What's wrong with my teleporter? Reading on the no communication theorem and quantum teleportation, I have difficulty understanding the following thought experiment.

A light source is emitting photons at a regular intervals $T$ in a vacuum. The photons travel through a geometry that disperses it in the desired directions (something like a diffraction aperture).
Two observers in the same frame of reference as the source possess clocks synchronized with that of the light source.
Observer $D1$ is nearer the source and will choose whether or not to observe each photon by time $t1$.
Observer $D2$ further from the source, will observe every photon from the source.   
Observer $D2$ will not, though, be able to observe photons that have interacted at $D1$ because all their energy will be intercepted at $D1$ and will not propagate as far as $D2$.  


*

*Can the photon theoretically at least, be delivered such that the interaction probability at $D1$ is 100% and $D2$ is 100%?  

*Would it not possible to send a bit stream from $D1$ to $D2$ if observer $D1$
2.a. intercepts photons to indicate "FALSE" and
2.b. moves their detector, allowing photons to pass to indicate "TRUE"?
This question is, can observer $D2$'s measurements determine what $D1$ did?

*Would the transmission time of the bit stream from $D1$ to $D2$ be independent of their separation (dependent rather on the difference in their distance from the source)?
If those assumptions were correct the message transmission could be made to exceed $c$ - so I suspect assumption 1 is wrong.
 A: Your error is in assuming that if D1 does not measure a photon, it will necessarily be measured by D2. But that is not true! If you're sending half the probability amplitude to D1, and half to D2, than whenever D2 tries to measure a photon, it will see one exactly half the time. Similarly, D1 will see one exactly half the time. You won't be able to use this information to determine whether D1 has made a measurement, since D2 sees a photon 50% of the time when D1 has made a measurement, and 50% of the time when D1 has not made a measurement.
A: The issue is your first assumption. While it's true that the photon can have an amplitude to be in multiple places at once, it can only be measured to be in one place.
For example, suppose we split the photon beam so half the amplitude goes to $D_1$ and half the amplitude goes to $D_2$. Then if $D_2$ measures continuously, he'll only find a photon half the time, as he collapses the photon's superposition in position space.
Now suppose that we have $D_1$ measure continuously. Then she'll detect a photon half the time, and collapse the superposition. If $D_1$ saw a photon, the resulting collapsed state is 'definitely no photon at $D_2$'. If $D_1$ didn't see a photon, there is also wavefunction collapse, and the resulting state is 'photon definitely at $D_2$'. These two states each occur 50% of the time.
Then when $D_2$ measures, he'll find a photon precisely half the time, exactly the result he would have gotten if $D_1$ hadn't done anything at all. He's no longer 'collapsing a superposition', but there's no way he can tell that. Hence no information is transferred from $D_1$ to $D_2$.
You can work out the analogous situation where $D_1$ and $D_2$ receive unequal amplitudes. No matter what $D_1$ does, she can't affect the probability that $D_2$ sees a photon.

Technical aside: the setup you're considering is identical to the quantum mechanics of the Bell pair
$$|\Phi \rangle = \frac{|01 \rangle - |10 \rangle}{\sqrt{2}}.$$
Usually, the zeroes and ones refer to the spin states of two separate particles.  In this case, they refer to the presence of absence of a photon at $D_1$ and $D_2$. Your error is proposing to use the state 
$$|\Phi \rangle = |01 \rangle - |10 \rangle$$
which is not properly normalized.
A: This fails the "is there something quantum mechanical about this test setup" basic requirement for asking if quantum teleportation allows superluminal transport of information. 
Your setup could be equally realized by a stream of non-quantum boxes which have been carefully placed on conveyor belts traveling near the speed of light, with the requirement that we place balls inside either one of the other but never both. Since there is no way to use this setup to transfer information superluminally, it fails to be a counterexample to anything.
