[pdonis]

> The Everett FAQ Q16 disagrees.

It looks to me like it "disagrees" mostly by mis-stating what other interpretations predict about the (highly idealized) thought experiments it describes. For example, it says that if we had a "reversible machine intelligence", we could use it to entirely reverse a measurement after it was performed. But on a collapse interpretation (such as Copenhagen), such a "reversible" measurement, like any reversible process, is not a measurement at all; it doesn't collapse the wave function. So the interpretations don't actually differ on this prediction; they just differ on how the result (which they both agree on) is interpreted.

The FAQ does raise one issue which might amount to a "difference in prediction", namely the issue of whether linearity is exact or not. The MWI requires that it is; any collapse interpretation requires that it is not. However, even here there is a problem about what constitutes a "prediction". The MWI claims that the other "worlds" exist, but we can't communicate with them; we can only detect them via interference effects, which basically means by keeping processes reversible, as above, so a collapse interpretation will just say that no collapse has occurred. But that is effectively the same as the collapse interpretation's prediction that the other worlds don't exist after a collapse has occurred. In other words, these two putatively different "predictions" are in fact experimentally indistinguishable.