I have a growing suspicion that familarizing yourself with quantum mechanics destroys your ability to solve logical puzzles. I base this on the fact that everyone who seems to know what they're talking about, is dodging the proposed exact method of using these findings to transmit information back in time, an on the subjective claim that there's nothing logical about quantum mechanics, and anyone claiming to actually understand it necessarily must be a total nutjob.
Anyway, if Alice and Bob compares their measurements of a particle pair, they will either find that the particles are most definitely NOT entangled, or that they might possible be entangled. Assuming we have a large batch of particles pairs - either the entire batch is entangled, or none of them are - when Alice and Bob compares measurements, they will either be able to say that the pairs are most definitely not entangled, or they'll be able to say with a certain amount of statistical certainty, proportional with the size of the batch, that the batch as a whole consists of individually entangled pairs.
Now, introduce Victor, who chooses whether or not to entangle entire batch of pairs, after the measurement has been made.
I realize this is not exactly what has been done, and would personally find it the whole experiment much more interesting if Victors random number generator had been replaced by a static "yes - DO entangle"-instruction stored in a system not accessible to Alice and Bob through conventional means, and check if Alice and Bob could confirm an entanglement which, although guaranteed to be about to take place, has not occured at their time of measurement.
However, assuming that the article and its findings are otherwise accurate, what's stopping one from sending information to the past using the method I just proposed?