[Xcelerate]

No one denies that when a measurement is performed, the measuring system and the measured system become entangled (and the aggregate system containing both continues evolving unitarily).

> When you "measure" the two halves of an EPR pair what you are really doing is performing two measurements on whatever system produced the EPR pair to begin with.

Here's the problem with that. Consider the measurements of a pair of space-like separated entangled photons. A choice can be made about the details of these measurements by the experimenters at the last second possible. For example, the rotation angle of a polarizer can be randomly chosen the instant before a photon strikes it, yet correlations with this choice show up in the other measurement that was taken far away. (Note that these correlations still don't allow you to send information faster than light.) But the "which angle" information was never contained in whatever system produced the two photons.

[lisper]

>No one denies that when a measurement is performed, the measuring system and the measured system become entangled (and the aggregate system containing both continues evolving unitarily).

It's not true that no one denies this. Adherents of the Copenhagen interpretation deny it. They claim that measurement involves a non-unitary phenomenon called "wave function collapse."

> I can't quite tell exactly what you all are arguing about.

Exactly this. You may not realize it, but not everyone understands that measurement and entanglement are intimately related (in fact, the exact same physical phenomenon). In fact, some people vehemently deny it. There are even some card-carrying physicists who vehemently deny it.

> entanglement is not a phenomenon that can be explained classically.

That is certainly true (though I've met people who deny this as well).

[Xcelerate]

My apologies, I was editing my post while you were replying to it.

> the aggregate system containing both continues evolving unitarily.

The Copenhagen interpretation is that the measured subsystem collapses in a non-unitary way. That doesn't imply the overall system did. I have no opinion* on whether a measured subsystem collapses unitarily or not — the answer to that question does not currently seem to be experimentally testable. But it is testable that the overall system continues evolving unitarily during a subsystem collapse (well, at least up until the overall system is measured), and this has been tested and verified many times.

(* Ok fine, I do have an opinion. I don't think unitarity breaks — ever. I think it appears to break through the process of decoherence. But I'm certainly not going to claim that as fact unless experiment can prove it somehow.)