zonde said:
you have doubts that MWI has well defined "observer" at all
No, I'm just pointing out that you can't use the word "observer" the way it is usually used when talking about the MWI. The same goes for many other ordinary language words, as I have already pointed out.
zonde said:
let's assume we can speak about "observers" in MWI.
No, you can't just "assume" this if you mean by "observer" what is usually meant, because the usual meaning of "observer" makes assumptions that the MWI violates. So "assuming" that you can just speak normally about observers in the MWI means you are being inconsistent.
zonde said:
I call a "copy of the experimenter" in single branch an "experimenter".
Yes, but you can't call the copy "the" experimenter because there isn't just one of them after the measurement. Nor can you say anything about "experimenters" that assumes that there is only one of them after the measurement, or that one particular copy is "the same" as the experimenter before the measurement and the other ones aren't. Doing any of those things means you're being inconsistent.
zonde said:
the copy of experimenter observing particular outcome is distinct from another copy of experimenter that observes different outcome - these copies should be called distinct "observers".
Ok, this is fine. But then you actually have to do it and accept the implications. See below.
zonde said:
what is prediction? It's not that particular outcome happens objectively. Prediction is that group of observers observe the same outcome.
With the above meaning of "observers", so that "group of observers" means "the copies of a group of observers that are all in the same branch", yes.
zonde said:
Who is going to be the experimenter and what measurements he is going to perform to observe this "group of observers observing the same outcome"?
It doesn't matter; all of the observers in the group (in a particular branch, since that's what we're using "observers" and "group" to mean) can communicate with each other (since they're all in the same branch) and verify that they all observed the same outcome. This communication is a classical process and doesn't require any quantum branch points anywhere, so it doesn't raise any quantum measurement issue.
zonde said:
we don't have any superobservers that are not already included into prediction.
We don't need any. See above.
zonde said:
We can't compare this prediction with reality in a theory neutral way.
Yes, we can,
for this particular prediction. But that's because you restricted this particular prediction to one branch. The kinds of predictions of the MWI we
can't compare with reality in a theory neutral way are predictions about the relative weights of multiple branches (or, for that matter, the existence of multiple branches).
zonde said:
Where in the math is that relation?
In your choice of the measurement operator.