Quantum mechanics is weird. Richard Feynman, one of its most important discoverers, said, "I think I can safely say that nobody understands quantum mechanics." Well, it just got weirder.
This article says that effects can occur before their causes. Don't ask me how that can possibly work; I have no idea. But the experiment seems to show that it's possible.
As always with entanglement, it's important to note that no information is passing between Alice, Bob, and Victor: the settings on the detectors and the BiSA are set independently, and there's no way to communicate faster than the speed of light. Nevertheless, this experiment provides a realization of one of the fundamental paradoxes of quantum mechanics: that measurements taken at different points in space and time appear to affect each other, even though there is no mechanism that allows information to travel between them.
I wonder if it's possible to change the experimental setup a bit: can Alice and Bob take their measurements and then tell Victor what to do? If Alice and Bob measure that Victor didn't entangle, but tell him to entangle and he does, something has to give. It's a little unsatisfying to say they must not have been able to communicate because of speed-of-light issues. Let them communicate and see if we can violate causality on the macro level.