These two statement appear to contradict each other. Can the experimenter control the beam splitter or not? You said it had a 50% chance, which sounded like it was random.
If the experimenter can quickly change the beam splitter to "preserve information" after the photons are in flight, and even after the signal photon has hit it detector, then you have FTL or backwards-in-time communication. I'm certain this is impossible by current physics knowledge.
Otherwise it sounds like the other entangled particle experiments which seemed like FTL but in fact weren't because no information could be transmitted FTL.
No, the beam splitters are arranged beforehand with four detectors, two of which give information about the paths of the 'twin' photons and which slit the original photons (before being 'split') went through and two of which do not provide definite path information (i.e. could have gone through one or the other slit). So, the configuration of the apparatus in the experiment insures that both information and no information are available to the experimenter at the same time. The experimenter then has the choice to either preserve information or to erase it. Again, though, this can be done AFTER the first photon has reached its own detector and been registered in a special device as a wave (since no information available at that point). Remember, the first photon will always reach its detector before its 'twin' reaches its detector because it has a slightly shorter path to travel and so will be checked first, yet this result is able to be changed by the experimenter's choice AFTER it has happened.
