Yes, you can't understand the message unless we have a mutually known
key... But if I'm only sending you one of two messages, 0 or 1, then we
can both "know what the message I want to send you is," having agreed upon
that subluminally, but then use this agreement to send superluminal
communication.
A very simplistic example of this would be something like "1 if by land, 2
if by sea." If we were to try something like this using QT, we could agree
that I would either send you a "land" photon with an already agreed upon
polarization or a "sea" photon with a different, previously deterimined,
polarization. Of course, if relativistic speeds/distances are involved,
that means that I would be able to tell you that the British were going to
attack by land *before they had even chosen to do so* from your
perspective... The causality problems this could create are enormous.
To simplify this even further, suppose we have only ONE agreed upon
polarization. "If I send you a particle with properties X, Y and Z, then
the British have attacked by land; otherwise, they have attacked by sea."
This agreement could have been made subluminally, but the information to be
transferred would be superluminal, and again, I would be able to tell you
how the British would attack long before they had actually done so.
Similarly, you don't have to know every word in the english language in
order to write words... You just need to agree on an alphabet.
-YEAH BABY I CAN'T IMAGINE THAT YOU'D WANT ME TOO-