I was reading a book about the same experiments commented by page russell, and reached the same doubts (or ida) as him. He described it the same that had popped to my mind. The original poster had a similar idea, but was not very good at describing it (and I didn't see the need to have two pairs of entangled photons on different paths, when one would be enough). I see the question remains unanswered...
This is not about trying to detect the spin and leveraging the fact that the other photon will have a specific spin if measured along the same axis. I'd have to phone him for confirmation and be bound by speedligtht. This about working with a stream of entangled photons (A and B, or A and B streams) and measuring one stream (for example, the position). The idea is to use a measurement on the A stream to cause the B photons’ (part of the B stream) (the entangled peers) probability waves to collapse. I do not care what the measurement outcome is, and I do not care about any correlation between individual A/B entangled photons pairs (for which I'd have to use the phone or a fax to find out). I do care that the probability waves of photons of the B stream (say all the entangled peers going south) are collapsed. If I tried to work this out of a single pair of entangled photons, it would tell me nothing. As soons as the A stream starts to be measured (this means I start measuring the position of the A photon stream), the B stream running through two slits (light years away) suddenly stops forming the interference pattern, and starts forming additive patterns. After a number of photons, I should be able to detect the pattern.
What page rusell did is point to an experiment that implies that the interference pattern of the B stream does stop forming when a polarizer acts on the A photons, and that this happens at a speed faster than light.
What I was expecting to find when I googled for this topic, is why this reasoning is flawed, why is it a paradox, and what is causing the missunderstanding. But it seems to me that while probably flawed (because it's too obvious to not have been addressed earlier) nobody has been able to explain why/debunk. Some asked for the hard data. As rusell pointed out, the experiments used to make the reasoning have already been done. If for another reason in mind, the outcome of the experiment will not change. What’s the point in rerunning them?
If you can collapse the probability wave of a photon instantly (or other particle) through a distant entangled peer; and if collapsing the waves a photons part of a stream preventing an interference pattern to form (and the inverse when not collapsed), then you can apparently pass information faster than light, provided you have created the stream with sufficient time in advance (ie: to communicate anything at speeds faster than light to Alpha Centaury, I'd first need to be in between the sun and AC, and create entangled particles directly pointing at each star. After two years have passed I could send the first "instant message").
So if look at it, this is not so much about “faster than light communications”, but about "delayed messages". I send the photons unknowingly of what the message will be. So in the future, I'll be able to alter the stream (measuring/not measuring at different intervals, and having waves collapsed/not collapsed streams at the other end) to "modulate" my message. You can see this as the message traveling at the speed of light, (entangled photons separating from each other at the speed of light) but having the final message defined at a later stage (when they arrive Earth and I dcide to measure/not measure the stream for a few seconds). But clearly cannot send a message at a speed faster than light to anywhere I haven't prepared in advance.
Now, if measuring a stream of entangled photons at one end did nothing to alter the interference patterns forming at the other end (ie, the rusell reference and the robert green book and other books, etc are all wrong), then all this would be nonsense.
--
Now the reason why I thought this was possible is that the information is not traveling faster than light really. It's traveling through time. Just like delayed quantum choice, a decision in the future seems to affect something in the present (or past), which seems add odds with common sense. It’s as if "the reality” knew what that future choice would be (deterministic), and this “superluminar” example sits in the same theater.
If you think it through a different lens, you could say the photons travel at the speed of light, and that makes them timeless. It's the same now and 10 years from now for them. So from the photon perspective, both things occur instantly (being born, and being detected). For us, they may travel for a millisecond, a second, a year, etc. So it's no wonder that what for us "seems to be the future", for the particles its always the present. So the particle and the entangled peer always knew what we’ll do in the future…it’s the same exact moment for them.
When you use the entangled particles streams to modulate a signal (stream B collapsed/not collapsed as shown by interference pattern, when the A stream is measured/not measured) the same happens. When the particle (actually the entangled pair) is born, it already knows if it will fall in the "measured photons" bucket or the "not measured photons" bucket. Accordingly, the "spooky action at a distance" does not mean information traveling faster than light, but it means information traveling across time, which means that for us, a change here instantly produces a change there (non local), but what really happened is that the photons already knew our (we believe…future) choice. Particles such as photons can predict their future with 100% accuracy. And that means they know what we'll do in what we call the future.
So the faster than light information that rusell and the original poster described enable me to ask a friend in Alpha Centaury how much is 2 x 2, and have him answer my call instantly (or is it a few seconds or minutes?). But of course, I'd have needed to send entangled photons from a location half across Alpha Centaury and our Sun. And also agreed what the collapsed/not collapsed sequences would mean (putting something at half the distance would take >2 years and the entangled particles would take another 2 years to reach Alpha and Sol. And an agreement on what the sequences means would also take >4 years to reach Alpha Centaury.
Basically, you could see the superbowl live from any place on the galaxy provided you planned it with sufficient time in advance (1 year in advance, 1 light year distances, etc). The photons ability to “read their future” would allow the exact contents of the message to be determined in what it’s for us "the future".