Bringing science and philosophy back together

I've got a lot of mixed feelings on this one. There are aspects I strongly agree with, others I think are more questionable, and in the end I simply don't get the author's main point. What, exactly, does he want scientists to do differently ? I think the assumptions he describes are exactly what everyone already makes, implicitly, because there's no way to proceed otherwise. If the change really required is "urgent", then that suggests we're doing something fundamentally wrong, but he never says what that is. I do agree, though, with the sentiment that there are lessons from the scientific approach that could be applied more broadly.

Scientists today might not hold that theories can be ‘deduced’ from phenomena by induction, but they do hold that evidence alone (plus explanatory considerations) decides what theories are accepted and rejected in science. In other words, they take for granted one or other version of standard empiricism, the doctrine that evidence decides in science what theories are to be accepted and rejected, with the simplicity, unity or explanatory power of theories playing a role as well, but not in such a way that the world, or the phenomena, are assumed to be simple, unified or comprehensible. The crucial point, inherited from Newton, is that no thesis about the world can be accepted as a part of scientific knowledge independently of evidence, let alone in violation of evidence.

Well, yeah. I'd put more emphasis on the "plus explanatory considerations" bit personally, but basically... yeah. Theories that flatly contradict the evidence are wrong. Theories that agree with the evidence get to live another day (and can even be eventually proven correct, after a fashion). If they make testable predictions, so much the better. How could anyone argue with an evidence-based approach ? How would you ever justify saying, "I believe in this theory even though it gives totally different results from reality even in very carefully controlled conditions ?". You never would. Even the most irrational people at least wave their hands and say that something went wrong : the stars weren't aligned correctly, someone was too skeptical... no-one openly avoids making excuses.

This Newtonian conception of science bequeathed to philosophy a fundamental problem about the nature of science that, for most philosophers, remains unsolved down to today. It is the problem of induction... For example, if the accepted theory is Newton’s law of gravitation, one rival, up till now just as empirically successful as Newton’s theory, might assert: everything occurs as Newton’s theory predicts until 2050, when gravitation abruptly becomes a repulsive force. Another such rival might assert: everything occurs as Newton’s theory predicts except for gold spheres in outer space over 1,000 tons in mass that attract each other in accordance with an inverse cube law (instead of the inverse square law of Newton’s theory). These rivals are horribly disunified, and somewhat implausible: they are, however, for the moment, just as empirically successful as Newton’s theory.

Surely, if ever there were grounds to apply Occam's Razor, it would be here. You can't just tack on extra stuff to a theory for shits and giggles, you have to justify it. That's also partly what evidence is for : not just testing predictions, but inspiring ideas in the first place. You can't invoke entities beyond necessity. If your theory demands the existence of giant gold spheres for some reason, then fair enough, but if it doesn't, you don't have any grounds to include them. A theory should have explanatory power, so unless giant gold spheres actually play a role in that, you've gone from "legitimate scientific theory" to "making stuff up".

Evidence cannot verify a theory. It cannot even select a theory – since infinitely many disunified rivals will always fit the available evidence equally well, or even better. This famous problem – Hume’s problem of induction – in effect decisively refutes the Newtonian conception of science, still accepted by the scientific community today. 

I really fail to see how. Yes, you can just make up whatever extra nonsense you want and tack it on to a theory, but unless your additions actually change the theory's explanatory power, then you haven't really made a new theory at all. It's the fundamental mechanism the theory proposes that's important : everything else is superfluous. For example it's easy to change the theory of Newtonian gravity such that it doesn't obey an inverse square law, i.e. that it goes to the power of -2.01 or -2.02 and so on (it's also easy to refute this). But it's not at all easy to propose a wholly new mechanism, which is why it took centuries to go from the concept of gravity as a force to gravity as the curvature of spacetime.

The author then considers including Occam's Razor but doesn't think that it helps much :

But there is still a problem. If physics, in particular, persistently accepts unified theories only, even though endlessly many disunified rivals are available that fit the available facts just as well, or even better, this must mean, whether it is acknowledged or not, that physics makes a big, highly problematic assumption about the nature of the Universe. It means that physics makes the big assumption: the Universe is such that all disunified theories are false. There is some kind of underlying unity in nature.

Yes and no. Yes because it's still not clear why this is any kind of problem : of course we assume the laws of physics are the same everywhere, because without this you can simply invoke magic to explain anything; making assumptions isn't a problem in itself. But also no, because the evidenced-based nature of scientific inquiry makes it conceivable that these assumptions can be disproven. We can start by postulating laws that apply everywhere, but we could in principle discover regions where they don't apply.

This assumption of underlying unity is, however, accepted independently of evidence, even in a sense in violation of evidence (in that it clashes with endlessly many disunified theories even more empirically successful than the theories we accept). 

But it's not in violation of the evidence unless you have evidence all those other theories are true. Again, subtle modifications are not enough to make a new theory. Neither are adding in extraneous untestable components that don't change the explanation. You can't modify science the way you can add unrelated amendments to American laws.

The conclusion is inescapable: science cannot proceed without making, implicitly or explicitly, a persistent metaphysical assumption of unity – ‘metaphysical’ because it is too imprecise to be verified or falsified by evidence. 

Yes, and there's no problem with that.

The current orthodox conception of science, inherited from Newton, and still taken for granted by scientists today, that science must appeal only to evidence, and must not make metaphysical assumptions about the nature of the universe independently of evidence, is untenable, and must be rejected.

I'm not sure this is wrong so much as it's an incredibly silly way of phrasing things. I've discussed some of the basic assumptions of science before (though the author's approach is more sophisticated and interesting than mine), but I think they are implicitly obvious to everyone : awareness of them doesn't change the scientific approach; they are already part of the scientific approach for very good reasons.

Aim-oriented empiricism, as I have called this new conception of science, represents the metaphysical assumption in the form of a hierarchy of assumptions. As we go up this hierarchy, assumptions become less and less substantial, and so more and more likely to be true, and more and more nearly such that their truth is required for science, or the pursuit of knowledge to be possible at all. In this way, we create a framework of assumptions (and associated methods) high up in the hierarchy, very likely to be true, within which much more substantial assumptions (and associated methods) low down in the hierarchy can be critically assessed and, we might hope, improved.

I've also discussed a plausibility index of scientific theories and how they move up and down the ladder according to evidence. Extending this to methods as well as conclusions is perfectly reasonable, and provides a clear reminder that we're making assumptions. But I see no way to avoid making the assumptions we've already chosen to make, so I don't see how it changes anything. It's an interesting article, but I come away no clearer than when I started as to what the problem is, why it needs solving, how it can be solved, and what the consequences will be if it is.

Bring back science and philosophy as natural philosophy - Nicholas Maxwell | Aeon Essays

There are decisive grounds for holding that we need to bring about a revolution in philosophy, a revolution in science, and then put the two together again to create a modern version of natural philosophy. Once upon a time, it was not just that philosophy was a part of science; rather, science was a branch of philosophy.