On being the sole observer in MWI

[Talisman]

Hi all,

Suppose we have some particle in state $$|\Psi\rangle=\frac{1}{\sqrt{2}} (|U\rangle + |D\rangle)$$

It starts interacting with its environment, including Experimenter 1 (E1). From Experimenter 2's viewpoint, this can be represented as:

$$|\Psi\rangle=\frac{1}{\sqrt{2}} (|U\rangle \otimes |Env_U\rangle \otimes |E1_U\rangle + |D\rangle \otimes |Env_D\rangle \otimes |E1_D\rangle)$$

Because of decoherence, FAPP he can also consider it to be a classical mixture of the two outcomes I guess. But in principle there may be some interference experiment he could do that would reveal the superposition. On the other hand, once he interacts with the system, from his perspective the joint system (including himself) should be represented as one or the other "branch," say:

$$|\Psi\rangle=\frac{1}{\sqrt{2}} (|U\rangle \otimes |Env_U\rangle \otimes |E1_U\rangle \otimes |E2_U\rangle)$$

Doing an interference experiment no longer makes sense, even in principle. Any such experiment would have to include his "other self(s)," which is meaningless from "within the branch."

My first question is: is this considered well-accepted within MWI? If so, then from E2's perspective, he is the only "observer" in a particular sense (i.e., only when he interacts with the system can -- and must -- he stop treating it as a superposition.) This is straightforward and unsurprising perhaps, but I want to clarify.

My second question (really, observation that you should feel free to poke holes in) is a metaphysical one, so strictly speaking doesn't belong on this forum.

[Moderator's note: And for that reason, it has been deleted.]

 

[Talisman]

It seems I am the sole observer of this thread .

 

[PeterDonis]

is this considered well-accepted within MWI?

Yes; the "copy" of the observer in a particular branch can model the system using just the state in his branch, in order to predict all of his future observations. The rest of the branches don't matter.

My second question (really, observation that you should feel free to poke holes in) is a metaphysical one, so strictly speaking doesn't belong on this forum.

Yes, and for that reason, I have deleted it from the OP.

 

[Talisman]

Yes; the "copy" of the observer in a particular branch can model the system using just the state in his branch, in order to predict all of his future observations. The rest of the branches don't matter.

Thank you. This "sole observer (in my branch)" phenomenon is a bit unsettling, and I'm surprised it doesn't bother more MWI proponents. I can think of a number of reasons why it wouldn't (for example, it's not asymmetric or otherwise weird if you take a god's-eye view), but it does for me.

Yes, and for that reason, I have deleted it from the OP.

I'll try to formulate it less metaphysically.

Suppose we run a series of experiments with particles in state $$\frac{1}{\sqrt{2}} (|U\rangle + |D\rangle)$$ and we see a stream of |U> results. If Copenhagen is true, then the longer this keeps up, the more evidence there is that something (the apparatus or theory) needs adjustment. But in MWI, it makes no sense to say "it's unlikely that I'm in this branch"; overall, all of your copies are obeying the proper statistics, and one of you had to end up here. Therefore if MWI is true, then such a string of results should not cause you to rethink the theory.

Cheers,
T

 

[PeterDonis]

This "sole observer (in my branch)" phenomenon is a bit unsettling, and I'm surprised it doesn't bother more MWI proponents. I can think of a number of reasons why it wouldn't (for example, it's not asymmetric or otherwise weird if you take a god's-eye view), but it does for me.

You're not alone. In my experience this is one of the biggest issues people have with the MWI.

If Copenhagen is true, then the longer this keeps up, the more evidence there is that something (the apparatus or theory) needs adjustment. But in MWI, it makes no sense to say "it's unlikely that I'm in this branch"; overall, all of your copies are obeying the proper statistics, and one of you had to end up here. Therefore if MWI is true, then such a string of results should not cause you to rethink the theory.

I agree with this argument; in fact it's one I've tried to make before (not here on PF, but in other discussions I've had over the years). Basically this amounts to asking how you derive the Born rule (which is basically what we use in practice to justify inferring particular quantum states from measurement results) in the MWI. There is quite a bit of literature on this, but I don't find any of it convincing (although many MWI proponents do).

 

[Talisman]

Basically this amounts to asking how you derive the Born rule (which is basically what we use in practice to justify inferring particular quantum states from measurement results) in the MWI.

I see how it's related, but is it the same? Namely, even if they derived the Born rule, it only applies to individual measurements. It is still in no meaningful sense "unlikely" to find oneself in a world with all |U> results, and in no sense "likely" that one finds oneself in a world whose outcomes reflect said Born rule.

This sounds circular, so I will rephrase. Regardless of what probability rule is derived, it is in no meaningful sense probable that "the world" (i.e., my current branch) reflects that probability distribution. It seems one must allow meaning for statements like "what are the odds I'll end up on a given branch," which is what it set out to avoid.

I'm sure I'm just not thinking clearly enough.

 

[PeterDonis]

Regardless of what probability rule is derived, it is in no meaningful sense probable that "the world" (i.e., my current branch) reflects that probability distribution.

When I referred to papers that try to "derive the Born rule" within the context of the MWI, I meant deriving it for both directions, so to speak--for computing the probabilities for outcomes from the quantum state, and for inferring the quantum state from the relative frequencies with which outcomes are observed. The latter is essential if we are going to assign quantum states based on preparation procedures: for example, if we want to verify that an electron source puts out electrons in the spin-z up state, by passing them through a spin measuring device oriented in the z direction and verifying that they all come out in the up beam. The literature published by MWI proponents indicates that they recognize this is an issue that needs to be addressed.

 

[akvadrako]

Here is a thought experiment I sometimes suggest to help people grasp the subjective viewpoint in the WMI. I think it gets to the root of the problem, but can be explained with just classical concepts. I do ask you to assume some future technological advances, but I think it should be clear they are at least physically possible.

First, assume you decide you want two copies of yourself. So you take a trip down to the local cloning clinic, where they freeze your body and make a complete scan, including of your mind. The original is then discarded. Now they create two copies ##A## and ##B##, and wake them up. Both of them are future versions of you, but you will only ever become one of them. In this scenario the chances are equal, ## P(A) = P(B) ##. This generalises to more than 2 copies.

I don't think the above is hard to accept if you ignore the philosophical aspects. Almost the same thing is happening in the WMI, though there are a few differences; ##P(x)## is calculated differently and you can have outcomes with unequal likelihoods. Still, in symmetric cases all outcomes are equally likely and unsymmetric outcomes are handled in two popular derivations of the Born rule by rewriting them in terms of symmetric outcomes ##E##. So if ##A## is twice as likely, it's calculated like ## P(A) = 2 * P(E) ## and ## P(B) = P(E) ##.

 

[akvadrako]

But in MWI, it makes no sense to say "it's unlikely that I'm in this branch"; overall, all of your copies are obeying the proper statistics, and one of you had to end up here. Therefore if MWI is true, then such a string of results should not cause you to rethink the theory.

My main point is really to address this. If you try thinking about your subjective experiments in terms of classical cloning, it's easier to accept. Every time you perform an experiment, imagine multiple clones are created and they see all possible results, with the proper ratios. Science is still possible and a vanishingly small percentage of clones will persistently see haywire behavior.

 

[Talisman]

My main point is really to address this. If you try thinking about your subjective experiments in terms of classical cloning, it's easier to accept. Every time you perform an experiment, imagine multiple clones are created and they see all possible results, with the proper ratios. Science is still possible and a vanishingly small percentage of clones will persistently see haywire behavior.

Thanks, that is clear. My point is that in MWI it is (usually) considered not meaningful to say something like "I could have been a different clone," and thus meaningless to ask about the probability of being this one.

Instead, one seems to gets a stronger form of the anthropic principle (the solipsistic principle?): "the odds of me being here are 1, since otherwise I wouldn't be me." (This also leads to fun thought experiments like quantum immortality.)

I see there are various attempts in the literature to address this. For example, Sean Carroll's has written about "self-locating uncertainty"[0]. My problem with it looks to be in essence the one given here[1]. Namely, their argument rests on the supposed distinguishability of two ways of writing the very same state:

$$|O\rangle \otimes (|A\rangle + |B\rangle)$$
$$(|O\rangle \otimes |A\rangle) + (|O\rangle\otimes|B\rangle)$$

They claim that because we can write it the latter way, the observer should consider himself "branched" even before he encounters the system in question. This seems to be having one's caking and eating it too: because I haven't branched I avoid the problem of "which me is the real me?" but because I have "mathematically branched" I am allowed to introduce a concept of probability.

Sebens and Carroll’s self-locating uncertainty seems to be meant to be an uncertainty about an objective fact about reality: either I am the observer on the up branch or the observer on the down branch, but I’m not sure which. But then how can these be possible facts about reality, when the choice of branches is agreed to be not objective, but rather an arbitrary mathematical choice which – remember – can be made in many generally incompatible ways in generic models?

I feel myself being pulled into a rabbit hole, so maybe best I put this on pause :).

[0] https://arxiv.org/abs/1405.7577 "Self-Locating Uncertainty and the Origin of Probability in Everettian Quantum Mechanics"
[1] https://arxiv.org/abs/1408.1944 "Does it Make Sense to Speak of Self-Locating Uncertainty in the Universal Wave Function? Remarks on Sebens and Carroll"

 

[PeterDonis]

their argument rests on the supposed distinguishability of two ways of writing the very same state:

I don't think this is correct. The state of the observer has to be entangled with the state of the observed system in order for "branching" to occur. So, schematically, the process would look something like this:

$$
\vert O \rangle \otimes \left( \vert A \rangle + \vert B \rangle \right) \rightarrow \left( \vert O_A \rangle \otimes \vert A \rangle \right) + \left( \vert O_B \rangle \otimes \vert B \rangle \right)
$$

where ##O## without a subscript indicates that the observer has not yet observed the ##A / B## system; ##O_A## indicates that the observer has observed result ##A##; and ##O_B## indicates that the observer has observed result ##B##.

In other words, making an observation changes the observer's state. If you don't include that in your analysis you won't get the right answers.

 

[Talisman]

I don't think this is correct. The state of the observer has to be entangled with the state of the observed system in order for "branching" to occur. So, schematically, the process would look something like this:

$$
\vert O \rangle \otimes \left( \vert A \rangle + \vert B \rangle \right) \rightarrow \left( \vert O_A \rangle \otimes \vert A \rangle \right) + \left( \vert O_B \rangle \otimes \vert B \rangle \right)
$$

where ##O## without a subscript indicates that the observer has not yet observed the ##A / B## system; ##O_A## indicates that the observer has observed result ##A##; and ##O_B## indicates that the observer has observed result ##B##.

In other words, making an observation changes the observer's state. If you don't include that in your analysis you won't get the right answers.

Yes, this is precisely my problem with it. Have a look at the paper or the associated blog post. Here's an image:

The point is that in between steps ["decoherence" and "measurement complete"], the wave function of the universe has branched into two, but the observer doesn’t yet know which branch they are on. There are two copies of the observer that are in identical states, even though they’re part of different “worlds.” That’s the moment of self-locating uncertainty.

"Measurement complete" is the point of entanglement with the observer, i.e. his point of actual branching. The "self-locating uncertainty" step is just a rewriting step (distributing the tensor product), but they claim it has decision-theoretic consequences. It seems to be a sleight of hand: branching before really branching. I don't see how this is reasonable, and if it weren't for the paper I cited, I would end up assuming I simply misunderstood:

But then how can these be possible facts about reality, when the choice of branches is agreed to be not objective, but rather an arbitrary mathematical choice

As Carroll takes pains to point out, the main question here isn't how to get the specifics of the Born rule, but how to interpret the meaning of "probability." If you try to do it after the split ("measurement complete"), you face the meaninglessness of "I could have been another me," and so you are forced to find a meaning before the split / entanglement.

 

[akvadrako]

Thanks, that is clear. My point is that in MWI it is (usually) considered not meaningful to say something like "I could have been a different clone," and thus meaningless to ask about the probability of being this one.

Perhaps you are assuming some things that proponents of the WMI don't. In both setups, it's equally meaningful (or not) which version you will become: both of them. Before the split you know you'll have a 50% chance of waking up as A or B, but you also know versions of you will wake up as both. After the split you'll have self-locating uncertainty about which you are until new information arrives.

If you don't see how that's the same as in WMI, I'm not sure how the math can clarify it.

 

[akvadrako]

"Measurement complete" is the point of entanglement with the observer, i.e. his point of actual branching.

The point of actual branching is the decoherence step; that's when it physically happens. Self-locating uncertainty can happen well after that and in fact does, until you have gathered enough information (and broadcast it into the environment) to determine which version you have become.

 

[Talisman]

Perhaps you are assuming some things that proponents of the WMI don't. In both setups, it's equally meaningful (or not) which version you will become: both of them. Before the split you know you'll have a 50% chance of waking up as A or B, but you also know versions of you will wake up as both. After the split you'll have self-locating uncertainty about which you are until new information arrives.

I recommend reading the paper. "Self-locating uncertainty" refers very specifically to a point before one "wakes up as A or B." It is the key innovation of the paper. What you call "waking up" is there called "measurement complete," and at that point there is no uncertainty (with respect to the branches in consideration) by definition.

The point of actual branching is the decoherence step; that's when it physically happens.

Notice I said his actual point of branching, i.e., the point at which he becomes entangled with the system. Clearly the environment has branched during decoherence.

Please look again at his image I shared, and his accompanying description.

 

[Talisman]

If you don't see how that's the same as in WMI, I'm not sure how the math can clarify it.

Right, the point is that math cannot clarify it. The problem is that probability in the usual sense doesn't make sense in light of the standard MWI formulation.

Here's Sean Carroll on the issue:

For that matter, in what sense are there probabilities at all? There was nothing stochastic or random about any of this process, the entire evolution was perfectly deterministic. It’s not right to say “Before the measurement, I didn’t know which branch I was going to end up on.” You know precisely that one copy of your future self will appear on each branch. Why in the world should we be talking about probabilities?

He answers this with self-locating uncertainty, as mentioned above.

 

[akvadrako]

He answers this with self-locating uncertainty, as mentioned above.

That is one possible answer, but why do you think it's different than the cloning story? You should have the same concern in both cases and since the classical cloning example also clearly has an answer, so does the WMI story.

 

[Talisman]

That is one possible answer, but why do you think it's different than the cloning story? You should have the same concern in both cases and since the classical cloning example also clearly has an answer, so does the WMI story.

I can offer one sense in which they differ, but I'm not interested in defending it. In the classical case, one does not have to be God to measure the "probability" of each outcome: anyone, including the clones, can in principle take a measurement of the probability of each clone. In MWI, it is not possible even in principle for any of the clones to make such a measurement. (I'm sure we can reformulate the classical case in devious ways to make them more analogous, but I'm not trying to argue this point here.)

I'm just observing that this "one possible answer" seems to be the standard in MWI, which is why they publish papers about things like "self-locating uncertainty" (and why other distinguished physicists feel compelled to raise counterarguments, etc.).

 

[PeterDonis]

Here's an image

The decoherence step looks misplaced. If the observer is going to get entangled with the observed system (and environment), that has to happen before decoherence. But I haven't read through the papers you linked to in detail; I'll try to when I get a chance.

 

[Talisman]

The decoherence step looks misplaced. If the observer is going to get entangled with the observed system (and environment), that has to happen before decoherence. But I haven't read through the papers you linked to in detail; I'll try to when I get a chance.

The image matches the description in the paper / blog. He's just saying that the particle and apparatus encounter many unaccounted-for degrees of freedom before the observer interacts.

Think of observing the spin of a particle, as in our example above. The steps are:

1. Everything is in its starting state, before the measurement.
2. The apparatus interacts with the system to be observed and becomes entangled. (“Pre-measurement.”)
3. The apparatus becomes entangled with the environment, branching the wave function. (“Decoherence.”)
4. The observer reads off the result of the measurement from the apparatus.

 

[PeterDonis]

He's just saying that the particle and apparatus encounter many unaccounted-for degrees of freedom before the observer interacts.

If that's the case, then there are really two "measurements", not one, in the scheme being described. The first is when the apparatus becomes entangled with the environment, which decoheres the measurement made by the apparatus on the system. The second is when the observer becomes entangled with whatever he becomes entangled with in the course of reading off the result of the measurement--which doesn't seem to be described in any detail at all. As I said, I haven't read the papers in detail, but if we are looking for a place where a lot of handwaving is occurring, this would seem to be a good place to look.

 

[Talisman]

That part seems fine to me.

Anyway, I seem to have gotten bogged down in the details. The basic point was something like this: in Copenhagen if the experiment keeps pumping out the same result, a God-like view should suspect something is wrong. In MWI, noticing that there's one copy of one guy with unexpected results (from his perspective) is not necessarily a problem. This should be true regardless of how the details shake out.

Cheers!

 

[PeterDonis]

in Copenhagen if the experiment keeps pumping out the same result, a God-like view should suspect something is wrong.

In Copenhagen if the experiment keeps pumping out the same result, we conclude that the systems being measured are coming from the source in an eigenstate of the measurement operator--the one corresponding to the observed result. For example, if our spin measurer keeps measuring spin-z up, we conclude that our electron source is putting out spin-z up electrons. And if we could not do this, practically speaking, we could not do quantum experiments, because we could not use our observations to tell us what quantum states our sources were producing.

The question is what justifies this procedure if the MWI is true.

 

[Talisman]

In Copenhagen if the experiment keeps pumping out the same result, we conclude that the systems being measured are coming from the source in an eigenstate of the measurement operator--the one corresponding to the observed result. For example, if our spin measurer keeps measuring spin-z up, we conclude that our electron source is putting out spin-z up electrons. And if we could not do this, practically speaking, we could not do quantum experiments, because we could not use our observations to tell us what quantum states our sources were producing.

Right, this falls under "something" (the source or apparatus or theory or ...) being wrong. If you have a stream measuring every particle in z+, and an x-measuring device immediately after doesn't produce 50-50 results, something is wrong.

In MWI there will be some branch where any arbitrarily surprising result happens, and so from the broader perspective nothing is necessarily wrong.

 

[PeterDonis]

If you have a stream measuring every particle in z+, and an x-measuring device immediately after doesn't produce 50-50 results, something is wrong.

Ah, I see. Yes, under the MWI there should be a branch in which this occurs, but of course nobody has ever observed such a thing.

 

[Talisman]

Ah, I see. Yes, under the MWI there should be a branch in which this occurs, but of course nobody has ever observed such a thing.

Yeah, not here anyway .

 

[akvadrako]

In the classical case, one does not have to be God to measure the "probability" of each outcome: anyone, including the clones, can in principle take a measurement of the probability of each clone. In MWI, it is not possible even in principle for any of the clones to make such a measurement.

That's true but I don't see how it's relevant. One doesn't ever need to make those global measurements to make sense of their universe.

The origin of probability is the same in both deterministic systems and both have branches where unlikely chains of events occur. This demonstrates that the existence of such branches in no way gives evidence the theory is invalid. Nor should you expect to ever experience them in a finite amount of time.

I'm just observing that this "one possible answer" seems to be the standard in MWI, which is why they publish papers about things like "self-locating uncertainty" (and why other distinguished physicists feel compelled to raise counterarguments, etc.).

This is one popular answer, but not the standard one. Other approaches from Zurek and Deutsch are also often cited.

 

[Talisman]

Reviving this because there's something I'm unclear on.

What is the justification for saying that a copy of myself (on one branch) must model the system using only his branch? A straightforward answer might be: he is incapable of knowing, even in principle, what is going on in the other branch. For example, maybe something randomly quantum tunneled in, and I'd never know it.

Of course, such a thing can also happen even before I branch, at the last moment before I try to exhibit interference to demonstrate that the system is in a superposition.

I'm trying to make the original argument more airtight. Namely, that from the perspective of a copy of myself, my encountering the experiment is the point at which the state effectively collapses (in a way that's indistinguishable to me from an actual collapse).

 

[PeterDonis]

What is the justification for saying that a copy of myself (on one branch) must model the system using only his branch?

Because that's the definition of a "branch": the state of the copy of you in that branch is the one in which you observe everything else in the state that it's in in the same branch.

maybe something randomly quantum tunneled in

Quantum tunneling doesn't go from one branch to another. Nothing goes from one branch to another. Once branches are created, they never interact. (If they are still interacting, they aren't "branches" yet; the "experiment" is still in progress.)

such a thing can also happen even before I branch, at the last moment before I try to exhibit interference to demonstrate that the system is in a superposition.

I don't know what you're referring to here. If there is still interference, the branching hasn't happened yet; the definition of "branching" is "there is no more interference".

the perspective of a copy of myself, my encountering the experiment is the point at which the state effectively collapses (in a way that's indistinguishable to me from an actual collapse)

Yes; again, that's the definition of a "branch".

 

[Talisman]

Sorry, let me try again.

Suppose there are two experimenters, A and B. A encounters the experiment first, and branches. A must model things with his branch alone. B (who has not encountered the experiment) is still modeling it as a superposition. If B is sufficiently technologically advanced, he ought to be able to perform an interference experiment demonstrating the superposition, or even reverse all the entanglements.

But do MWIers generally agree that A could not perform such an interference experiment on himself? It sounds like you're saying yes. But here's Scott Aaronson channeling David Deutsch:

Here Deutsch had the following thought: suppose you could do a quantum-mechanical interference experiment on yourself. That is, rather than sending a photon or a buckyball or whatever through Slit A with some amplitude and through Slit B with some other amplitude, suppose you could do the same thing with your own brain. And suppose you could then cause the two parallel “branches” of your experience to come back together and interfere. In that case, it seems you could no longer describe your experience using the traditional Copenhagen interpretation, according to which “the buck stops”—the wave of amplitudes probabilistically collapses to a definite outcome—somewhere between the system you’re measuring and your own consciousness.

https://blogs.scientificamerican.co...ery-ridiculously-big-question-i-throw-at-him/

To run an interference experiment on yourself would require knowing precisely what's going on in the other branch (and still model it as two). That seems nonsensical to me (and equally meaningless to you), but I can't really find much literature that addresses it, and the above seems to suggest that some smart people disagree.

Do you know of any literature that addresses it directly?

 

[PeterDonis]

A encounters the experiment first, and branches. A must model things with his branch alone. B (who has not encountered the experiment) is still modeling it as a superposition. If B is sufficiently technologically advanced, he ought to be able to perform an interference experiment demonstrating the superposition, or even reverse all the entanglements.

And if that can be done, then according to the MWI, A never branched in the first place. The MWI says branching only occurs when the experimental result is irreversible. In other words, according to the MWI, if B can do something to demonstrate the superposition, then A, B, and the experiment are all one quantum system that does not branch at all.

here's Scott Aaronson channeling David Deutsch

Aaronson is assuming that your consciousness would somehow "branch" even though, according to the MWI, no branching occurs in this scenario (for the same reason as above). If there is still interference possible, then no branching has occurred.

The real question here is what it would be like to experience such an experiment. The answer is that we don't know, because we don't know how consciousness is implemented in the brain, or whether it is even sensitive to quantum-level properties.

 

[Talisman]

Hmm, I'm caught by the FAPP thing again.

Given that the evolution is unitary, it should be possible in theory for an extremely advanced B to demonstrate the superposition, right? I'm trying to figure out whether anyone would argue that it is impossible for (an extremely advanced...) A to do the same in principle.

If we're only talking FAPP then the "sole observer" question is irrelevant. All sufficiently complex systems are observers FAPP.

 

[PeterDonis]

Given that the evolution is unitary, it should be possible in theory for an extremely advanced B to demonstrate the superposition, right?

Not once the branches have decohered. Decoherence, at least in the current best version of the MWI (as I understand it) is what actually makes the branching irreversible. Or, more precisely, entanglement of the measured system and the measuring apparatus, followed by decoherence.

I'm not familiar enough with the literature on decoherence to know if it addresses the apparent contradiction between irreversible branching/decoherence and unitary evolution of the state as a whole, at least from the standpoint of the MWI.

 

[Talisman]

Not once the branches have decohered. Decoherence, at least in the current best version of the MWI (as I understand it) is what actually makes the branching irreversible. Or, more precisely, entanglement of the measured system and the measuring apparatus, followed by decoherence.

I'm not familiar enough with the literature on decoherence to know if it addresses the apparent contradiction between irreversible branching/decoherence and unitary evolution of the state as a whole, at least from the standpoint of the MWI.

From my reading, decoherence is always merely entanglement that is uncontrolled and irreversible "for all practical purposes," and thus branching is FAPP.

We ought to be able to overcome this by making our experimenters sufficiently technologically advanced. For them, decoherence and branching don't happen with a system as "small" as the ones we're discussing. That way, when we say that "A" branches, we just mean that she has become entangled with the system.

So, given that Alice can say "the electron is spin-up," can she also meaningfully model the larger system (that includes herself) as a superposition?

It sounds like the answer won't be settled here!

 

[AlexCaledin]

It's always the whole universe that is branching, together with the whole awareness - which, ironically, seems to be the best explanation of the "mysterious collapse" which emerges FAPP for this branch of awareness which has nothing to do with any other branches anyway whether they are "real" (in Heaven knows what sense) or not (each branch works strictly according to Einstein's relativity so it's not that spooky after all).