5 ms·

I disagree. A lot of progress in physics was by providing another formalism, which predicted nothing, but provided another picture, provides a theoretical found

by stared 6y ago

I disagree. A lot of progress in physics was by providing another formalism, which predicted nothing, but provided another picture, provides a theoretical foundation, etc. Take the Langrangian formalism for classical mechanics (you write a book about Noether, right?) - it is equivalent to the causal laws of motions, yet provide a powerful practical and theoretical tool. Or: - Schroedinger vs Heisenberg description of QM - Path integral formulation - Feynman diagrams (here you can argue that it is not even a formulation, "just a visual representation") The problem with Stephen Wolfram's theory of everything is not that it does not explain new things. The problem is that (as it is right now) it does not explain new things (he should wish to get there). The problem is that it follows the pattern: 1. Simple graph automata rules. 2. A loose connection with a few physics properties. 3. ??? 4. Theory of everything! Profit!!!

leephillips 6y ago
I don't think we actually disagree; in any case, I can't find anything in your comment that I don't agree with. I think an important point is that the alternative formulations you list are provably equivalent to the theories that they reformulate; they can not make different predictions, but make make some calculations much easier and lead to expanded insights. It was a big deal when Schrödinger proved the equivalence of the two versions of QM (although his proof was incomplete; a rigorous proof was accomplished by von Neumann). As you suggest, the problem with Wolfram's computational approach is that since it is not equivalent to existing theory, except perhaps in some special cases, it must be a different theory, with different predictions. And, as of almost 20 years since _A New Kind of Science_, it hasn't predicted anything.
- stared 6y ago
  For a long time, I had thought that A New Kind of Science is an excellent math popularization book on cellular automata, prompting to do computational experiments in mathematics, and having an emphasis on beautiful visualizations. Also, his science education posts (e.g. on analyzing FB data) are wonderful. I just read a few chapters. And later, I was surprised by the criticism. It turned out that "A New Kind of Science" is not a tongue-in-cheek title (vide "Freakonomics"), but that SW actually believes that it not only is a new kind of science, but his science (Turning, Conway, Feynman, anyone...?). Some cellular automata simulations are nice. And in general, many concrete in physics are not general, yet are useful e.g. for numerics (with known limitations). They are not considered crackpot (even if not useful), as long as they make clear what they predict, what - not. Any toy model usurping its right to be the TOE, without a sufficient proof (not just the lack of proof it is not) is likely to be crackpotism. (Side note: for quite a few years I used heavily Mathematica. Only a decade ago I switched to Python, for a few reasons, but mostly for open-source (and quite a few of its implications). Jupyter Notebook, which IMHO was a game-changer for data science in Python, clearly had quite a few inspirations from Mathematica.)
  - leephillips 6y ago
    Yes, I think the idea behind notebooks like Jupyter came from Mathematica. Many people believe that Mathematica would have been huge if it had been open sourced. Now it's probably too late. For example, you can use many kernels with Jupyter, not only Python, so it's a very flexible interface. I enjoy toy models, either simplified equations or computational models like CA. I even made a toy model for the spread of corona virus, so I can look at the effects of isolation, death rate, etc. on the spread of infection, and to give my students something to play with during isolation: https://lee-phillips.org/corona/ https://lee-phillips.org/corona/ But I don't claim that this is a serious epidemiological model. Like CAs and the like, I'm hoping that it is fun to play with and might, just possibly, lead to some insight or new idea. But then that new hypothesis would have to be studied using a more serious model.

by stared 6y ago

leephillips 6y ago

I don't think we actually disagree; in any case, I can't find anything in your comment that I don't agree with. I think an important point is that the alternative formulations you list are provably equivalent to the theories that they reformulate; they can not make different predictions, but make make some calculations much easier and lead to expanded insights. It was a big deal when Schrödinger proved the equivalence of the two versions of QM (although his proof was incomplete; a rigorous proof was accomplished by von Neumann). As you suggest, the problem with Wolfram's computational approach is that since it is not equivalent to existing theory, except perhaps in some special cases, it must be a different theory, with different predictions. And, as of almost 20 years since _A New Kind of Science_, it hasn't predicted anything.

stared 6y ago
For a long time, I had thought that A New Kind of Science is an excellent math popularization book on cellular automata, prompting to do computational experiments in mathematics, and having an emphasis on beautiful visualizations. Also, his science education posts (e.g. on analyzing FB data) are wonderful. I just read a few chapters. And later, I was surprised by the criticism. It turned out that "A New Kind of Science" is not a tongue-in-cheek title (vide "Freakonomics"), but that SW actually believes that it not only is a new kind of science, but his science (Turning, Conway, Feynman, anyone...?). Some cellular automata simulations are nice. And in general, many concrete in physics are not general, yet are useful e.g. for numerics (with known limitations). They are not considered crackpot (even if not useful), as long as they make clear what they predict, what - not. Any toy model usurping its right to be the TOE, without a sufficient proof (not just the lack of proof it is not) is likely to be crackpotism. (Side note: for quite a few years I used heavily Mathematica. Only a decade ago I switched to Python, for a few reasons, but mostly for open-source (and quite a few of its implications). Jupyter Notebook, which IMHO was a game-changer for data science in Python, clearly had quite a few inspirations from Mathematica.)
- leephillips 6y ago
  Yes, I think the idea behind notebooks like Jupyter came from Mathematica. Many people believe that Mathematica would have been huge if it had been open sourced. Now it's probably too late. For example, you can use many kernels with Jupyter, not only Python, so it's a very flexible interface. I enjoy toy models, either simplified equations or computational models like CA. I even made a toy model for the spread of corona virus, so I can look at the effects of isolation, death rate, etc. on the spread of infection, and to give my students something to play with during isolation: https://lee-phillips.org/corona/ https://lee-phillips.org/corona/ But I don't claim that this is a serious epidemiological model. Like CAs and the like, I'm hoping that it is fun to play with and might, just possibly, lead to some insight or new idea. But then that new hypothesis would have to be studied using a more serious model.