hypnagogue
Physics is in no dire need to answer metaphysical questions, of course not (sounds rather obvious in those terms). Physics gets along fine without any major contributions from philosophy-- agreed. I'm not trying to point out a major flaw in physics or advocate a renovation of how it's done.
Well the majority of the philosophers of science (by the way almost all have scientific training being physicists and so on at base) are no more so sure that the difference between science and metaphysics is so clear cut as Sir Karl thought once and how many scientists still believe even now.
The reality is that the history of science,what happens now at the frontiers of science and even logic itself have shown that popperian falsificationism is at least not satisfactory.Any unbiased course in philosophy of science point to this,presenting also the logical difficulties of falsificationism (as happened once with logical positivism-totally unsatisfactory on logical grounds).
Even if we accept the revised version of falsificationism,which require the unity of the theories (in the Ayer's sense,but without making any reference to weak 'verification' or meaning) we still have the problem of key unobservables in very successful otherwise scientific theories.
Even if we could 'verify' them somehow indirectly in a crucial experiment (of the type of that of Rutherford which finally convinced even physicists to accept the existence of atoms) we still have the problem of theory ladeness.
Logically it is still possible that some very fruitful otherwise theoretical concepts (such 'entities' are not testable in isolation) do not exist in reality.Thus is conceivable,as a possibility,that further developments would require to renounce at them altoghether in future theory making (not only at some minor attributes attached to them)!This 'mere' possibility (toghether with the possibility of Quinean underdetermination) and the Duhemian underdetermination (which is a fact in spite of some here failing to understand it's importance) is more than enough to not be able to talk of a clear demarcation between science and metaphysics.
Neither are we allowed to grant to a scientific method based on fallibilism and falsificationism (even methods going well beyond the mere popperian falsificationism) a perpetual epistemological privilege.Moreover if strong Quinean underdetermination (irrespective [of] data) were proved to be more than a mere possibility then we would not be even entitled to talk of epistemological privilege of science (based on fallibilism and intersubjectivity)...
Some feyerabendists take the strong underedetemination as granted by I argue that we do not have yet sufficient reasons to believe this (finding an alien civilisation would be very instructive for,if the strong underdetermination thesis is right,then we should expect that their set of scientific-in physics at least-enunciations are totally different,anyway very different,from our's,after the 'translation' in a human language of course).
Now all this does not mean that modern philosophers of science want the demise of falsificationism,no.Popper's version of falsificationism involve a methodological
choice,counting only as a mere logical argument,favorizing falsifiable hypotheses.Based on the common sense,logical,expectation that a good model of the universe must have internal unity and coherence,must make novel,testable,predictions and resist severe attempts at potential falsification ('corroboration' of the theory).It [has problems] but is,still,a good criterion,though in itself insufficient.
The idea is that falsificationism alone is one of the [key] criterions we should use but this in no way should restriction the freedom of movement (in the light shed by the clear limits of falsificationism).Indeed sometimes ad hoc hypotheses deserve to be part of science,to be the first choice program more exactly (though such ad hoc hypotheses are metaphysical in the popperian methodology).
Not all ad hoc hypotheses are on the same level of rationality,in spite of the fact that we might have problems even here we can still define a clear ladder of rational preferences (whilst still taking in account all proposals,even those who seem not reasonable at a certain moment).
The key factor is that ad hoc hypotheses should solve some key 'puzzles' or anomalies.Additionally we can define an orientative 'official' ladder of preferences,that is what programmes ought to be pursued as the first choice and so on (if exist more proposals).But this 'official' list (based on previous experience) in no way reastrict the freedom of scientists who can have their 'private' list of preferences.The idea is that finally to obtain a program which fully comply with the standard popperian requirements.In enough many cases ad hoc hypotheses initially did lead to full programmes,as history of science do prove plenty.
This was the case with the quanta hypothesis or with the neutrino hypothesis at their time (from what I've read Pauli writes somewhere that even if he and his colleagues were '[embarassed]' first of the totally ad hoc character of the neutrino hypothesis they still thought that it was much more than mere 'metaphysics').
Sure it is possible that some of those might not prove capable to help the conjectures they are part of to become theoretically and empirically evolving on medium term.However there is no need to keep them inside science on long term (or continuing to prefer them as the first choice program) if we have sufficient reasons against this (the program is stagnant,degenerative or other programmes are progressive,much more successful).
There is no mistake to proceed so for anyway in the light of what I said before (the 'normal science' itself is a mere program on which scientists spend more of their efforts trying to develop it further).Basically nothing is indispensable.
Let me explain my point of view from a different perspective (above what I said before).Well all (at least in physics,but this is valid in broad lines for science in general) all existing scientific programmes counting as 'normal science' have at least puzzles (facts not accommodated yet within theory,the 'problems' of Popper) if not anomalies (but much more often they have).
The first task of scientists,in the scientific tradition,is the accomodation of those puzzles within the existing system,paradigm.Thus even in cases that a paradigm considered as 'normal science' become degenerative (more and more puzzles,at least,do accumulate) scientists still try to stick to the letter with the existing paradigm,still try to accommodate all those puzzles or anomalies inside the existing normal science.
This does not really let enough space for creativity,scientists tend to not follow more variants in the same time and devise potentially new hypotheses (whilst still spending their most efforts by trying to accommodate the puzzles within the existing paradigm).On the contrary they tend to grant to the existing paradigm the capacity to explain those puzzles later,in the vrtue of scientific tradition and epistemological conditioning during the years of learning.
This could lead at least to important detours.I'd argue that instead of attaching to the existing paradigm and not trying to devise alternatives until much later (by pursuing the existing program until much later as the main choice,the only choice in many cases) it would be much more useful from the good beginning (at all times as a matter of fact) to devise as many as possible alternatives (as I've argued before we can define a ladder of preferences for them) though they seem totally unreasonable at a certain moment.
I argue that is a legitimate action sometime to prefer other programmes,as the primary choice,in the process of theoretical development,which solve a certain 'puzzle',by using new theoretical concepts if 'puzzles' begin to pile up,even when there exist only few.The first choice on that ladder of preferences among different ad hoc hypotheses (as the main choice only,without renouncing to implicate actively on other paths) is the addition of new 'constructs',totally coherent with previous knowledge.
They might seem ad hoc at a certain moment but further theoretical and empirical work could prove that they represent indispensable constructs,enabling a greater fecundity for the theories incorporating them.It deserves sometime to follow such paths,especially when such ad hoc constructs 'fit' naturally with the other parts of the existing accepted enunciations.
In the latter cases scientists might accept those constructs as part of the existing normal paradigm itself.As such not only that they try to find further direct or indirect empirical support for that construct [in a key experiment] (or merely theoretical support),that is try to make them 'reliable concepts' within that paradigm,but they USE THEM (even when still ad hoc) together with other enunciations,in other,TOTALLY UNRELATED,'problems' (in other,higher level,sciences for example).
There is no mistake to do so since scientists use indispensable nonobservables in certain successful theories to explain 'puzzles' in other sciences even though those entities could be only some constructs which happen to 'work'.This is possible we do not have sufficient reasons to think otherwise,unfortunately the 'no miracle' argument is,still,not sound.
I mean that those unobservables do make testable predictions in conjunction with other enunciations but they are not testable in isolation-there is no known separate experiment which to 'confirm' directly or indirectly their existence.Anyway indirect confirmation involve an inductive process,using often Mill's methods,and might [also] be theory ladden.Nothing impedes the possibility that intial ad hoc hypotheses could lead to huge 'leap' ahead.What is clear for all unbiased philosophers is that the history of science proves plenty that a too rigid method is not the fastest way toward 'success' (understood as mere fecundity).
) then I think another one is in order. Actually, I see your point. I now no longer believe that ecomonics is worth studying. Economics is useless because it has no role in explaining the mystery in high-Tc superconductors! 
