Hacker News Reader

Comment by btilly | Hacker News Reader

If you believe in the Everett Interpretation, then he is wrong.

The Everett Interpretation, aka Many Worlds, holds that both observer and observed are quantum mechanical systems. From this it follows that the act of observation does not produce a "collapse", but it does separate the observer into multiple observers that can't interact with each other again.

So Schrödinger's cat is in a superposition of alive and dead before the box is opened, and after it is opened the observer is in a superposition of one who saw the cat alive, and the other who saw the cat dead. It feels bizarre, but there are no contradictions. And it is what quantum mechanics predicts.

For those who believe in this interpretation, there is a simple test of quantum gravity. Set up 2 Cavendish experiments to measure gravity. Based on whether there is a click in a Geiger counter, choose which one to put a cannon ball next to. Measure both.

This has been done, and we only see the gravity from the cannon ball that we placed, and not the alternate location it might have been placed at.

But believers in other interpretations of quantum mechanics will disagree that this experiment tests anything at all.

View on HN

I’m a “believer” in the Everett Interpretation (on the grounds that it’s the simplest interpretation) but I can’t see how such an experiment could prove anything. This experiment seems to assume we can keep the whole apparatus including the Geiger counter and the mechanism to move the cannon ball from becoming entangled with the environment (decohering). This seems very far outside the realm of current technology.

It shows that when there is a superposition of quantum mechanical states, you also get a superposition of two different space-time structures.

The reason why the Everett Interpretation is needed for the experiment to be valid is that according to it decoherence results is a superposition of quantum mechanical states, and not a collapse. If you believe in another interpretation, there has been a collapse, and so the experiment demonstrated nothing.

Once decoherence has occurred I don’t think it’s meaningful to say that the system exists in a superposition of states. The multiple decohered branches of the wave function can no longer interfere with each other and so there’s no way for them to affect each other, which is why we call them separate worlds. Within each decohered branch there’s an experimental apparatus and a human that sees one specific outcome. This was Everett’s main insight that each observer in each branch experiences what looks like a collapse but is really just their brain becoming entangled with one or other of the measurements.

According to quantum mechanics, they can no longer interfere with each other.

But we are bringing gravity into the mix. Gravity works differently than the other forces. Do space-time structures also exist in a superposition of states? Or does QM work against an essentially classical gravitational space-time background?

Without a theory combining quantum mechanics and gravity, the answer is not obvious. The fact that the superpositions do NOT interfere gravitationally means that either there was no quantum mechanical superposition (collapse actually happened) or that gravity can also exist in superimposed states. If you believe in the Everett interpretation, then there must have been a quantum mechanical superposition, and this tells us something non-trivial about a quantum theory of gravity.

> This has been done, and we only see the gravity from the cannon ball that we placed, and not the alternate location it might have been placed at.

Doesn't this presume that your brain deciding where to place the cannon ball was an uncertain quantum event?

Doesn't it also presume quantum gravity does not exist?

Your brain is not deciding where the cannon ball is, it is predicting where it is and them making that prediction a certainty. That is the collapse of the wave function, turning a probability into a certainty.

No:

> Based on whether there is a click in a Geiger counter, choose which one to put a cannon ball next to.

In Everettian Mechanics, decoherence occurs long before the box is opened. The photons in the box interact with the cat. That entanglement with the environment which causes the branching of the wave function occurs then.

The observer that opens the box is already either in a branch where the cat is asleep (to borrow from Sean Carroll) or awake.

The interesting thing to come to understand is that probability doesn't actually seem to exist as a real feature in the universe. Rather, probability is a measure of what observers believe they will or are seeing. Probability, in Everettian Mechanics, is about perception of the universe, not a feature of the universe itself.

The cat is all entangled with the other stuff in the box, but by assumption the stuff inside the box is not entangled with the stuff outside the box.

But the observer is in all “probabilities” no?

There are observers in all probabilities of the experiment, but each observer only observes their world because decoherence interferes with the superposition of the other worlds. Worlds being the entanglement of the entire experimental setup spreading to everything else.

> If you believe in the Everett Interpretation, then he is wrong.

If.

> And it is what quantum mechanics predicts.

It is an interpretation of what the equations predict. It is very much not the only possible interpretation, so it is not "what quantum mechanics predicts".

> But believers in other interpretations of quantum mechanics will disagree that this experiment tests anything at all.

Within other interpretations of quantum mechanics, this does test nothing at all.

Where would energy come from for creating each universe?

I think this question stems from a fundamental misunderstanding of the Everett interpretation...the idea of "many worlds" (a phrase I do not favor specifically because of this very misunderstanding!) evokes this imagery of an entirely new universe BURSTING forth dramatically upon the collapse of the wave function, and as an entirely new universe is "created," the amount of extent energy is literally doubled (or multiplied by many times more). Indeed, where does this energy come from?

What the Everett interpretation suggests is that in the same way you are comfortable with the function f(x)=y having (infinitely) many values (you don't HAVE TO choose an x, ie the function doesn't collapse), the wavefunction simply is what it is and doesn't collapse either. What we observe as the collapse of a function is simply an artifact of being a conscious being that can only observe one value of the of the wavefunction at any particular moment.

None of these unique values (observables) of the wave function can interact with each other (in the same way the value of x=2 doesn't "interact" with the value of x=8 for f(x)=y), so energy cannot be gathered or duplicated at any particular point of the wave function, so this energy creation/duplication is no issue.

The question is a misunderstanding. In the pure wave (Everett) interpretation, there are no universes being created when the wave function branches. Rather, different regions of the wave function become separated (decohered) from each other. https://physics.stackexchange.com/questions/41588/many-world...

Sean Carroll, following Everett, puts it in the most concise form: (i) systems are described by wave functions, (ii) wave functions obey the Schrödinger equation [1]. "Many-worlds", universes, observations, observers, and so on become just entia multiplicanda [2], superfluities.

[1] https://www.youtube.com/watch?v=nOgalPdfHxM, one set of rules in quantum mechanics at 36:00; at 1:13:14 where does the energy come from? energy of the set of all universes is conserved.

[2] https://en.wikipedia.org/wiki/Occam%27s_razor

Maybe the implementation uses copy-on-write to minimise the amount of work required? ;-)

Why would it necessarily require any energy?

Energy is what we have to expend to effect change within a universe. There may be no requirement for energy to split into a branching multiverse.

"Where would energy come from for creating each universe?"

Why does it have to come from somewhere? What if it just is?

Among other problems. It's just bollocks on the level of "but you didn't say it couldn't be this, nanananana."

I think because of the abuse of "many worlds" in science fiction media (especially as of late) as a convenient plot device, people develop this idea that it's an unserious proposal wrt the foundations of physics. As far as I have read, it strikes me as perhaps the most parsimonious interpretation of QM out there.

Genuinely curious, what do you mean by "but you didn't say it couldn't be this, nanananana...?" The Everett interpretation isn't some fantastical notion spun up by a scifi writer...it is simply the consequence of removing collapse of the wavefunction as an objective event from the picture. And if it turns out our observations wouldn't be changed by removing this feature, then perhaps it was an extraneous feature in the first place!

Assuming many worlds, and many here means: enormous amounts, far exceeding the number of particles in the universe, is everything but parsimonious. There happens to be a model that fits some data, but that's it. It's a grotesque assumption to avoid a conflict in a man-made theory. It's a funny thought, but no more than that.

There's also nothing special about observing. Our consciousness isn't super-natural, so the idea is in desperate need of some other underpinning.

And probabilities: if this is one of many, many worlds, the distribution of events as we can observe them is heavily skewed. The next observations should follow a radically different pattern, unless you also assume that each split influences the probabilities of future events.

> Assuming many worlds, and many here means: enormous amounts, far exceeding the number of particles in the universe, is everything but parsimonious.

"Many worlds" is a misnomer. MWI is just wavefunction realism + unitary evolution. There's only one world, and really only one dynamical object: the wavefunction. It evolves according to some unitary operator, and that's the whole story. No splitting, no collapse, no objective classical transition, just quantum mechanics taken at face value.

> There's also nothing special about observing.

Yes, that's the MWI position.

Is it possible to observe anything at all in MWI? If yes, what does it mean to "observe something" in MWI? If not, is there any physical content in MWI?

Observations in MWI are just ordinary physical interactions. Specifically they're perturbative-regime interactions between the thing being observed and large thermalized systems (e.g. humans). From the perspective of the thermal bath, you get exponential suppression of everything but the eigenstates of the interaction Hamiltonian, which is why our observations "look classical".

Will other large thermalised system (e.g. rocks) also experience observations that "look classical"?

Another MWI proponent here is talking about how the "look classical" thing is "simply an artifact of being a conscious being that can only observe one value".

Is there something special about "large thermalised systems" (and/or humans)? How large have they to be to allow for "our observations"? Where is the boundary between the thing being observed and the observing thing?

MWI seems to still face most of the difficult questions - it not all.

> Will other large thermalised system (e.g. rocks) also experience observations that "look classical"?

The inner life of rocks is somewhat beyond our reach, I'm afraid. But they'll induce decoherence in the same way as a human, yes.

> Is there something special about "large thermalised systems" (and/or humans)?

Aside from being large and thermalized? No.

> How large have they to be to allow for "our observations"?

It's a continuum. The more internal degrees of freedom you have, and the more thoroughly they're mixed, the faster you'll decohere things.

> Where is the boundary between the thing being observed and the observing thing?

At the level of fundamental physics, there isn't one. "Observation" is an approximate and thermodynamic notion.

> But they'll induce decoherence in the same way as a human, yes.

From the perspective (?) of the (non-human) thermal bath, will that decoherence result in a single (diagonal, mixture) state or in a particular state of those N separate states that would "look classical"?

Decoherence doesn't make things "look classical" by itself - at least until you define what "looking" is.

Depends on whether you're talking about the "perspective" of the whole joint |Rock>|Interaction Eigenstate 1> + |Rock>|Interaction Eigenstate 2> ... system or just one of its components.

> Decoherence doesn't make things "look classical" by itself - at least until you define what "looking" is.

Of course; but that's true of every scientific theory. Decoherence solves the preferred basis problem, not the hard problem of consciousness.

> Decoherence solves the preferred basis problem, not the hard problem of consciousness.

Ok, I guess I misunderstood the scope of "There's also nothing special about observing."

It's not clear to me if you (MWI) would say that rocks had defined positions when nobody was observing them - or whether the question of things having definite positions (and the very existence of those things) wouldn't even make sense in the absence of consciousness.

> Ok, I guess I misunderstood the scope of "There's also nothing special about observing."

There's nothing special about the physical processes constituting a scientific experiment. They're unitary evolution like everything else. Whether there's anything special about conscious experience (for my money: obviously yes) is outside the scope of physics.

> or whether the question of things having definite positions (and the very existence of those things) wouldn't even make sense in the absence of consciousness.

If you want to be perfectly precise, the MWI does not contain discrete things at all, any more than the Earth objectively has discrete continents and seas. But the most accurate map is not always the most useful one.

If the MWI doesn’t contain the Earth, the seas or the molecules of water it may not be providing a very useful map until we add a lot of things to it.

Correct, it is an enormous amount...Under Everett, the "number of worlds" practically exists on a continuous spectrum, so basically an infinitude of worlds. But this doesn't violate parsimony in terms of building a model of reality. I don't have the greatest comprehension of the history of science, but I imagine the behavior of water and gas was not initially explained as being the collective behavior of moles (6.02x10^23) and moles of individual molecules. The parsimony of thermodynamics is measured by the simplicity of the underlying equations that describe their behavior, not the staggering number of atoms involved.

As for observation, I didn't suggest there was anything special about observing. In fact, this is part of what makes Everett among the most parsimonious interpretations: While some (not all) other interpretations are tasked with explaining the nature of observation (what counts as observation, how quickly does collapse propagate, etc), under Everett, there is no notion of observation at all. My point about consciousness is that, being conscious, we are forced to have a point of view which depends on where we exist in/on the wave function. I wasn't suggesting that consciousness has any actual effect whatsoever on the wavefunction.

I admit that the nature of probability is among the most difficult parts of Everett to wrestle with, though I'm not sure I understand why subsequent observations should follow a radically different pattern? It's still the same wavefunction with the same distributions as before...Schrodinger's equation is the same wherever you are on the wavefunction.

Many worlds just falls out of superposition and entanglement. Those happen in quantum systems, and we are made of quantum systems. To avoid that you have to stipulate something additional that avoids putting macroscopic objects like devices, cats and brains into superpositions.

> it is simply the consequence of removing collapse of the wavefunction as an objective event from the picture. And if it turns out our observations wouldn't be changed by removing this feature, then perhaps it was an extraneous feature in the first place!

> What we observe as the collapse of a function is simply an artifact of being a conscious being that can only observe one value of the of the wavefunction at any particular moment.

If we simply assume that our observations are as if a wave function happened we can indeed have our Schrödinger cake and eat it too.

Where do the eggs come from when you cut a cake?

This is a good summary. I like to put it the following way as an alternative: instead of imagining that you are the researcher, imagine that you are the cat.

Why would anyone believe such a thing?

Because it’s very straightforward and has less assumptions and weird features than most other interpretations, and it has seen the most advancements in quantum fundamentals research

Complete layman here -- does that theory say that every time a quantum effect does or does not occur, in fact both possibilities happen but in different worlds?

The entire universe instantly splits up in two for every quantum event anywhere in it?

Edit: I read more comments and no, it's not literally that.

Yes, it does mean that

These are just mathematical tools we use to make predictions. Assuming they represent objective reality and somehow prove an infinite number of universes doesn’t seem very scientific to me.

That's way too instrumentalist. The math also describes how atoms work and many other things. Modern physical understanding is based on those descriptions. Cosmology and nuclear physics would be useless without an understanding.

How do the predictions work if they aren't in some way modeling the way reality is? Why does the technology based on them work? Instrumentalism gives no answers to those questions. Science is about understanding the world, and using that to make predictions.

They work because they use human logic which has evolved to understand the universe just enough to “work” and where that logic doesn’t work we use probabilities to ensure we’re only calculating the odds and deriving from there. You can devise probabilistic systems of equations to describe stock prices and account for when you could be wrong, but this is not you describing some objective set of equations the market emanates from.

It working doesn’t mean it in anyway reflects some objective truth (if such a thing even exists.) It just breaks things down into small enough parts you can then model those parts and make predictions. This does not mean those small parts are “real” just that they are convenient ways to describe the flowing of energy for our purposes.

> atoms work and many other things

Atoms are still just a human created mathematical abstraction we use to describe and make predictions about matter. All there really is is energy and atoms are just a mathematical tool we use to study energy.

This is quite literally the oldest debate in natural science and it seems like one half of the debate just decided they’re right a few centuries ago and stopped caring about the fact that’s not something they have (or can) ever proved.

They’re tools, don’t get me wrong they are very useful ones, but the second you start pretending they’re anything else or that your interpretation of them is how reality really works you become just another religious zealot and that goes against everything science is supposed to be about.

> Atoms are still just a human created mathematical abstraction we use to describe and make predictions about matter. All there really is is energy and atoms are just a mathematical tool we use to study energy.

We can see atoms with electron microscopes. A mathematical tool to study energy doesn't say anything. A tool describing how chemistry arises from atomic structure does.

> This is quite literally the oldest debate in natural science and it seems like one half of the debate just decided they’re right a few centuries ago and stopped caring about the fact that’s not something they have (or can) ever proved.

Not proved, but imagine using this sort of argument against evolution being a true account of life's history on Earth. Of course it won't get everything right, but in general there is simply no other way to explain how life forms changed over time. Similarly, there's no other way to understand microscopic physics than quantum mechanics. Maybe a more complete theory combining gravity and QM would be more true, just like Newtonian physics was incomplete and superseded by relativity.

Yet earlier you asserted one side of the argument without proving it nor even acknowledging that there is a debate

I mean, that's obviously a huge matter of debate, but most of us physicists at least hope that our theories reflect objective reality in some way. Also, there are of course explicitly ontological aspects of quantum mechanics, such as PBR theorem and Bell's inequalities

Because that's what quantum mechanics in it's purest form tells us. Avoiding the many worlds requires tagging on extra stuff not in the equations.

Quantum mechanics tells us that, in order to predict the outcome of a measurement, we have to compute a specific probability based on the amplitude of the wavefunction.

We can explain this probability as some kind of collapse, or we can explain it as some measure of the number of observers making the measurement in parallel "worlds". Neither is inherently closer to the math.

> Quantum mechanics tells us that, in order to predict the outcome of a measurement, we have to compute a specific probability based on the amplitude of the wavefunction

Even this is already wading into interpretational waters. The math says nothing about whether a given POVM should be thought of as a measurement or an interaction (or both, or neither).

Sure, it's not part of the math, but it is part of the physics. There is a very clear distinction in the theory between "interactions", governed entirely by the Schrodinger equation (or its relativistic versions, or QFT), and a "measurement", where you have the apply the Born rule not only to predict what the measurement device will show, but also to correctly predict what will happen in a subsequent experiment on the same particle.

Without the Born rule, QM makes nonsensical predictions that don't explain what happens. Decoherence helps explain certain things directly from the math without appealing to the extra Born rule, but not fully, and not quantitatively (not the specific probabilities).

> Sure, it's not part of the math, but it is part of the physics.

"the physics" here is a specific interpretation

> Decoherence helps explain certain things directly from the math without appealing to the extra Born rule, but not fully, and not quantitatively (not the specific probabilities).

You can derive the Born rule using decoherence

> "the physics" here is a specific interpretation

No, "the physics" here is the way we are supposed to tie the math to physical observations. You can't explain all of our experiments if you don't make a distinction between interactions which preserve quantum properties and measurements which lead to classical observations. For a specific example, you can't explain the double-slit experiment with a detector at one slit if you don't make this distinction.

> You can derive the Born rule using decoherence

No, you can't. I have actively looked for such derivations and found none (at least none that are considered credible), and there is literature that explicitly says that this hasn't been convincingly done, e.g. [0]:

> My main conclusion is that there is no way to derive the Born rule without additional assumptions. It is true both in the framework of collapse theories and, more surprisingly, in the framework of the MWI. The main open question is not the validity of various proofs, but what are the most natural assumptions we should add for proving the Born rule.

[0] https://pages.jh.edu/rrynasi1/HealeySeminar/literature/Vaidm...

> the way we are supposed to tie the math to physical observations

Again, that's a matter of interpretation. The different quantum interpretations say different things here.

> No, you can't.

Yes, you can. Look up Quantum Darwinism.

PS. throwing "nuh uhs" back to a professional in the field is not appropriate. Get your respect in order.

> Again, that's a matter of interpretation. The different quantum interpretations say different things here.

In my view, a physical theory has two parts: a mathematical model, and a series of rules to relate the model physical experiments. For example, the theory of Newtonian mechanics isn't just F=m×a, it also includes the observation of how to measure the mass of an object and its speed etc.

Without these extra rules, you just have a mathematical theory, not a physical theory.

> Yes, you can. Look up Quantum Darwinism.

I did, and as far as I understand (e.g. from here [0]), it is an exciting candidate for explaining the Born rule based only on the other equations, but it's not fully accepted and it has its own limitations so far. Still, very interesting, thanks for pointing me to it.

> PS. throwing "nuh uhs" back to a professional in the field is not appropriate. Get your respect in order.

I did not throw a "nuh uh", I explained what I looked for and why I came to my conclusion, including a citation from a professor in the field. I think that's about as far from disrespecting you as I can get, while still disagreeing. Not to mention, I didn't know you were a professional in the field.

[0] https://www.quantamagazine.org/quantum-darwinism-an-idea-to-...

Why should we not just take it as a probability based abstraction for something we don’t have a true understanding of and move on? Anything else seems like weird theology to me.

Sure, it's perfectly valid to say we don't understand it - but the ambition of physics is to actually understand things like that. So if we agree that it is something that we don't know yet, then it remains a target for new theory development or new experiments.

Physics is ultimately not about predicting the outcomes of experiments. It goes farther than that: it is about coming up with a model of the world that humans can understand. The experiments and predictions are just there to validate that the model actually corresponds to the world, but they are not an end in itself.

And all of this speculation about what the Born postulate actually means has in fact lead to new experiments and new theories. If we had only ignored it, we probably wouldn't have discovered Bell's theorem, and we probably wouldn't have discovered decoherence.

1) Because physics has had an ontology, what there really is, since like Aristotle. Even if ontology changes, posing one has worked for roughly 2500 years to partially guide science. It really does inform all kinds of experiments, even thought experiments. BUT, you might be right the trade off you propose is more worth it

2) Reichenbach's principle says any correlation must have a cause. Even without an idea of an ontology like in 1) this looser principle is hard to give up as well. It explains the mystery behind Simpson's for example. And why without the causal model behind correlations, you could easily take the wrong drug for you specifically.

Again though, these are guiding principles and trade offs I am curious to see how we could progress without them

I really think it’s best to leave such metaphysical speculation out of it (or rather leave that to metaphysics.) Sure it could be right, and reality is definitely weird as hell, but it just seems like theology to me when you start pushing such things as fact when nothing says it is.

Don’t get me wrong I have weird beliefs about the universe too but I would always stress they’re my weird beliefs and not something we’ve proved. The many worlds interpretation is pretty mainstream at this point and it leads the masses to believe it’s either true or that scientists are batshit crazy. Not a good look imo.

As opposed to tagging on a literally incalculable number of extra universes for every possible quantum interaction.

Reply on news.ycombinator.com