Many Worlds Interpretation and act of measuring

[bhobba]

Added Later:

Since this concerns Quatum Suicide I thought I would do a preamble about it to lay the background.

Here is the description of it with the relevant bit posted:
http://rationalwiki.org/wiki/Quantum_suicide
What makes this interesting, is that if the many-worlds interpretation of quantum mechanics is true, then at the point at which a decay might happen, the universe splits in two — into one universe in which it decays and I die, and another in which it does not decay and I live. Assuming there is no afterlife, I will cease to exist in one universe but not in another. So, the argument goes, although there will be others who will exist in the universes in which I die, I will only ever exist in the universes in which I survive, so I will only ever observe the universes in which I survive. From my perspective, I will never die, I will always be saved from death by quantum indeterminancy.

One point to note that reduces it to absurdity is its symmetrical - you can just as well argue you never live. But it is irrelevant. Each observation causes the world to split and each subworld evolves on its own with no effect on the others.

A cleaner version would be if the appartus like a spin detector destroyes itself when it detects up and you supply it with a stream of particles to measure. There will always be a world where its destroyed, and a world where its not. But since they can never communicate - its makes zero difference.

Wallace knows, and details, the trick involved here in his book. It relies on a certain feeling we have about death that makes it philosophically complicated and the idea is to entangle those complications with this interpretation (see page 371). The key point however, again from the same page, and this time I quote 'I should stress though, the question, however interesting, does not bear on the epistemic status of the Everett interpretation'

Of course being a philosophy type that enjoys such questions he examines it a bit more and looks at those that disagree with his (and my) position. You can read it if that sort of thing interests.

The bottom line is it proves diddley squat, and to be blunt, is simply what I call philosophical waffle.

Anyway that's just a preamble to set the stage so to speak

I don't have a copy of Wallace's book at hand, but I suggest that you read again his final chapter entitled, "A Cornucpia of Everettian Consqeuences", particularly, "10.2 Exotic Consqeuences of Quantum Probability", "10.2.1. Cosmoloigical Probabilities and Anthropic Reasoning" and "10.2.2. Quantum Russian Roulette". I have no doubt that it will make my comments crystal clear to you.

I reacquainted myself with them last night.

They do not do what you suggest. For example, like me he gives the suicide argument short thrift (not as short as I do - he is slightly kinder) challenging it, correctly IMHO, as sensationalist and being of little consequence. The argument is symmetrical - you can equally have the experience of being dead. He is a bit kinder than me, conceding the experience of being dead is not that well defined - me I say you experience nothing.

Anyway none of this affects the MW interpretation which in no way depends on if rational observers are present or not.

Tegmark, another highly respected MWI expert, discusses a significant difference for an observer, under the MWI compared to other interpretations, in this paper:
The Interpretation of Quantum Mechanics: Many Worlds or Many Words?
http://arxiv.org/abs/quant-ph/9709032

I read it. A waste of time.

What it does is exactly what it says:
Common objections to the MWI are discussed. It is argued that when environment-induced decoherence is taken into account, the experimental predictions of the MWI are identical to those of the Copenhagen interpretation except for an experiment involving a Byzantine form of “quantum suicide”. This makes the choice between them purely a matter of taste, roughly equivalent to whether one believes mathematical language or human language to be more fundamental.

It does not prove what you said: 'An exception occurs when a singnificant fraction of possible outcomes do not contain the observer.'

The quantum suicide argument I already have dealt with - but even aside from that it is irrelevant to your point. But for completeness will go through it in detail. The paper says 'This time the shut-up-and-calculate recipe is inapplicable, since probabilities have no meaning for an observer in the dead state, and the contenders will differ in their predictions.' That is incorrect. It makes no difference if the observer is alive or dead to the predictions of the theory. Sure the assistant will not hear anything if they are dead - but QM doesn't concern itself with that - it only predicts the probabilities of quantum observations. The observation occurs at the apparatus that measures the spin - that is where decoherence occurs and the world spits - everything is classical from that point. Its exactly the same trap people fall into when discussing Schroedinger's Cat. The observation there occurs at the particle detector - the world is classical from that point on - the cat is never in a strange alive and dead superposition. Yes there will always be a world where the assistant is alive and hears the click - but so? It is utterly irrelevant to anything. And, this is the symmetry bit, there will also be a world where the assistant is dead and feels nothing. Again so? We have differing worlds where differing things happen. Big deal.

Now please, this time can you give a link to a paper that proves your claim, and give a summary of the argument, because I won't waste my time again.

Thanks
Bill

 

[bhobba]

It really doesn't matter what form the observer or observation device takes, since if in a significant proportion of the worlds, it doesn't exist then no such observation can be made.

Please describe to me a world where an observational device doesn't exist? An observation can occur, when for example, a few stray photons from the CBMR decoheres a dust particle and gives it a definite position. In fact observations are occurring around us all the time and that is how the classical world we experience comes about. The answer to Einstein's question is the moon not there when nobody looks is its being looked at all the time by the environment.

Or are you talking about the same observational device? In the quantum suicide the assistant is an observational device - not the one doing the quantum observing - that is at whatever is measuring the spin - but an observational device nonetheless. Its obvious whether or not things in the world are destroyed or not by the observation is of no relevance.

One can of course remove that constraint and have the observational device destroy itself if it detects something. Again all it means is in another world it can't make that observation. Big deal.

Thanks
Bill

 

[craigi]

Bill,

You now have it from me, Tegmark and you have it from Wallace, your chosen authority on the subject, in the very book that you read and insisted that he made no reference to it.

[mentor's note - edited a bit to keep it on topic]

 

[bhobba]

You now have it from me, Tegmark and you have it from Wallace, your chosen authority on the subject, in the very book that you read and insisted that he made no reference to it..

He specifically disagreed that quantum suicide is of any relevance. I explained exactly why its of no relevance.

[mentor's note - edited a bit to keep it on topic]

 

[carllooper]

If MW is purely deterministic, what determines what will happen in which universe? So If something can go left or right, what will determine which thing will happen to the universe that I am in now?

The whole MW thing seems to me they put another layer in between that mechanism, to kind of put a distance between us and that whole pure randomness thing?

For example it might still be completely random what will happen in which universe? Unless you can observe this mechanism, it does not really do anything about the whole pure randomness problem at all. You just moved it around a bit.

As I understand it each possible world (in a probability model) is an actual world in the many worlds model. All possible worlds exist. An infinity of such. The concept of splitting is somewhat awkward, for what would make a world split? If a world is deterministic then it has no way of splitting into two or more different worlds. However a slightly different take can resolve this apparent problem. We can pose a multiplicity of worlds that are partly the same and partly different in an infinity of varying ways. That which is the same in each world (the cat is alive in both worlds) is that aspect of each world which is not in fact two worlds (or sub-worlds) but one sub-world (A1 = A2 = A). Deutsch suggests such a definition of identity - that given a number of identical things we're really talking about just one thing - if indeed we do mean "they" are identical.

So we can propose (or select), for example, two worlds (out of an infinity of all possible worlds) where there is an overlap between these two worlds - where a part of each is identical, and everywhere else not so. Following Deutsch we can say the identical parts are one world (or sub-world) - and the non-identical parts belong to two worlds. Set theory can provide us with the requisite concepts for describing such a situation.

Within one of these worlds we have a situation where the non-intersecting part of such is completely free to interact, in a classical way, with the intersecting part, but in each world the outcome is different, due to what differed between each world.

So, for example, as I await the results of a lottery I entered, there is an infinite number of worlds in which I am awaiting the results of a lottery (so each of which contain the same sub-world) but in each world, outside the sub-world which is identical (or 'shared'), there is the sub-world which differs (some of which the lottery goes in my favour, some which don't). This unshared sub-world will eventually infect the shared sub-world (in which I'm awaiting the result of a lottery) and 'split' the otherwise shared sub-world. That part of a world in which I'm awaiting the lottery results (in an identical fashion), becomes no longer identical.

In one world I win the lottery and in another I don't, before which it was just one of me awaiting the results of the lottery (according to our definition of identity).

C

 

[mfb]

for what would make a world split?

Decoherence.

If a world is deterministic then it has no way of splitting into two or more different worlds.

The contrary, it is impossible to have unitary evolution of quantum mechanics without something that can be called "splitting". This is exactly the deterministic evolution, in contrast to nondeterministic ones like collapses where you just get one world.

 

[carllooper]

Decoherence.
The contrary, it is impossible to have unitary evolution of quantum mechanics without something that can be called "splitting". This is exactly the deterministic evolution, in contrast to nondeterministic ones like collapses where you just get one world.

The question was rhetorical rather then requiring an answer, and I answer it according to a version of the many worlds model. An alternative answer is decoherence but that doesn't address the question on how a many worlds model might otherwise resolve the same question. We need to distinguish between the "universe" and a "world". The universe is the set of all worlds. A world is one of such worlds. The evolution of the wave function is in terms of the universe, and in terms of such is deterministic. In the Many Worlds model we're talking about each world, on it's own being deterministic and classically so: ie. where the same conditions must give the same results. However the take I pursue is in relation to those worlds where the same conditions are not world wide nor entirely absent, but where each of the considered worlds are deterministic (in a classical way) yet provide a solution to how the "same" world can split. The answer being that the considered worlds are not entirely the same world - only in part - be it a very large part or a very small part.

We can propose an infinity of worlds that are entirely identical but following a Deutschean definition of identity we must drop the concept of there being many such worlds: if they really are identical they are really the same world: ie. just one world. (A1=A2=A3 ... = A). For every one of these singular worlds there will be an infinitely more worlds that are entirely different. But more interesting are not these entirely different worlds but those that are partly identical (and of course partly different). The identical parts become singular (following Deutchean identity) and the different parts become plural. This provides for a concept of deterministic splitting and no need to introduce randomness as a mechanism.

C

 

[RUTA]

Here are a couple papers that are interesting http://philsci-archive.pitt.edu/9542/1/Decoherence_Archive.pdf and http://arxiv.org/abs/1504.01063

 

[rootone]

Here are a couple papers that are interesting http://philsci-archive.pitt.edu/9542/1/Decoherence_Archive.pdf and http://arxiv.org/abs/1504.01063

Skimmed it, an interesting read, not math heavy, will go back to it I hope, marked it anyway.
"subjective uncertainty", hmm

Thanks.

 

[RUTA]

My understanding of these papers, correct me if I'm wrong, is that they are a response to a response. That is, Adrian Kent (and others) pointed out that splitting worlds in accord with the Born rule would leave most branches producing (empirically) the wrong probabilities. For example, if the correct probability is 50-50 for a given experiment, there would be many branches that would not see 50-50 outcomes. How would we know for sure that the probability we measure is the "right" one? The response to that was "subjective uncertainty" and these two papers are a response to that.

 

[stevendaryl]

My understanding of these papers, correct me if I'm wrong, is that they are a response to a response. That is, Adrian Kent (and others) pointed out that splitting worlds in accord with the Born rule would leave most branches producing (empirically) the wrong probabilities. For example, if the correct probability is 50-50 for a given experiment, there would be many branches that would not see 50-50 outcomes. How would we know for sure that the probability we measure is the "right" one? The response to that was "subjective uncertainty" and these two papers are a response to that.

But the objection applies equally to any probabilistic theory. If you a flip a coin some number of times, it's possible to get arbitrarily long sequences of heads-up. Strictly speaking, no finite amount of information can confirm or refute a probabilistic theory. That's true whether or not we consider many-worlds. In practice, we use a cut-off and declare that a probabilistic theory has been refuted if the chance that it is correct is below the cut off. But this leaves a possibility that we come to the wrong conclusion--accept a false theory, or reject a true theory--just because we by chance had an "atypical" run.

When we consider many-worlds, there will obviously be some worlds where the results of experiments will differ significantly from the predictions of QM, and in those worlds, the researchers will come to the wrong conclusion that QM is mistaken.

 

[stevendaryl]

But the objection applies equally to any probabilistic theory. If you a flip a coin some number of times, it's possible to get arbitrarily long sequences of heads-up. Strictly speaking, no finite amount of information can confirm or refute a probabilistic theory. That's true whether or not we consider many-worlds. In practice, we use a cut-off and declare that a probabilistic theory has been refuted if the chance that it is correct, given the experimental data^* is below the cut off. But this leaves a possibility that we come to the wrong conclusion--accept a false theory, or reject a true theory--just because we by chance had an "atypical" run.

When we consider many-worlds, there will obviously be some worlds where the results of experiments will differ significantly from the predictions of QM, and in those worlds, the researchers will come to the wrong conclusion that QM is mistaken.

^*Speaking like a Bayesian, that is. We can switch that around, and talk about the probability of getting those experimental results, under the assumption that the theory is true.

 

[Rajkovic]

Seriously, this idea of many universes, universe splitting in two, is far beyond the fantasy, more like fairy tale walt disney. lol.
You know how MWI should be called?
FOI = 'failure of interpretation'

That's the price of trying to understand the quantum world. I don't even know why you discuss it.

 

[RUTA]

You're right, we do talk as if there is a "weird" place in the universe where someone is always seeing heads when they flip a coin. Likewise, if all possibilities are realized with equal weight and the universe is infinite, then there are many places that don't agree with the 50-50 outcome of flipping a coin. And, we can't say by virtue of our experience, that indeed the probability is 50-50 just because that's what we observe. Yet, we're talking as if our 50-50 observation represents the "real" probability and those other "anomalous" regions are occurring according to our probability. It's exactly Kent's complaint with MWI, which is a legitimate complaint, so what we must *really* believe, despite claims otherwise, is that *every* region in the universe finds empirically the 50-50 outcome -- there are no "anomalous" regions. Otherwise, we can't do empirical probabilistic science.

 

[Rajkovic]

You're right, we do talk as if there is a "weird" place in the universe where someone is always seeing heads when they flip a coin

hahaha, where are this "weird place" ?
and from what I've read on many forums and wikipedia, the theory says that there are many Universes, and all other possibilities happen in OTHER universes, not in our universe.

"MWI's main conclusion is that the universe (or multiverse in this context) is composed of a quantum superposition of very many, possibly even non-denumerably infinitely[2] many, increasingly divergent, non-communicating parallel universes or quantum worlds."
Many-worlds interpretation - Wikipedia, the free encyclopedia

 

[RUTA]

hahaha, where are this "weird place" ?
and from what I've read on many forums and wikipedia, the theory says that there are many Universes, and all other possibilities happen in OTHER universes, not in our universe.

That's the point I was making ...

 

[bhobba]

"MWI's main conclusion is that the universe (or multiverse in this context) is composed of a quantum superposition of very many, possibly even non-denumerably infinitely[2] many, increasingly divergent, non-communicating parallel universes or quantum worlds."
Many-worlds interpretation - Wikipedia, the free encyclopedia

Wikipedia is usually reliable - but on that its wrong. The worlds are NOT in superposition. That's exactly what a mixed state after decoherence isn't.

Thanks
Bill

 

[stevendaryl]

You're right, we do talk as if there is a "weird" place in the universe where someone is always seeing heads when they flip a coin. Likewise, if all possibilities are realized with equal weight and the universe is infinite, then there are many places that don't agree with the 50-50 outcome of flipping a coin. And, we can't say by virtue of our experience, that indeed the probability is 50-50 just because that's what we observe. Yet, we're talking as if our 50-50 observation represents the "real" probability and those other "anomalous" regions are occurring according to our probability. It's exactly Kent's complaint with MWI, which is a legitimate complaint,

But I don't see how it's more of a complaint against MWI than any other interpretation. In any probabilistic theory, to make sense of the data in light of the theory, we have to assume that the results we have are "typical".

so what we must *really* believe, despite claims otherwise, is that *every* region in the universe finds empirically the 50-50 outcome -- there are no "anomalous" regions. Otherwise, we can't do empirical probabilistic science.

I don't see how the fact that some people experience anomalous results doesn't prevent US from doing science. We don't need to assume that everyone observes 50/50 outcomes from coin flips in order for us to reason about probability. As a matter of fact, probability theory implies almost certainly that there WILL be anomalous regions. If you flip coins long enough, eventually you'll have a run of 1,000,000 heads in a row. If the world lasts long enough, and people continue to flip coins, eventually there will be a generation in which nobody alive remembers ever seeing a coin land on tails. That's just probability. It isn't special to MWI.

 

[stevendaryl]

Wikipedia is usually reliable - but on that its wrong. The worlds are NOT in superposition. That's exactly what a mixed state after decoherence isn't.

I'm sure you already know this, but your remark sounds as if it's saying something different: If the universe starts out in a pure state, then it will ALWAYS be in a pure state. That's what unitary evolution implies. The appearance of mixed states comes from "tracing out" environmental degrees of freedom. But that's not something that the universe does, that's something that WE do.

 

[bhobba]

that's something that WE do.

No - that's something the theory implies from a theorist doing it - as with any deduction from a mathematical model.

There is a deep problem here about how a random environment comes about in such an explanation. That requires explanation in a deterministic theory. My hunch is QFT vacuum fluctuations have something to do with it, and may even provide a natural way to factor systems. Just a thought - as I have said many times I think this stuff requires a lot more research before firm conclusions can be reached.

Thanks
Bill

 

[RUTA]

But I don't see how it's more of a complaint against MWI than any other interpretation. In any probabilistic theory, to make sense of the data in light of the theory, we have to assume that the results we have are "typical".
I don't see how the fact that some people experience anomalous results doesn't prevent US from doing science. We don't need to assume that everyone observes 50/50 outcomes from coin flips in order for us to reason about probability. As a matter of fact, probability theory implies almost certainly that there WILL be anomalous regions. If you flip coins long enough, eventually you'll have a run of 1,000,000 heads in a row. If the world lasts long enough, and people continue to flip coins, eventually there will be a generation in which nobody alive remembers ever seeing a coin land on tails. That's just probability. It isn't special to MWI.

But, if you *really* believe that, then why do you believe the probability we find here is the correct one? No, we're clearly (although tacitly) assuming everyone and anyone in the universe will discover the same probability, which means there is no place that always sees heads.

 

[stevendaryl]

But, if you *really* believe that, then why do you believe the probability we find here is the correct one? No, we're clearly (although tacitly) assuming everyone and anyone in the universe will discover the same probability, which means there is no place that always sees heads.

Well, that belief is actually inconsistent.

 

[stevendaryl]

Well, that belief is actually inconsistent.

Let me expand on that reply. Suppose you believe that, for a sufficiently large number of trials, the relative frequency of an event will approach the probability. Then what's the probability of getting all heads when you flip a coin 1 million times? It's $P = 2^{-10^{6}}$. To say that it never happens is inconsistent with saying that it happens approximately once in $2^{10^6}$ times.

 

[bhobba]

Let me expand on that reply. Suppose you believe that, for a sufficiently large number of trials, the relative frequency of an event will approach the probability. Then what's the probability of getting all heads when you flip a coin 1 million times? It's $P = 2^{-10^{6}}$. To say that it never happens is inconsistent with saying that it happens approximately once in $2^{10^6}$ times.

Its this applied math thing. In many problems, and this is just one, you pick a very small number and assume FAPP its zero. It isn't really - but you assume it is. For example velocity is dx/dt. But you approximate it by delta x/delta t because that's all you can do.

Thanks
Bill

 

[stevendaryl]

Its this applied math thing. In many problems, and this is just one, you pick a very small number and assume FAPP its zero. It isn't really - but you assume it is. For example velocity is dx/dt. But you approximate it by delta x/delta t because that's all you can do.

Right. And making the assumption that sufficiently small probability events don't happen is a perfectly good heuristic, provided that we're only running an experiment a small number of times. But the assumption becomes inconsistent if the number of trials is very large. An example is the lottery. There's a 1 in a million chance of winning the New York State lottery. So if you have a small sample size, say the set of all your closest friends and relatives, you can get away with saying that nobody in that set is going to win the lottery--it's effectively a probability zero event. But if your sample size is the entire state of New York, it's obviously inconsistent to believe that nobody will win.

 

[bhobba]

Right. And making the assumption that sufficiently small probability events don't happen is a perfectly good heuristic, provided that we're only running an experiment a small number of times.

Remember conceptually you can make the very small number as small as you like so you simply choose one that is good for whatever number of times you are doing it.

Thanks
Bill

 

[RUTA]

If you believe you can find the "real" probability empirically, then you believe one of two things:

1. Anyone in the universe can do so
2. Only those in the "right" places can do so

I pick 1.

 

[atyy]

But I don't see how it's more of a complaint against MWI than any other interpretation. In any probabilistic theory, to make sense of the data in light of the theory, we have to assume that the results we have are "typical".

It depends on how much one buys the assignment of a probability to an amplitude within Many-Worlds, since one typically uses probablity, not amplitude, to assign typicality. In Copenhagen and Bohmian Mechanics, the amplitudes do pick up probability interpretations, so in those interpretations, one can argue that quantum mechanics describes what we see because the subsystems we observe are typical.

Also, there is still the preferred basis problem, since decoherence is not perfect. So one could choose a basis with cats that are dead and alive, and one still has to explain why there are no conscious observers in those worlds.

 

[stevendaryl]

If you believe you can find the "real" probability empirically then you believe one of two things:

1. Anyone in the universe can do so
2. Only those in the "right" places can do so

I pick 1.

But 1. is provably wrong. It's logically inconsistent to believe that.

 

[stevendaryl]

But 1. is provably wrong. It's logically inconsistent to believe that.

Suppose you have a million people, and they each flip a coin 20 times to figure out the probability of heads and tails. Then typically,

1. 1 person will see all heads. This person will assume that the probability is 1 of getting heads.
   
2. 20 people will see 19 heads and 1 tail. These people will assume that the probability is 95% of getting heads.
3. 190 people will see 18 heads and 2 tails. These people will assume that the probability is 90% of getting heads.
4. etc.

Around 185,000 people will correctly come up with 50% probability. A much larger number will come up with a probability between 0.4 and 0.6.

But it definitely will not be the case that everyone comes up with the observed probability of 50% heads.

 

[stevendaryl]

It depends on how much one buys the assignment of a probability to an amplitude within Many-Worlds, since one typically uses probablity, not amplitude, to assign typicality. In Copenhagen and Bohmian Mechanics, the amplitudes do pick up probability interpretations, so in those interpretations, one can argue that quantum mechanics describes what we see because the subsystems we observe are typical.

Also, there is still the preferred basis problem, since decoherence is not perfect. So one could choose a basis with cats that are dead and alive, and one still has to explain why there are no conscious observers in those worlds.

I think you're mixing up two different questions:

1. Are Born probabilities derivable from MWI?
2. Are MWI inconsistent with the Born probabilities?

I thought that some people were saying that MWI is inconsistent with the Born probabilities, because some branches/worlds will have relative frequencies that don't agree with the Born predictions.

 

[TrickyDicky]

But 1. is provably wrong. It's logically inconsistent to believe that.

Suppose you have a million people, and they each flip a coin 20 times to figure out the probability of heads and tails. Then typically,

1. 1 person will see all heads. This person will assume that the probability is 1 of getting heads.
   
2. 20 people will see 19 heads and 1 tail. These people will assume that the probability is 95% of getting heads.
3. 190 people will see 18 heads and 2 tails. These people will assume that the probability is 90% of getting heads.
4. etc.

Around 185,000 people will correctly come up with 50% probability. A much larger number will come up with a probability between 0.4 and 0.6.

But it definitely will not be the case that everyone comes up with the observed probability of 50% heads.

I don't think the second quote proves 1 wrong or logically inconsistent. IMO it introduces misleading assumptions that make the coin example inconsistent itself. You are conflating frequencies after a finite number of events are the same as probabilities. The probabilities are always an ideal limit at infinity. The person getting 20 heads will not assume probability 1 of getting heads if he knows anything about probabilities. We don't have to assume that the results we have are "typical". The actual results are specific outcomes, not probabilities, if finding 20 or 200 heads implied we had to assume the probability is one of getting heads probability would be a very different discipline, but nobody does conclude that, the concept of 50% chace for coins is an ideal limit, it is independent of the outcomes found as a concept, it is of course approximately(never exactly) validated by outcomes, it is a tendency, the only way to prove it wrong would be performing infinite trials which is impossible.

 

[stevendaryl]

I don't think the second quote proves 1 wrong or logically inconsistent. IMO it introduces misleading assumptions that make the coin example inconsistent itself. You are conflating frequencies after a finite number of events are the same as probabilities.

That's what I was arguing AGAINST. My point is that you CAN'T assume that relative frequencies will correctly tell you the probability.

The probabilities are always an ideal limit at infinity. The person getting 20 heads will not assume probability 1 of getting heads if he knows anything about probabilities.

Maybe not, but if you're trying to figure out whether you have a fair coin, or not, you would very likely decide that it was not.

 

[TrickyDicky]

That's what I was arguing AGAINST. My point is that you CAN'T assume that relative frequencies will correctly tell you the probability.

But then your example goes against what you are arguing for. And option 1 from RUTA is the only one consistent, certainly not 2.

Maybe not, but if you're trying to figure out whether you have a fair coin, or not, you would very likely decide that it was not.

Exactly, because the concept of probabiliy is independent of the specific outcome you get. Getting those outcomes make you think there's something wrong with your physical coin, not assume the probability is different from 0.5.

 

[atyy]

I think you're mixing up two different questions:

1. Are Born probabilities derivable from MWI?
2. Are MWI inconsistent with the Born probabilities?

I thought that some people were saying that MWI is inconsistent with the Born probabilities, because some branches/worlds will have relative frequencies that don't agree with the Born predictions.

I think we were both talking about (2). Is one able to say that branches are unlikely in which observers cannot verify the Born rule?