finds out that some toy system is Turing-complete (previously cellular automata, now graphs);
says “therefore, this toy system could simulate a universe”;
gives a few extremely shallow analogies between some properties of the system and some known laws of physics (on the level of: “this thing grows, which is just like the expansion of the universe… this thing is moving at constant speed, which is just like the speed of light… and this thing changes unpredictably, which is just like quantum uncertainty”);
declares that he has thereby solved the deepest mysteries of physics, and expects applause.
The problem with his explanations is that it is the Turing-completeness itself that has the “can simulate universe” trait, and the choice of a specific toy model is completely irrelevant; the very point of Turing-completeness is that in some sense all toy models are equivalent to each other. And the Turing-completeness was known long before Wolfram, just like the hypothesis that the universe might be a computation. His only contribution is the only part that does not matter: finding yet another Turing-complete system. Dozens of them are already known.
The shallow analogies between the toy model and some known laws of physics are not only silly, but also irrelevant. The simulated universe does not have the same traits as the simulating machine—a gray rectangular computer can simulate a yellow sphere—and Wolfram talks about how the gray color is kinda similar to yellow. But precisely that should be irrelevant, if the universe is a computation.
I’m not asking because you not having it read it would ‘disqualify’ your comment or anything. I’m just curious.
I can see some substance in what you wrote in this comment (and I recognize you as a generally very thoughtful commenter), but this still seems exaggerated and at least a little uncharitable.
Something that he did in NKS, and then further confirmed in his recent post about him writing the book, was just look at a bunch of examples. I thought that was an admirable instance of ‘original seeing’, if nothing else.
(And maybe even Wolfram himself has trouble keeping up with all of the relevant academic literature.)
I can understand that a lot – or maybe even literally all – of that work had been previously been done. The notes in the book do cite an awful lot of people tho! But I’m still not sure that anyone else had previously compiled it into a single work, let alone one that was accessible to people without an account thru, e.g. a university, to access academic journals. That seems, to me, to be valuable, in and of itself.
It does seem like Wolfram may have (nearly) permanently and irrevocably burned all of the bridges between himself and academia, and that’s very sad. And others have pointed out that he’s, if not just an asshole, possibly clinically disordered.
But enough people have also expressed, in various forms, that basically everything Wolfram has published, in any form, is completely bullshit and can be reasonably ignored entirely. That’s certainly making me think I shouldn’t bother sharing anything connected to him here, but I’m still not entirely convinced that something like this video, that another commenter on this post shared, is just entirely a mistake and stupid (and worthy of vitriol):
I have read some articles by Wolfram, but not the books.
The video you linked, if I understand it correctly, says that (unlike with the rule 110), the latest Wolfram’s theory is not about simulating the universe, but rather that the structure of the space-time at/below the Planck length is actually a hypergraph evolving using some simple rewriting rules.
The idea that the universe may be discrete at small levels (and that the smoothness we observe is just applying the laws of statistics to zillions of Planck units) was already popular; the challenge is to propose the specific rules for the discrete levels so that their aggregates are compatible with the known laws of physics.
The guy in the video made three statements that sounded interesting:
that if you take those kinds of hypergraphs and rewriting rules where the order of applying the rules does not matter, and ask “what happens when the size of the graph goes to infinity”, you can derive the equation for Einstein–Hilbert action;
that if you have multiple possible rules… I forgot the details… it results in something like the many world of quantum physics;
that if Wolfram’s model is true, we should expect to find situations where the universe is not 3-dimensional, but has a slightly different Hausdorff dimension (or, as the YouTube subtitle generator calls it, “house dwarf dimension”), like 3.001 or 2.999.
My reaction:
The second statement sounds unimpressive (many possible alternative rules should logically result in many possible outcomes), but that could be because I misunderstood something.
The first statement is more impressive; deriving a specific equation sounds cool. So it seems to me that the best way to convince someone to take Wolfram’s theory seriously is to show them the derivation. (It wasn’t show in the video, it was just said that it exists.)
I am not sure about the third one. Seems to me that hypegraphs should allow an arbitrary number of dimensions. So on one hand, finding situations with different number of dimensions would be evidence in favor of the theory. On the other hand, how does the theory explain that usually the number is 3?
Summary:
Show me (not me specifically, but rather any physicist you want to convince to take Wolfram’s model seriously) how you can derive the equation for Einstein-Hilbert action from the assumption that you have a hypergraph and rules that are independent on order of their use. (Actually derive the equation, not just claim that it can be done.) That should impress people, I think.
I think of it as more that – in what they’re calling “multiway systems” – there are (often) several rules for how the hypergraph evolves and you can apply those rules in different orders. Instead of either picking a set of rules that is ‘invariant’ with respect to the order in which they’re applied, or trying to pick a clever algorithm for deciding that order, they consider the full ‘multiway hypergraph’ of all possible ‘application orderings’.
There’s some other impressive derivations they’ve done, supposedly, tho several are matching other prospective ‘quantum gravity’ theories (and several of those are really more like potential-quantum-gravity mathematical formalisms instead of strictly physics theories).
Seems to me that hypegraphs should allow an arbitrary number of dimensions.
You’re right! They seem to be both ‘exploring’ all/LOTs of individual hypergraph systems, as well as looking for one that might match our observations of our own universe.
Jonathan Gorard claimed I think in one of the videos (shared by someone else in another comment on this post) that basically all of their work (e.g. academic papers, working papers, and code) is available on their website:
You might like the notes I wrote up in another comment on this post about one of a few ideas someone else shared. The video is a podcast ‘interviewing’ the ‘math lead’ of the project, Gorard. The two videos with Gorard on the same podcast:
I found the second one to be particularly impressive just based on the details of the theory/model that Gorard shared. The first is still great too; just more of an introduction of the ideas.
You’re right! They seem to be both ‘exploring’ all/LOTs of individual hypergraph systems, as well as looking for one that might match our observations of our own universe.
From my perspective, this is a bad thing. If you can explain everything, you explain nothing. Science is about making predictions, not excuses. Not trying to be dogmatic here, I just don’t see how a theory that says “the universe could have any number of dimensions” is going to help you build something useful, such as a microwave oven.
If you had the hypergraph theory plus the extra conditions that make the predictions fit our universe, that would be useful. Because it could be used to actually predict things in our universe. I am not even asking for new predictions here, just for the theory to say something other than the very filters that were used to select the right kind of rules, and for that something to apply to our universe.
If you are looking for the rules that prove X, Y, Z, and after finding them exclaim “behold, my theory predicts X, Y, Z”, all that it proves is that X, Y, Z are not fundamentally incompatible with your model (maybe because your model is compatible with almost everything).
It would be valuable if you looked for rules that prove X and Y, and found out that all such rules also predict Z. Then you could say “my theory explains why if X and Y are true in our universe, then Z must be too”. That sounds interesting, if Z does not seem like an obvious consequence of X and Y.
The fictional Library of Babel contains the textbook with the Theory of Everything. Knowing this is not useful at all.
If you had the hypergraph theory plus the extra conditions that make the predictions fit our universe, that would be useful.
That’s what I meant by the part of my previous comment that you quoted.
But, aside from the physics, they’re also exploring the more general mathematics/computation of ‘multiway systems’ too. They’re the pioneers of the math and I think they’re effectively ‘nerd sniping’ themselves with interesting math/computer-science and that’s become something like a ‘parallel’ effort from the focus on deriving/finding a GUT that unifies (our specific) QM and GR.
They’ve found specific “extra conditions” that identify specific ‘hypergraphs’ (really the multiway system that generates a ‘multiway’ set/tree/whatever of hypergraph evolution) and those specific hypergraphs match up to, e.g. General Relativity (in at least some ways).
In the second podcast video with Gorard, he describes ‘experiments’ they’ve done with things they think might be equivalent to the particles in the Standard Model but it seems like they weren’t trying to (at that point), ‘derive’ them from ‘scratch’. That is apparently something they want to do, but it seems (very) plausible that reaching ‘particles’ from the Big Bang, and simulating that on a computer, might be its own future mega-project.
It seems to me like Wolfram’s project isn’t doing anything that all of the other GUTs or quantum-gravity theories/formalisms aren’t doing too. They’re all trying to find a more fundamental theorem from which the Standard Model and General Relativity can be derived, i.e. ‘emerge’ automatically.
AFAIK, ‘string theory’, in the fully general sense of all similar theories, doesn’t itself constrain the number of dimensions of the universe. Why is Wolfram’s theory/model different?
If you are looking for the rules that prove X, Y, Z, and after finding them exclaim “behold, my theory predicts X, Y, Z”, all that it proves is that X, Y, Z are not fundamentally incompatible with your model (maybe because your model is compatible with almost everything).
It seems like even if the only thing they end up doing is finding a ‘compressed’ version of QM+GR that that would in fact be a monumental achievement? I don’t know why that wouldn’t also predictably imply other new predictions.
AFAIK, ‘string theory’, in the fully general sense of all similar theories, doesn’t itself constrain the number of dimensions of the universe. Why is Wolfram’s theory/model different?
I think I have heard people making similar objections against the string theory. I don’t know much about string theory to evaluate whether the objections are deserved. If they are, I would treat it as an argument against string theory, not as an argument if favor of Wolfram’s theory.
I don’t know why that wouldn’t also predictably imply other new predictions.
We are still talking about the potential of the new theory (possibly also string theory).
The original complaint is that the physics community does not take Wolfram’s ideas seriously. Some people seem to do, and they seem to promise amazing results soon. From my perspective, there are two options: if those amazing results are actually delivered and published (the actual proofs, not just a youtube video claiming that they could be done), the physics community will start paying attention. But if the amazing results never materialize (despite smart people working on the problem), then maybe it was all hype and no substance.
If there is tomorrow an equation proving theory of relativity from the first principles, and no one in academia pays attention to it, that would be the right moment to discuss the proof at Less Wrong.
I’ve now watched several videos, particularly a few with the ‘math lead’ of the Wolfram’s physics project, and I now do not think that the “physics community does not take Wolfram’s ideas seriously”. It seems like they do in fact have a good set of contributors and are making great progress (relative to expectations).
You’re absolutely right that it could be “all hype and no substance”, tho I don’t think that’s even a remote possibility beyond just physics. (The ‘math’ alone seems interesting, which is ‘more than enough’ IMO for any possible ‘math’ that someone might want to study/explore.)
A paper they’ve published (and one that’s available on the project’s website) about (some aspects of) relativity:
In my opinion, Wolfram essentially does this:
finds out that some toy system is Turing-complete (previously cellular automata, now graphs);
says “therefore, this toy system could simulate a universe”;
gives a few extremely shallow analogies between some properties of the system and some known laws of physics (on the level of: “this thing grows, which is just like the expansion of the universe… this thing is moving at constant speed, which is just like the speed of light… and this thing changes unpredictably, which is just like quantum uncertainty”);
declares that he has thereby solved the deepest mysteries of physics, and expects applause.
The problem with his explanations is that it is the Turing-completeness itself that has the “can simulate universe” trait, and the choice of a specific toy model is completely irrelevant; the very point of Turing-completeness is that in some sense all toy models are equivalent to each other. And the Turing-completeness was known long before Wolfram, just like the hypothesis that the universe might be a computation. His only contribution is the only part that does not matter: finding yet another Turing-complete system. Dozens of them are already known.
The shallow analogies between the toy model and some known laws of physics are not only silly, but also irrelevant. The simulated universe does not have the same traits as the simulating machine—a gray rectangular computer can simulate a yellow sphere—and Wolfram talks about how the gray color is kinda similar to yellow. But precisely that should be irrelevant, if the universe is a computation.
Have you read NKS?
I’m not asking because you not having it read it would ‘disqualify’ your comment or anything. I’m just curious.
I can see some substance in what you wrote in this comment (and I recognize you as a generally very thoughtful commenter), but this still seems exaggerated and at least a little uncharitable.
Something that he did in NKS, and then further confirmed in his recent post about him writing the book, was just look at a bunch of examples. I thought that was an admirable instance of ‘original seeing’, if nothing else.
(And maybe even Wolfram himself has trouble keeping up with all of the relevant academic literature.)
I can understand that a lot – or maybe even literally all – of that work had been previously been done. The notes in the book do cite an awful lot of people tho! But I’m still not sure that anyone else had previously compiled it into a single work, let alone one that was accessible to people without an account thru, e.g. a university, to access academic journals. That seems, to me, to be valuable, in and of itself.
It does seem like Wolfram may have (nearly) permanently and irrevocably burned all of the bridges between himself and academia, and that’s very sad. And others have pointed out that he’s, if not just an asshole, possibly clinically disordered.
But enough people have also expressed, in various forms, that basically everything Wolfram has published, in any form, is completely bullshit and can be reasonably ignored entirely. That’s certainly making me think I shouldn’t bother sharing anything connected to him here, but I’m still not entirely convinced that something like this video, that another commenter on this post shared, is just entirely a mistake and stupid (and worthy of vitriol):
Eigenbros ep 117 - Jonathan Gorard (Wolfram Physics) - YouTube
I have read some articles by Wolfram, but not the books.
The video you linked, if I understand it correctly, says that (unlike with the rule 110), the latest Wolfram’s theory is not about simulating the universe, but rather that the structure of the space-time at/below the Planck length is actually a hypergraph evolving using some simple rewriting rules.
The idea that the universe may be discrete at small levels (and that the smoothness we observe is just applying the laws of statistics to zillions of Planck units) was already popular; the challenge is to propose the specific rules for the discrete levels so that their aggregates are compatible with the known laws of physics.
The guy in the video made three statements that sounded interesting:
that if you take those kinds of hypergraphs and rewriting rules where the order of applying the rules does not matter, and ask “what happens when the size of the graph goes to infinity”, you can derive the equation for Einstein–Hilbert action;
that if you have multiple possible rules… I forgot the details… it results in something like the many world of quantum physics;
that if Wolfram’s model is true, we should expect to find situations where the universe is not 3-dimensional, but has a slightly different Hausdorff dimension (or, as the YouTube subtitle generator calls it, “house dwarf dimension”), like 3.001 or 2.999.
My reaction:
The second statement sounds unimpressive (many possible alternative rules should logically result in many possible outcomes), but that could be because I misunderstood something.
The first statement is more impressive; deriving a specific equation sounds cool. So it seems to me that the best way to convince someone to take Wolfram’s theory seriously is to show them the derivation. (It wasn’t show in the video, it was just said that it exists.)
I am not sure about the third one. Seems to me that hypegraphs should allow an arbitrary number of dimensions. So on one hand, finding situations with different number of dimensions would be evidence in favor of the theory. On the other hand, how does the theory explain that usually the number is 3?
Summary:
Show me (not me specifically, but rather any physicist you want to convince to take Wolfram’s model seriously) how you can derive the equation for Einstein-Hilbert action from the assumption that you have a hypergraph and rules that are independent on order of their use. (Actually derive the equation, not just claim that it can be done.) That should impress people, I think.
I think of it as more that – in what they’re calling “multiway systems” – there are (often) several rules for how the hypergraph evolves and you can apply those rules in different orders. Instead of either picking a set of rules that is ‘invariant’ with respect to the order in which they’re applied, or trying to pick a clever algorithm for deciding that order, they consider the full ‘multiway hypergraph’ of all possible ‘application orderings’.
There’s some other impressive derivations they’ve done, supposedly, tho several are matching other prospective ‘quantum gravity’ theories (and several of those are really more like potential-quantum-gravity mathematical formalisms instead of strictly physics theories).
You’re right! They seem to be both ‘exploring’ all/LOTs of individual hypergraph systems, as well as looking for one that might match our observations of our own universe.
Jonathan Gorard claimed I think in one of the videos (shared by someone else in another comment on this post) that basically all of their work (e.g. academic papers, working papers, and code) is available on their website:
The Wolfram Physics Project: Finding the Fundamental Theory of Physics
Some papers from the “Technical Documents” page:
Some Relativistic and Gravitational Properties of the Wolfram Model [PDF]
Wolfram Physics Project | A Class of Models with the Potential to Represent Fundamental Physics
Some Quantum Mechanical Properties of the Wolfram Model [PDF]
You might like the notes I wrote up in another comment on this post about one of a few ideas someone else shared. The video is a podcast ‘interviewing’ the ‘math lead’ of the project, Gorard. The two videos with Gorard on the same podcast:
Eigenbros ep 117 - Jonathan Gorard (Wolfram Physics) - YouTube
Eigenbros ep 138 - Wolfram Physics Project Pt. 2 (w/ Jonathan Gorard) - YouTube
I found the second one to be particularly impressive just based on the details of the theory/model that Gorard shared. The first is still great too; just more of an introduction of the ideas.
From my perspective, this is a bad thing. If you can explain everything, you explain nothing. Science is about making predictions, not excuses. Not trying to be dogmatic here, I just don’t see how a theory that says “the universe could have any number of dimensions” is going to help you build something useful, such as a microwave oven.
If you had the hypergraph theory plus the extra conditions that make the predictions fit our universe, that would be useful. Because it could be used to actually predict things in our universe. I am not even asking for new predictions here, just for the theory to say something other than the very filters that were used to select the right kind of rules, and for that something to apply to our universe.
If you are looking for the rules that prove X, Y, Z, and after finding them exclaim “behold, my theory predicts X, Y, Z”, all that it proves is that X, Y, Z are not fundamentally incompatible with your model (maybe because your model is compatible with almost everything).
It would be valuable if you looked for rules that prove X and Y, and found out that all such rules also predict Z. Then you could say “my theory explains why if X and Y are true in our universe, then Z must be too”. That sounds interesting, if Z does not seem like an obvious consequence of X and Y.
The fictional Library of Babel contains the textbook with the Theory of Everything. Knowing this is not useful at all.
That’s what I meant by the part of my previous comment that you quoted.
But, aside from the physics, they’re also exploring the more general mathematics/computation of ‘multiway systems’ too. They’re the pioneers of the math and I think they’re effectively ‘nerd sniping’ themselves with interesting math/computer-science and that’s become something like a ‘parallel’ effort from the focus on deriving/finding a GUT that unifies (our specific) QM and GR.
They’ve found specific “extra conditions” that identify specific ‘hypergraphs’ (really the multiway system that generates a ‘multiway’ set/tree/whatever of hypergraph evolution) and those specific hypergraphs match up to, e.g. General Relativity (in at least some ways).
In the second podcast video with Gorard, he describes ‘experiments’ they’ve done with things they think might be equivalent to the particles in the Standard Model but it seems like they weren’t trying to (at that point), ‘derive’ them from ‘scratch’. That is apparently something they want to do, but it seems (very) plausible that reaching ‘particles’ from the Big Bang, and simulating that on a computer, might be its own future mega-project.
It seems to me like Wolfram’s project isn’t doing anything that all of the other GUTs or quantum-gravity theories/formalisms aren’t doing too. They’re all trying to find a more fundamental theorem from which the Standard Model and General Relativity can be derived, i.e. ‘emerge’ automatically.
AFAIK, ‘string theory’, in the fully general sense of all similar theories, doesn’t itself constrain the number of dimensions of the universe. Why is Wolfram’s theory/model different?
It seems like even if the only thing they end up doing is finding a ‘compressed’ version of QM+GR that that would in fact be a monumental achievement? I don’t know why that wouldn’t also predictably imply other new predictions.
I think I have heard people making similar objections against the string theory. I don’t know much about string theory to evaluate whether the objections are deserved. If they are, I would treat it as an argument against string theory, not as an argument if favor of Wolfram’s theory.
We are still talking about the potential of the new theory (possibly also string theory).
The original complaint is that the physics community does not take Wolfram’s ideas seriously. Some people seem to do, and they seem to promise amazing results soon. From my perspective, there are two options: if those amazing results are actually delivered and published (the actual proofs, not just a youtube video claiming that they could be done), the physics community will start paying attention. But if the amazing results never materialize (despite smart people working on the problem), then maybe it was all hype and no substance.
If there is tomorrow an equation proving theory of relativity from the first principles, and no one in academia pays attention to it, that would be the right moment to discuss the proof at Less Wrong.
I’ve now watched several videos, particularly a few with the ‘math lead’ of the Wolfram’s physics project, and I now do not think that the “physics community does not take Wolfram’s ideas seriously”. It seems like they do in fact have a good set of contributors and are making great progress (relative to expectations).
You’re absolutely right that it could be “all hype and no substance”, tho I don’t think that’s even a remote possibility beyond just physics. (The ‘math’ alone seems interesting, which is ‘more than enough’ IMO for any possible ‘math’ that someone might want to study/explore.)
A paper they’ve published (and one that’s available on the project’s website) about (some aspects of) relativity:
Some Relativistic and Gravitational Properties of the Wolfram Model [PDF]
(I haven’t read it myself yet.)