The paradoxes (misunderstandings) associated with observations in quantum mechanics, the quantum measurement problem (what happens when "a wavefunction collapses”) and action-at-a-distance (in the contemporary interpretation of entanglement) cannot be explained by a self-consistent theory in a classical “universe" model. But, it’s simple enough in a multiverse model. Note: a multiverse model needn't have any redundancies, unlike the Many Worlds interpretation. In a multiverse, purely causal and local explanations suffice and "quantum weirdness", e.g. action at a distance, is also eliminated and all physics becomes local.

• Considering the problems inherent to observation altering the observed… An obvious (to me) problem is that the events leading to any observation alway precede the observation. If events are not defined until they are observed, then the past can have no specific state until the state information reaches “the" observer. At a quantum level, this is most dramatically represented in delayed choice experiments. Those require the unobserved state information to remain isolated from observation… a very special state of events.

• The current model requires that there can only be one “real” observer for any event in the whole universe. Consider Schrödinger’s Cat. Suppose the experiment is done in a specially designed laboratory, enclosed by a Schrödinger-catbox-quality wall. The wall is information-proof. Within the lab, the experiment is performed. Outside the lab, everyone knows the experiment has been completed. Yet, for them, the cat and the scientists that have observed it, exist in an indeterminate state. So, are the observations of the scientists within the lab insufficient to collapse the wavefunction? When the lab is opened and the rest of the world can learn the results, is that when the experimenters assume a definite state, including their memories? That is when the cat’s state becomes real for the external observers.

• The absolute requirement that the wavefunction collapse extend across limitless distance, instantly, is obviously absurd to anyone familiar with special relativity. Simultaneity doesn’t exist except for individual reference frames. For any two events, (A) & (B), in spacetime there these three categories of observers exist: Observers for whom the events are simultaneous, observers for whom (A) precedes (B), and those for whom (B) precedes (A). Relativity requires that all three of these perspectives are equally valid. There are no alternatives: either action at a distance is a misinterpretation of quantum entanglement or special relativity is entirely wrong in its fundamental assumptions.

• Relativity requires that all events in spacetime, future and past, exist as an unchanging backdrop. This is a consequence of non-simultaneity. For events to be past for some observers, while still in the future for others, requires that the same events be available to all possible frames of reference*. [*Einstein hated this because he (correctly) saw it and human free will as mutually-exclusive. I don’t think he would have liked it any better had he considered it in a multiverse context.]

All of these problems are resolvable only in multiverse models in which every allowed outcome instantiates for some observers, though their world views must be mutually incompatible. Any model that accepts both relativity and quantum mechanics must include the fact that all possibilities exist, measurably (’Bell’s Experiment’ or ‘Inequality’) and that every observed state exists in spacetime, not only after observation, but prior to observation, for every observer.

To me, the obvious conclusion is that observation alters the observer, not the other way around. We have QM modeled outside-in when it must be viewed inside-out (science is based on observation, assumptions about the nature of reality beyond the observable are just assumptions). From the observer’s perspective. The indeterminacy of unobserved, though measurably real, events just shows that observation doesn’t define the event to anyone but the observer. Observation determines the event the observer gets. The states that the observer can entangle with. We don’t see the alternatives because that would require the observer to entangle with at least one causally-inconsistent quantum. Observable reality cannot violate any conservation symmetries, so it can only be observed to be causally-consistent, for each observer.

So far as I can tell, the only arguments against a multiverse interpretation are aesthetic and require discarding causality as a physical principle. Without the presumption of causality there is no logical basis for believing observations have anything to do with physical events. Basically, physicists have thrown away physics in order to preserve the classical universe model and just declaring that QM cannot be understandable.

The obvious advantages to an observer-centric interpretation of QM is that it's causal, consistent with Schrödinger's equation & all QM observations... making clear that action-at-a-distance is an inevitable artifact of perspective and removing any weirdness from even the delayed choice quantum eraser experiments. The model makes QM consistent with relativity, so long as spacetime is expanded to include all allowed events, though only those consistent with any observer's state will be observable by that observer (we each have access to "cross-sections" consistent with us).

This doesn't eliminate the problem of observed vacuum energy being different from calculations by about 15 orders of magnitude, but that's a different error... modeling a microscopic region of vacuum as a microscopic universe instead of as a tiny part of a 14 billion light year radius volume... but one problem at a time.