Scott Aaronson on Many Worlds
Sep. 18th, 2012 08:37 pmI've heard quite a few people proposing that the many worlds interpretation of quantum mechanics is "obviously right": many physicists, and very vociferously Elizier Yudkowsky. Unfortunately, other respected physicists have derided the whole thing as bunk.
Fortunately, Scott Aaronson, who I generally trust to know what he's talking about on quantum physics has obligingly written about it: http://www.scottaaronson.com/blog/?p=1103 (via cyphergoth).
It definitely wades into physics that I can't follow. It wades into physics that he doesn't know, for that matter.
However, as best I could tell, a few highlights. I found the best quote in his first comment:
I feel the part of the MWI argument that I found important to have been settled with that comment: I know many physicists will still resist, but purely in terms of "does this sound true to a professional physicist", I trust Scott's interpretation.
Of course, that still leaves the meat of Scott's post, where he disagrees with MWI advocacy, but I didn't find that as important. He thinks MWI advocates are jumping the gun in saying it's obvious and settled. I mention a couple of his major points below, but I don't think I can do them justice (I'm not sure I understand or agree) so you're better off reading his post directly.
Turtles all the way up
The meat of the matter seems to be, photons experience decoherence. As do electrons. As do buckyballs. But what about cats? Human brains? Is there a size bigger than which things don't experience decoherence?
There seem to be three possibilities:
1. Yes, it works all the way up, just like it looks like it should
2. No, there's a physical effect which stops at a particular size. Scott makes the point that we've no idea how to apply superpositions to physics including non-trivial gravity fields, which he thinks may be relevant. I think this is his objection, though it doesn't make sense to me?
3. There's some other explanation, such as hidden variables, or magic powers of consciousness which make waveforms collapse, which have been expressed by many very venerable theories of physics and many of the best physicists from the 1930s onwards, but are pretty clearly Not True, unless we're prepared to accept dubious propositions like "consciousness has magic properties not explained by the rest of physics" or "spooky information travels faster than the speed of light in untestable ways". This is the sort of thing which looked more reasonable than not 50 years ago, but lots of mathematical proofs have proved lots of things are impossible (given "reasonable" assumptions), and MWI advocates have spent a lot of time refuting.
Scott agrees with MWI advocates that by now it pretty much has to be 1 or 2. He thinks there's still a reasonable possibility of 2, given experiments we'd like to do but haven't done yet, but I don't understand why.
Occams Razor
Scott obligingly demolishes a common flawed application of occam's razor, namely applying it in an extremely literal sense to minimise the number of physical objects, rather than the complexity of physical laws.
As he points out, tables are made of atoms. Are they made of atoms all the time, or only when you look at them closely enough? The first theory has fewer physical objects (since a lot of the time you just have "table", not a lot of separate atoms), but the second has much simpler physical laws (since you don't have to define what "someone's looking at it" means).
But most people would agree that it only makes sense to assume the table is made of atoms all the time. Ie. we should minimise the complexity of physical laws, not the number of physical objects. This is the interpretation of Occam's Razor which (a) is what scientists usually mean by it and (b) has worked historically -- for instance, the discovery that stars were stars was unlikely in the "number of objects" way, but fine in the "simple laws of physics" way, and was right.
Physicists who agree with the concepts of MWI but disagree with MWI
He points out that some physicists basically agree with the "there's a wavefunction, end of story" interpretation, but think "many worlds" is the wrong concept for it. I think that's fine, they're probably right. I am using MWI as a handy label, I'm happy to use something else. I don't endorse all the trappings of MWI as the best way of explaining stuff (they may be, or may not be), I intend to refer to the sort of things these people would agree with.
For the record, the idea is that there's a waveform, and "worlds" is a convenient way of viewing that waveform as overlaid copies of mostly-separate waveforms that only interact with themselves, not with each other.
So the question "when, exactly, does the universe split into two 'worlds'" has the answer "at some fuzzy point during the process when the split becomes irrecoverable, ie. the worlds no longer interfere". But the fuzziness is in the concept that things should be divided into worlds, not in the underlying physics. MWI advocates blame this fuzziness on MWI detractors and vice-versa, but it's primarily terminological.
Philosophical Quibbles
Scott raises an objection about how we perceive consciousness if our brains may be nearly-duplicated in adjacent "worlds". To me this sounds like "I don't like the conclusion so I don't want the argument to be true" wishful thinking. But Scott explicitly agrees that that sort of wishful thinking is pointless. So it seems like he's saying something else, but I'm not sure what.
Edit: Fix spelling mistake in name.
Fortunately, Scott Aaronson, who I generally trust to know what he's talking about on quantum physics has obligingly written about it: http://www.scottaaronson.com/blog/?p=1103 (via cyphergoth).
It definitely wades into physics that I can't follow. It wades into physics that he doesn't know, for that matter.
However, as best I could tell, a few highlights. I found the best quote in his first comment:
I understand the case for MWI and agree with part of it: yes, MWI really is just the “obvious” story you would tell if you wanted to apply quantum mechanics to the entire universe. (The zillions of other worlds aren’t “added” per se; rather, they seem unavoidable once you accept that the Schrödinger equation applies always and everywhere.) Furthermore, all of the concrete alternatives to MWI on the market today are contrived and unsatisfactory in various ways.I think the first part of this is what MWI-advocates like Yudkowsky have been pushing, that the simplest interpretation of what we know about how wave functions work and how observations of them work, is that wavefunctions continue to evolve according to the usual equations all the time without exception, and what we call an "observation" is the observer becoming entangled with the experiment.
On the other hand, for the reasons explained in the post, I reject the further step some people take these days: of asserting that MWI is as obviously true as the Copernican system, and that anyone who refuses to see that is an idiot.
I feel the part of the MWI argument that I found important to have been settled with that comment: I know many physicists will still resist, but purely in terms of "does this sound true to a professional physicist", I trust Scott's interpretation.
Of course, that still leaves the meat of Scott's post, where he disagrees with MWI advocacy, but I didn't find that as important. He thinks MWI advocates are jumping the gun in saying it's obvious and settled. I mention a couple of his major points below, but I don't think I can do them justice (I'm not sure I understand or agree) so you're better off reading his post directly.
Turtles all the way up
The meat of the matter seems to be, photons experience decoherence. As do electrons. As do buckyballs. But what about cats? Human brains? Is there a size bigger than which things don't experience decoherence?
There seem to be three possibilities:
1. Yes, it works all the way up, just like it looks like it should
2. No, there's a physical effect which stops at a particular size. Scott makes the point that we've no idea how to apply superpositions to physics including non-trivial gravity fields, which he thinks may be relevant. I think this is his objection, though it doesn't make sense to me?
3. There's some other explanation, such as hidden variables, or magic powers of consciousness which make waveforms collapse, which have been expressed by many very venerable theories of physics and many of the best physicists from the 1930s onwards, but are pretty clearly Not True, unless we're prepared to accept dubious propositions like "consciousness has magic properties not explained by the rest of physics" or "spooky information travels faster than the speed of light in untestable ways". This is the sort of thing which looked more reasonable than not 50 years ago, but lots of mathematical proofs have proved lots of things are impossible (given "reasonable" assumptions), and MWI advocates have spent a lot of time refuting.
Scott agrees with MWI advocates that by now it pretty much has to be 1 or 2. He thinks there's still a reasonable possibility of 2, given experiments we'd like to do but haven't done yet, but I don't understand why.
Occams Razor
Scott obligingly demolishes a common flawed application of occam's razor, namely applying it in an extremely literal sense to minimise the number of physical objects, rather than the complexity of physical laws.
As he points out, tables are made of atoms. Are they made of atoms all the time, or only when you look at them closely enough? The first theory has fewer physical objects (since a lot of the time you just have "table", not a lot of separate atoms), but the second has much simpler physical laws (since you don't have to define what "someone's looking at it" means).
But most people would agree that it only makes sense to assume the table is made of atoms all the time. Ie. we should minimise the complexity of physical laws, not the number of physical objects. This is the interpretation of Occam's Razor which (a) is what scientists usually mean by it and (b) has worked historically -- for instance, the discovery that stars were stars was unlikely in the "number of objects" way, but fine in the "simple laws of physics" way, and was right.
Physicists who agree with the concepts of MWI but disagree with MWI
He points out that some physicists basically agree with the "there's a wavefunction, end of story" interpretation, but think "many worlds" is the wrong concept for it. I think that's fine, they're probably right. I am using MWI as a handy label, I'm happy to use something else. I don't endorse all the trappings of MWI as the best way of explaining stuff (they may be, or may not be), I intend to refer to the sort of things these people would agree with.
For the record, the idea is that there's a waveform, and "worlds" is a convenient way of viewing that waveform as overlaid copies of mostly-separate waveforms that only interact with themselves, not with each other.
So the question "when, exactly, does the universe split into two 'worlds'" has the answer "at some fuzzy point during the process when the split becomes irrecoverable, ie. the worlds no longer interfere". But the fuzziness is in the concept that things should be divided into worlds, not in the underlying physics. MWI advocates blame this fuzziness on MWI detractors and vice-versa, but it's primarily terminological.
Philosophical Quibbles
Scott raises an objection about how we perceive consciousness if our brains may be nearly-duplicated in adjacent "worlds". To me this sounds like "I don't like the conclusion so I don't want the argument to be true" wishful thinking. But Scott explicitly agrees that that sort of wishful thinking is pointless. So it seems like he's saying something else, but I'm not sure what.
Edit: Fix spelling mistake in name.