3 ms·

> That alone doesn't guarantee that those theories would fit future data. Sure, but that's sort of a triviality. It's true for any theory. > It's a two-way pr

> That alone doesn't guarantee that those theories would fit future data. Sure, but that's sort of a triviality. It's true for any theory. > It's a two-way process. You need experiments to improve theory, and you need theory to create experiments. I agree that it's a two-way process but I wouldn't say you need theory to create experiments and I think I've given sufficiently many counter examples to illustrate this. What we (as humans) do need theory for is to make sense of our experiments (or of the world, really) and discover new experimental setups that might uncover something interesting (not explained by the current theory). On the other hand, without theory there's no concept of "unexplained by theory" in the first place. Our experiments would probably resemble random walks (in the sense that they're not driven by theory) but that doesn't mean we couldn't uncover interesting phenomena, let alone run experiments at all. Getting back to the "making sense of the world" point: Take an AI for instance. It could probably be trained on experimental data alone and make suggestions for future experiments and give predictions, without a simple, coherent and human-understandable theory behind them[0]. This, too, illustrates that experiments would be possible (and worthwhile) even without theory. Note that I'm not at all trying to play down the role of theory. To the contrary, I'm saying all this as a former theoretical physicist myself. I just see experiment and theory as equals. [0]: Isn't this more or less what a human baby does as it learns?

LudwigNagasena 4y ago
> Sure, but that's sort of a triviality. It's true for any theory. It is not a triviality. OP and many other people insist that nothing is important unless there is empirical evidence. Was there any empirical evidence that the Higgs boson is real and all competing theories are wrong before physicists spent billions of dollars to find it? Yet it was one of the most convincing theories, even if most of the evidence was theoretical, intuitive and aesthetic rather than empirical. I think you, as a former theoretical physicist, can come up with a dozen theories that are uglier but could fit old data just as well. The most trivial way to do it is just to say that in those experiment we should’ve expected those results and so on without any generalization. But of course numerous more sophisticated wrong theories would be possible. There is no theory that is so bad that it can’t be “fixed” with a few epicycles. By the way, Ptolemaic vs Copernician model is another good example of two different theories with similar or exactly the same fit w.r.t. to old experiments and observations. > This, too, illustrates that experiments would be possible (and worthwhile) even without theory That just illustrates that you can build a theoretical model with lots of free parameters and low interpretability. > Note that I'm not at all trying to play down the role of theory. Right. And I was just responding to OP who completely downplayed it.

LudwigNagasena 4y ago

> Sure, but that's sort of a triviality. It's true for any theory. It is not a triviality. OP and many other people insist that nothing is important unless there is empirical evidence. Was there any empirical evidence that the Higgs boson is real and all competing theories are wrong before physicists spent billions of dollars to find it? Yet it was one of the most convincing theories, even if most of the evidence was theoretical, intuitive and aesthetic rather than empirical. I think you, as a former theoretical physicist, can come up with a dozen theories that are uglier but could fit old data just as well. The most trivial way to do it is just to say that in those experiment we should’ve expected those results and so on without any generalization. But of course numerous more sophisticated wrong theories would be possible. There is no theory that is so bad that it can’t be “fixed” with a few epicycles. By the way, Ptolemaic vs Copernician model is another good example of two different theories with similar or exactly the same fit w.r.t. to old experiments and observations. > This, too, illustrates that experiments would be possible (and worthwhile) even without theory That just illustrates that you can build a theoretical model with lots of free parameters and low interpretability. > Note that I'm not at all trying to play down the role of theory. Right. And I was just responding to OP who completely downplayed it.