[lisper]

> Even decoherence researchers agree that docoherence theory does not do this

Yes, but they are wrong. And it's not hard to see that they are wrong.

The crux of the argument is that the state predicted by QM:

|S1>|A1>|O1>|E1> + |S2>|A2>|O2>|E2>

where S is the system being measured, A is the measurement apparatus, O is the observer, and E is the environment, is not what is observed. What is observed is either:

|S1>|A1>|O1>|E1>

or

|S2>|A2>|O2>|E2>

neither of which is the predicted state above. Except that it is because |S1>|A1>|O1>|E1> is what is predicted to be observed by an observer in state |O1> and |S2>|A2>|O2>|E2> is what is predicted to be observed by an observer in state |O2>. It is not that the prediction is wrong, it is that you, a classical observer, are not sufficiently omniscient to see both observations. You can only see one or the other. And this too can be explained, though by quantum information theory rather than decoherence theory. In order to be a classical observer it is necessary to be able to copy (classical) information. The only way to do that is to discard some of the (quantum) information contained in the wave function. Being non-omniscient (i.e. being unable to directly observe a superposition) is a necessary precondition of being a classical observer.