I think the observes are in super position and measurement collapses us and the world. Thus there are no nonlocal effects rather we just get rid of multiple universes which were averages together and pick one, or at least the subset of possible universes that correspond to that particular measurement. Thus instead of non local effects one selects from multiple existing solutions at present and discard the rest.

I think this makes a lot more sense.

The other alternative was we are in a simulation and it has variable levels of resolution (wave verses particle) and it will computer higher resolutions on demand (eg observers), but that these on demand increases in resolution affect the simulation going forward. This seems to be overly complex thus I prefer the first interpretation.

I am a layman in this area so I am probably wrong.

There must already exist a formal name for this? Isn't this where multiverses comes into play?