Re: MITM impossible
The theory seems to be that if a 'Man-In-The-Middle' intercepts the message, by capturing the photons, they then have to send on photons of their own to try and make the other end think nothing happened. (Note that the message is encoded by the photons themselves as with any optical data transfer; the message is not encoded by the quantum states, as they are impossible to fix).
But they won't be able to replicate the quantum entanglements in the photons they send on, so if the sender measures the paired photons they kept, and the receiver measures the paired photons they received, and compare notes, they can tell if the message was intercepted and a fake copy passed on. However, this assumes they then have a perfectly secure way of exchanging the state data to do the comparison...
Here's an interesting question - if you can measure the state of a photon without impeding its passage, does its quantum state collapse such that the dual-slit experiment stops working? i.e. This would mean you could tell if a photon has already had its quantum state collapsed by observation. So I presume not; that any such non-destructive attempt will, like the dual-split experiment, just yield a 50:50 result.