/Laser_linewidth my guess is that it's 1/100000 of the energy is in the same band of the laser, that is 3e20Watts. Let's remove a few zeros to be sure, like 6 zeros, so my guess is something like 3e14Watts.
Comparing the guess of 3e14Watts of the Sun to the 7e6Watts of the laser. So it's not only difficult to see even a photon, there is a lot of noise in the background. I've measured signals like 1/100 of the noise level using a lock-in amplifier, and I think 1/1000 is possible, but 1/100.000.000 looks very difficult.
And now we must consider the quantum part, that is the interesting part of the post. If you generate a lot of pairs of entangled photons, after the trip you must match the photon that traveled with the photon you keep at home.
If you pick the right photon to compare, you theoretically get 100% of agreement, but my guess is that in a lab you will get 90% or less. After an interstellar trip, I'd be happy to get a 1% of correlation, and use a lot of redundancy and error correcting codes to fix it.
The problem is if you pick the wrong photon. You get perfect random noise. Absolutely no information. You can't fix it with redundancy. And with interstellar distances, you must generate a really huge number of photons just to be able to detect a photon in the other side of the communication.
The idea is to entangle all the photons of a star? All the photons in a narrow band?
From another star you can not magically distinguish the entangled photons. You can only make some measurements, and compare it with the measurements of the photons are entangled with, and get some correlations.
For example, you can create a pair of entangled photons in A, send one of them to B. In B the other person/alien can "transfer" the entanglement to a thermal photon from B and send it to A. Now in A you can measure the polarization of the photon you saved and the photon you received from B. Depending of how the other person/alien "transferred" the entanglement, you will get the same or opposite results when you measure the polarizations. So B can use "same" or "opposite" to transmit a bit of information to you.
The difficult part is to pick the correct photon in with a star making a gazillion of photons in the background.
You can create a gazillion of pairs of photons in A and send one half of each pair to B and get later the photon form B. But you must match exactly each received photon with the correct photon of the other half of the pair that you stored. Matching the wrong photon will produce only random noise.
IIUC the idea is that B will not entangle one photon but a gazillion of photons. It is possible to create in B an entangled state where the photon stored in A has the same or the opposite polarization of a gazillion of photons in B and send them. So now it would be easier to pick one of them.
The problem is that someone E in the middle can also look at the photons. Without the photon stored in A, the other person E can only measure random noise, it's impossible to know if B was transmitting "same" or "opposite". The problem is that the intermediate person E will see that in the polarizer all the photons pass or not pass together, never 10% of them pass and the other 90% don't pass.
The intermediate person E will see that half of the time all the photons pass and half of the time all the photons don't pass. This is a clear sign that the photons are entangled. So the secrecy of the quantum communication will be broken, in spite the content will never be discovered.