Yes, I know the thread name sounds like something you might see on the cover of Discover (or sadly, Scientific American these days). Hey, I could have called it: Proof that the future changes the past! Because that is precisely what happens. The subject relates to entanglement swapping. In typical entanglement swapping setups, two pairs of photons are polarization entangled (often labeled A & B and C & D). The pairs are created closely together in time from different (preferably) sources. The trick is to swap the entanglement so that A & D are polarization entangled, and they will violate a Bell Inequality. See Figure 1 attached for a general diagram, and Figure 2 for a more detailed schematic. Both of these are from a paper realizing this experiment: http://arxiv.org/abs/quant-ph/0409093" [Broken] 1. The reported results show violation of a Bell Inequality for photons (A & D) which each traveled over a 1 km fiber. These are post-selected based on results of a suitable Bell state measurement (BSM) on photons B & C. The BSM is what causes the previously otherwise independent photons A & D to become entangled. Note that in this particular experiment, the same pump laser is used to created both sets of photons pairs (A & B and C & D, each pair coming from different PDC crystals however). However, subsequent experiments have performed the same trick using 2 different lasers that have been pulse synchronized. 2. Here is where it gets interesting. According to QM, the path lengths of the various measuring devices can be set to any length in principle. And length implies time as well. The following can be made arbitrarily long (or short), for instance: a. Location of Alice measuring the polarization of photon A. b. Location of Bob measuring the polarization of photon D. c. Location of Charlie performing the Bell state measurement (BSM) on photons B & C. d. The distance separating the laser/PDC setup creating photon pairs A & B and the laser/PDC setup creating photon pairs C & D. 3. So here is what we do: we pick a. and b. to be short and c. to be long. Thus photons A and D have their polarization measured shortly after being created, and long before the BSM is performed on B & C. And we have d. be long as well so that A and D are created and destroyed (upon measurement) far before their light cones could have ever overlapped. See Figure 3. Charlie is designated as the person doing the BSM, and it takes a joint measurement of the Bell state to cause A & D to become entangled via entanglment swapping. Charlie's BSM does not itself directly identify the actual polarization results of A and D, which Alice and Bob may select to be anything. For our example, let's assume Alice and Bob always set their observations to be at the same angle but something Charlie does not know. In that case, Alice and Bob will see perfectly correlated/anti-correlated results for any A & D pair that were entangled by Charlie's BSM. Alice and Bob have already performed their measurements on A & D by the time B & C arrive at Charlie's BSM. Now, Charlie can decide whether or not to actually allow his BSM to entangle A & D. If he DOES perform the BSM, A & D WILL become entangled. They will then show either perfect correlation or perfect anti-correlation (depending on the actual Bell state measured, which occurs at random). Alternately, he can choose NOT to perform the BSM, and A & D will NOT become entangled. And yet A & D don't exist any longer. They were detected and destroyed BEFORE Charlie decides whether or not to perform the BSM. A & D were never in each other's light cone, and were in fact created from separate and independent laser sources. And yet, subsequent analysis can show that A & D were entangled in the cases when Charlie decides to perform the BSM (a Bell Inequality will be violated). In the cases when Charlies does NOT perform the BSM, a Bell Inequality will NOT be violated. Thus: Charlie's future decision appears to change the past by entangling photons which no longer exist and have never been in contact with a common light cone. Not an easy trick to explain by any physical mechanism, and in fact I don't believe any interpretation (other than perhaps the retrocausal/time symmetric variations) would even be able to predict this situation. Such is predicted by workman-like application of our poor ol' abused friend: Standard Quantum Mechanics. So I guess the reports of its demise and uselessness might be premature.