by thray » Thu Aug 02, 2018 8:10 am
Having gotten distracted, I'm only 14% through Sabine Hossenfelder's wonderful book, Lost in Math: How beauty leads physics astray. I'm planning to get back into it.
Something jogged my memory about an exchange on B's blog a couple of months ago, that I hope you'll find relevant.
Sabine Hossenfelder said...
t h ray,
What you get from experiment is data that allows you to tell which theories work and which don't. Experiment of course doesn't give you a theory by itself, and if that's what you think I said you misunderstood. (To be more precise, I was probably referring to models, not to theories, but not sure the distinction matters for the present purpose.)
8:19 AM, May 16, 2018
t h ray said...
B,
A theory is a model. Without it, how do you know what you're looking for? What "works"? Then, suppose one has a theory of models that work (e.g. string theory), what experimental data would falsify the underlying assumptions? One can't say that string theory is unfalsifiable, on the one hand -- and yet accounts for every known physical interaction, on the other.
So one expects to find an unknown physical interaction, using known experimental methods?
I think the problem is flatland thinking, a problem that Joy Christian identified long ago. So long as one is working in a framework assuming 3 dimensions, one gets 3-dimension results--reasonable to assume, except that Christian has shown an extended framework includes hyperspace. Which necessitates dropping the 3-dimension assumption, replacing it if you will, with an assumption of 3-sphere (4 dimension) dynamics. And he has outlined experimental ways to falsify 4-dimension dynamics in our locally real 3 dimensions--that's the model, independent of a theory.
Unless one is willing, like Einstein, to challenge assumptions--axioms--one is unlikely to make the leap out of flatland. Worse, one imagines the leap and calls it magical. After all, that model works just fine in 3 dimensions.
10:27 AM, May 16, 2018
Sabine Hossenfelder said...
t h ray,
I don't see the point of this discussion, but when I say "theory" I do not mean "model." A theory is a prescription for how to identify mathematical structures with observables. If you want to make a prediction with it, however, you need a model. Example: The standard MODEL is a quantum field THEORY. If a model doesn't fit with data, it may be the model that's wrong or it may be the theory that's wrong. That Anti-de-Sitter space (a model) doesn't describe our universe, for example, doesn't mean that general relativity (a theory) is wrong. And so on. I know this isn't standard terminology, but please take it as my definition. I think it agrees reasonably well with how most physicists use the terms.
11:04 AM, May 16, 2018
t h ray said...
B,
"Example: The standard MODEL is a quantum field THEORY. If a model doesn't fit with data, it may be the model that's wrong or it may be the theory that's wrong."
Okay. Except the standard model is not a (complete) quantum field theory. String theory is, and it fits all the data. It's too successful in fact -- predicts 500+ vacua, and provides no way to determine the lowest state. So why shouldn't we just trust the standard model? -- it doesn't have enough dimensions. And so on.
We gotta get out of flatland. A flatland model won't do anything but verify its own assumptions.
11:39 AM, May 16, 2018
Sabine Hossenfelder said...
t h ray,
Every model has a limited range of applicability and the standard model works just fine, I don't know what your problem is with it. As to your comments about string theory: you can't make predictions with a theory, you need a model...
11:49 AM, May 16, 2018
t h ray said...
B,
I don't have any more problem with the standard model than, as you say, its domain and range.
As I said previously, a theory of models (string theory) does predict models. A model is for testing, not predicting--one would not have gotten the model without predicting it. Theory is primary.