I think what I said applies when you take a first-person point of view. If you're a participant in a quantum suicide experiment, then if you expect a collapse interpretation to be an accurate description of reality, you should expect to eventually be hit by a bullet and die. But both MWI and QIT predict that you will continuously notice that the gun doesn't fire. The difference is not in the point of view taken, it's in the fact that the parts of the wavefunction that contain a (from first-person eye-view) future version of the participant actually are there.