This point seems to be lost on a bunch of physicists these days who wallow in their derivations and forget that the fundamental revolution in modern science was the empirical method. The silly philosophers who got overthrown never knew it.
Animals understand "the empirical method". Experimentation, pattern inference, cause and effect... Aristotle may have described it, but like most of the rest of the interaction of philosophy and science it's a post-hoc formalisation, mostly good for convincing philosophers that scientists are doing something reasonable.
That's why they should learn some philosophy of science.
Here's a philosophy-free formulation of science: (i.e. one where scientists don't have to know any philosophy)
- People set up symbol games with rules, that produce symbols.
- They also set up physical machines that produce symbols.
- When they match, they're happy.
I think that does a good job of capturing why physics was able to progress so well without concurrent progress in philosophy. At the end of a day, a physicist will have a number on their scratch paper and a number on the meter and the goal is to get them to match. Mainly the role of philosophy is to guard them from any distractions that would confuse them away from doing that, and perhaps to bend their brains into coming up with good symbol games.
Philosophy informs these questions. For instance, a scientific process modelled on Solomonoff induction would give you more insight on what kind of games you should set up and why, which games are better than others for the same symbol outputs and why, to what extend modifying a game is a valid approach, etc.
If you have multiple matching symbol games, maybe just either prove them mathematically equivalent (a la matrix mechanics and Schrodinger's equation), or if they aren't equivalent, write them both down and keep an eye on them until you develop an experiment that can distinguish between them. As far as I can see, that remains within the paradigm of "go forth and match the symbols."
But your rules provide no guidance to do so, which would require philosophy.
> As far as I can see, that remains within the paradigm of "go forth and match the symbols."
Absolutely, I'm not suggesting changing the spirit of the basic empirical game you outline, merely that a different set or a few additional rules would permit you to generate better games more efficiently. Blind trial and error would eventually work too, but we ideally want to converge on accuracy and precision as fast as possible. What this means in practice is a philosophical question worth answering.
It seems suspicious that this is classified as philosophy - it has an empirical foundation! You can even design experiments to test these "theories of metascience:" just check to see how fast the scientists that work according to them produce successful theories. So it sounds like what you're advocating for is a human science (i.e. very difficult to design experiments, but fundamentally based on empiricism), as opposed to a branch of philosophy.
Roughly speaking, the division I am making here goes: experiments plus arguments makes science, arguments without experiments makes philosophy. In this context, I can say, "scientists are only (professionally) interested in experiments and therefore uninterested in philosophy." However if something involves empiricism but is also labeled philosophy, that's just language and I gladly concede that scientists might be professionally interested in it.
I'm very skeptical that such a foundation would work faster or achieve more optimal results than an analytical approach. You can keep going meta, but so can I, and the analytical approach will always outpace the empirical one.
> Roughly speaking, the division I am making here goes: experiments plus arguments makes science, arguments without experiments makes philosophy.
I suppose it depends what you classify as "arguments". Science traditionally produces theories that provide predictive power, but also explanatory power. This latter aspect is often overlooked, but it's essential since it provides insights into further possible experiments. If your "arguments" includes an ontology for scientific theories, then I could agree with your breakdown, but notice that an ontology overlaps with philosophical inquiry.
Also, there's a new and growing branch called "experimental philosophy" which includes experiments that test intuitions on philosophical questions, so it's still not quite clear cut!
It sounds like you're describing and empirical meta-approach: that we should pick the approach that goes the fastest. I'm not proposing that analysis be thrown away, just that it is and should be controlled by empiricism (at every meta-level). Essentially, I'm arguing that science applies to itself recursively: if you take empiricism as axiomatic, then science can "choose" its own philosophy, just like it chooses its theories: empirically.
>Also, there's a new and growing branch called "experimental philosophy" which includes experiments that test intuitions on philosophical questions.
In that vein, you could claim that all philosophy is empirical: An Enquiry Concerning the Principles of Morals was plain evidence that Hume's brain had produced certain beliefs about morals. So, is philosophy of science an exploration of its namesake, or an exploration of the brains of philosophers of science? I say, whenever it has empirical grounding (in the success of science), it works like science and tells us things about how to do science, and when it's inwardly-focused it is just information about how philosophers' brains work.
>but also explanatory power
As far as I can tell, explanations are stories that make our brains feel comfortable with the procedures that we use to predict things. Why? Because perfect stories will be discarded without a second thought if another completely unpalatable theory comes along with better predictive power. Story-sentences about this stuff (for example, "the virtual particles pop in and out of existence in the vacuum") carry as much information about our brains as they do about the universe that they're supposedly describing.
It's not at all clear what "take empiricism as axiomatic" even means. This itself consists of a few a long-running philosophical debates:
https://plato.stanford.edu/entries/constructive-empiricism/
https://plato.stanford.edu/entries/rationalism-empiricism/
https://plato.stanford.edu/entries/logical-empiricism
> Because perfect stories will be discarded without a second thought if another completely unpalatable theory comes along with better predictive power.
I disagree. Newton's theories are still widely taught despite their inaccuracy, because their predictive power suffices and their explanatory power is superior to more accurate but sophisticated theories.
Furthermore, there's a good reason why the foundations of quantum mechanics are still hotly debated in some circles, despite the fact that such debates will have little practical impact on most practicing physicists. And such debates over explanatory power have yielded valuable insights into QM, like Bell's theorem which resulted from Bell's interest in the de Broglie-Bohm theory, and numerous other no-go theorems.
Certainly a theory with superior predictive power will be employed for predictions, but it will immediately launch an effort to provide a coherent explanatory framework.
The philosophy of science I am arguing for is empirically testable: so, as far as I can see that makes it science as opposed to philosophy.