5 ms·

There is a logic jump in the concept of the ruliad. First it goes: > the entangled limit of everything that is computationally possible: the result of followi

There is a logic jump in the concept of the ruliad. First it goes: > the entangled limit of everything that is computationally possible: the result of following all possible computational rules in all possible ways And then > it encapsulates not only all formal possibilities but also everything about our physical universe The assumption that everything about our physical universe is described by computational rules is not a given, and is never explored nor justified. Even the assumption that our perception follows computational rules is not really explored, just given. There is this old idea that with an universal computer that knew everything, one could predict anything. But that has been challenged by uncertainty principles, observer effect and all that. So it seems this castle has sand foundations.

Legend2440 3y ago
This is a well-known conjecture called the Physical Church-Turing Thesis. It is just a conjecture, but emperically it has held true to the limits of our knowledge of physics. It also still works even if the universe is probabilistic.
- p-e-w 3y ago
  > but emperically it has held true to the limits of our knowledge of physics Considering that our entire knowledge of physics is encoded as computational rules, because that's all we really know how to do, this statement is basically a tautology.
  - Legend2440 3y ago
    We do physics this way exactly because it works so well - it is "unreasonably effective", as the saying goes.
    - p-e-w 3y ago
      No, we do physics this way because we don't know any other way. When this way fails (e.g. quantum gravity, turbulence), we assume that the problem is that we haven't tried hard enough yet. Which may be true, but it's also possible that the computational approach simply doesn't work for those phenomena.
      - andybak 3y ago
        I'm not clear what you mean about turbulence? Quantum gravity involves as-yet unknown physics but turbulence is surely just a case of "the number of calculations we need to simulate it grows quicker than we can reasonably keep up with"? i.e. simple toy simulations work fine but we don't have a planet-sized computer to do something more extensive. (Just to clarify - this isn't a "classical vs quantum" thing - it's a "we know the equations vs we don't" thing. I'm sure QED simulations are fairly similar within a known domain)
        Sharlin 3y ago
        Presumably the GP refers to turbulence being infamously difficult to model analytically. But we have a great model for turbulent flow, the Navier–Stokes differential equations, it’s just that we can only solve them numerically in all but the simplest of cases. But when we do we get good results, so it’s not like turbulence is some mystical phenomenon beyond the reach of our standard mathematical tools!
        andybak 3y ago
        My knowledge is a bit weak but aren't there non-analytic ways to model turbulence? "Analytical" isn't the whole of maths. (I might not have a clear understanding of what "analytic" means so please correct me if I'm talking hogwash)
        Sharlin 3y ago
        The alternative is numerical approximation, as I mentioned. Like always in physics, any analytic (ie. "closed form") solutions would also necessarily have to be special-cases, approximations, and simplifications. The problem with turbulence is that it's so chaotic and complex that it doesn't seem to be reducible to simpler models while still retaining some predictive power. But that's not very surprising; we're talking about the chaotic motion of molecules at the Avogadro scale! After all, we can't even write closed-form solutions (or even approximations) to the motion of merely three bodies under gravity, never mind ten to the power of twenty-three. But what is fascinating about turbulence, though, is the "edge of chaos" – the boundary conditions at which laminar (non-turbulent) flow suddenly turns turbulent. On one side of the boundary analytic treatment is possible, on the other it is not.
- colordrops 3y ago
  But physics isn't a full description of reality. It certainly hasn't conquered consciousness and qualia and subjective experience.
  - kaibee 3y ago
    > It certainly hasn't conquered consciousness and qualia and subjective experience. I'm not sure how you're so certain of this. It is entirely possible that all we need to do to create a conscious mind is run a full molecular dynamics simulation of a detailed enough human brain scan. We know the equations to do this, we just don't have nearly enough computational power.
    - jchanimal 3y ago
      That would only be possible if qualia has no causation. In the case where our experiences have real effects they cannot be fully described by a Turing architecture.
    - prometheus76 3y ago
      Already you've sneaked in the presupposition that consciousness is only a result of your brain activity. Have you ever not eaten for two days? Or not slept for more than 24 hours? Have you ever had an itch under a cast that you couldn't itch, and been constantly distracted by it? You think that doesn't affect your consciousness?
      - kaibee 3y ago
        > Already you've sneaked in the presupposition that consciousness is only a result of your brain activity. What else could it be? The interesting question is how and why. > Have you ever not eaten for two days? Or not slept for more than 24 hours? Have you ever had an itch under a cast that you couldn't itch, and been constantly distracted by it? Sure, I've done all of those things, but they all cause physical changes in the brain. These physical changes then directly affect your qualia-type experience of reality.
ttctciyf 3y ago
> assumption that everything about our physical universe is described by computational rules If you understand "physical" to mean "stuff that physics studies" then the assumption is not so surprising - physics itself is the discovery and application of models to correctly predict or retrodict (i.e. compute) the simple measurable behavior of matter, so it is intrinsically unable to study anything unamenable to computation. In fact Wolfram is engaged in an attempt (as he sees it) to enlarge the computational paradigm of physics from models consisting largely of differential equations to a "multicomputational" paradigm (see [1]). I think your statement is much more defensible if you drop the word "physical": The assumption that everything about our universe is described by computational rules is not a given. A classic example might be our inability to sufficiently specify "qualia". If our experiences were fully describable by computational rules I suspect we would by now have discovered how to specify them so that we could agree (or disagree) that 'the red I see' looks like 'the red you see', but our inability to do this is notorious. Nonetheless my (and your) experience of red is part of the universe. 1: https://writings.stephenwolfram.com/2021/09/multicomputation-a-fourth-paradigm-for-theoretical-science/ https://writings.stephenwolfram.com/2021/09/multicomputation...
- bangkoksbest 3y ago
  There are physical models are not computable (in the sense of producing number values in a finite algorithm), particularly in quantum physics, e.g. the wave function.
  - ttctciyf 3y ago
    As a physical model, as opposed to a purely mathematical entity, don't you need to add operators to the wave function in order to have applicability to observable phenomena? I'm no physicist, but applying operators to the wave function seems like a computation to me?
    - bangkoksbest 3y ago
      I mean, theoretical physics models are still physics models. Functional analysis is still a valid mathematical tool for physics modeling even if it doesn't (usually) produce finitely calculable solutions. Physicists assign physical interpretation to certain parameters, even if they are not (yet) observable. I suppose there's probably a higher mathematical representation going on here that is finitely computable though. Something like a symbolic representation of the solution functions for the wave function are calculable even if the parameter values are not.
wnkrshm 3y ago
Even in a perfectly deterministic universe, the accessibility of that determinism to models hinges on knowledge of the state of the universe. We're surrounded by highly nonlinear interactions that we just approximate as linear - if we want to truly predict anything, in general, we'd need infinite precision on the variables that describe the state of these systems, because an infinitely small perturbation in the initial conditions can cause an almost (limited by conservation laws) arbitrary deviation in the predicted state.
jprete 3y ago
He also snuck in the claim that alignment means “AI that thinks like a human” which is not at all what people mean by the word.
Xcelerate 3y ago
> But that has been challenged by uncertainty principles, observer effect and all that. Any system which allows universal computation necessarily has subsystems that cannot make perfect predictions about the full system. This doesn’t even require quantum mechanics; humans would run up against this problem in a fully classical universe. Loosely put, since we’re part of the system ourselves, we can’t gather all of the information necessary for perfect prediction without interfering with whatever we’re trying to predict. However, as far as I’m aware, we haven’t yet closed all the loopholes related to superdeterminism, meaning quantum randomness might not be truly random in the sense that an observer completely outside of our universe could in theory procure an algorithm that predicts the exact eigenvalues for every quantum measurement we make. Since we’re inside the system, no luck for us though. But I like to make a subtle distinction between “unpredictable random” (max Kolmogorov complexity) and “truly random” (ungenerated via any Turing computable algorithm and unpredictable with infinite knowledge). The latter is a bit philosophically difficult to prove exists. The former is quite easy to prove exists; think about how a one time pad has max Kolmogorov complexity (assuming a good source of random noise) but it’s conditional Kolmogorov complexity with the key is much smaller.
smaddox 3y ago
> The assumption that everything about our physical universe is described by computational rules is not a given, and is never explored nor justified. That's in the linked article about the ruliad, among other places in his writings about their physics project: https://writings.stephenwolfram.com/2021/11/the-concept-of-the-ruliad/ https://writings.stephenwolfram.com/2021/11/the-concept-of-t...

Legend2440 3y ago

This is a well-known conjecture called the Physical Church-Turing Thesis. It is just a conjecture, but emperically it has held true to the limits of our knowledge of physics. It also still works even if the universe is probabilistic.

p-e-w 3y ago
> but emperically it has held true to the limits of our knowledge of physics Considering that our entire knowledge of physics is encoded as computational rules, because that's all we really know how to do, this statement is basically a tautology.
- Legend2440 3y ago
  We do physics this way exactly because it works so well - it is "unreasonably effective", as the saying goes.
  - p-e-w 3y ago
    No, we do physics this way because we don't know any other way. When this way fails (e.g. quantum gravity, turbulence), we assume that the problem is that we haven't tried hard enough yet. Which may be true, but it's also possible that the computational approach simply doesn't work for those phenomena.
    - andybak 3y ago
      I'm not clear what you mean about turbulence? Quantum gravity involves as-yet unknown physics but turbulence is surely just a case of "the number of calculations we need to simulate it grows quicker than we can reasonably keep up with"? i.e. simple toy simulations work fine but we don't have a planet-sized computer to do something more extensive. (Just to clarify - this isn't a "classical vs quantum" thing - it's a "we know the equations vs we don't" thing. I'm sure QED simulations are fairly similar within a known domain)
      - Sharlin 3y ago
        Presumably the GP refers to turbulence being infamously difficult to model analytically. But we have a great model for turbulent flow, the Navier–Stokes differential equations, it’s just that we can only solve them numerically in all but the simplest of cases. But when we do we get good results, so it’s not like turbulence is some mystical phenomenon beyond the reach of our standard mathematical tools!
        andybak 3y ago
        My knowledge is a bit weak but aren't there non-analytic ways to model turbulence? "Analytical" isn't the whole of maths. (I might not have a clear understanding of what "analytic" means so please correct me if I'm talking hogwash)
        Sharlin 3y ago
        The alternative is numerical approximation, as I mentioned. Like always in physics, any analytic (ie. "closed form") solutions would also necessarily have to be special-cases, approximations, and simplifications. The problem with turbulence is that it's so chaotic and complex that it doesn't seem to be reducible to simpler models while still retaining some predictive power. But that's not very surprising; we're talking about the chaotic motion of molecules at the Avogadro scale! After all, we can't even write closed-form solutions (or even approximations) to the motion of merely three bodies under gravity, never mind ten to the power of twenty-three. But what is fascinating about turbulence, though, is the "edge of chaos" – the boundary conditions at which laminar (non-turbulent) flow suddenly turns turbulent. On one side of the boundary analytic treatment is possible, on the other it is not.
colordrops 3y ago
But physics isn't a full description of reality. It certainly hasn't conquered consciousness and qualia and subjective experience.
- kaibee 3y ago
  > It certainly hasn't conquered consciousness and qualia and subjective experience. I'm not sure how you're so certain of this. It is entirely possible that all we need to do to create a conscious mind is run a full molecular dynamics simulation of a detailed enough human brain scan. We know the equations to do this, we just don't have nearly enough computational power.
  - jchanimal 3y ago
    That would only be possible if qualia has no causation. In the case where our experiences have real effects they cannot be fully described by a Turing architecture.
  - prometheus76 3y ago
    Already you've sneaked in the presupposition that consciousness is only a result of your brain activity. Have you ever not eaten for two days? Or not slept for more than 24 hours? Have you ever had an itch under a cast that you couldn't itch, and been constantly distracted by it? You think that doesn't affect your consciousness?
    - kaibee 3y ago
      > Already you've sneaked in the presupposition that consciousness is only a result of your brain activity. What else could it be? The interesting question is how and why. > Have you ever not eaten for two days? Or not slept for more than 24 hours? Have you ever had an itch under a cast that you couldn't itch, and been constantly distracted by it? Sure, I've done all of those things, but they all cause physical changes in the brain. These physical changes then directly affect your qualia-type experience of reality.

ttctciyf 3y ago

> assumption that everything about our physical universe is described by computational rules If you understand "physical" to mean "stuff that physics studies" then the assumption is not so surprising - physics itself is the discovery and application of models to correctly predict or retrodict (i.e. compute) the simple measurable behavior of matter, so it is intrinsically unable to study anything unamenable to computation. In fact Wolfram is engaged in an attempt (as he sees it) to enlarge the computational paradigm of physics from models consisting largely of differential equations to a "multicomputational" paradigm (see [1]). I think your statement is much more defensible if you drop the word "physical": The assumption that everything about our universe is described by computational rules is not a given. A classic example might be our inability to sufficiently specify "qualia". If our experiences were fully describable by computational rules I suspect we would by now have discovered how to specify them so that we could agree (or disagree) that 'the red I see' looks like 'the red you see', but our inability to do this is notorious. Nonetheless my (and your) experience of red is part of the universe. 1: https://writings.stephenwolfram.com/2021/09/multicomputation-a-fourth-paradigm-for-theoretical-science/ https://writings.stephenwolfram.com/2021/09/multicomputation...

bangkoksbest 3y ago
There are physical models are not computable (in the sense of producing number values in a finite algorithm), particularly in quantum physics, e.g. the wave function.
- ttctciyf 3y ago
  As a physical model, as opposed to a purely mathematical entity, don't you need to add operators to the wave function in order to have applicability to observable phenomena? I'm no physicist, but applying operators to the wave function seems like a computation to me?
  - bangkoksbest 3y ago
    I mean, theoretical physics models are still physics models. Functional analysis is still a valid mathematical tool for physics modeling even if it doesn't (usually) produce finitely calculable solutions. Physicists assign physical interpretation to certain parameters, even if they are not (yet) observable. I suppose there's probably a higher mathematical representation going on here that is finitely computable though. Something like a symbolic representation of the solution functions for the wave function are calculable even if the parameter values are not.

wnkrshm 3y ago

Even in a perfectly deterministic universe, the accessibility of that determinism to models hinges on knowledge of the state of the universe. We're surrounded by highly nonlinear interactions that we just approximate as linear - if we want to truly predict anything, in general, we'd need infinite precision on the variables that describe the state of these systems, because an infinitely small perturbation in the initial conditions can cause an almost (limited by conservation laws) arbitrary deviation in the predicted state.

jprete 3y ago

He also snuck in the claim that alignment means “AI that thinks like a human” which is not at all what people mean by the word.

Xcelerate 3y ago

> But that has been challenged by uncertainty principles, observer effect and all that. Any system which allows universal computation necessarily has subsystems that cannot make perfect predictions about the full system. This doesn’t even require quantum mechanics; humans would run up against this problem in a fully classical universe. Loosely put, since we’re part of the system ourselves, we can’t gather all of the information necessary for perfect prediction without interfering with whatever we’re trying to predict. However, as far as I’m aware, we haven’t yet closed all the loopholes related to superdeterminism, meaning quantum randomness might not be truly random in the sense that an observer completely outside of our universe could in theory procure an algorithm that predicts the exact eigenvalues for every quantum measurement we make. Since we’re inside the system, no luck for us though. But I like to make a subtle distinction between “unpredictable random” (max Kolmogorov complexity) and “truly random” (ungenerated via any Turing computable algorithm and unpredictable with infinite knowledge). The latter is a bit philosophically difficult to prove exists. The former is quite easy to prove exists; think about how a one time pad has max Kolmogorov complexity (assuming a good source of random noise) but it’s conditional Kolmogorov complexity with the key is much smaller.

smaddox 3y ago

> The assumption that everything about our physical universe is described by computational rules is not a given, and is never explored nor justified. That's in the linked article about the ruliad, among other places in his writings about their physics project: https://writings.stephenwolfram.com/2021/11/the-concept-of-the-ruliad/ https://writings.stephenwolfram.com/2021/11/the-concept-of-t...