Paul Feyerabend’s Against Method is a scandal. Most people who have heard of it know its tagline: “Anything goes”. As I mentioned in my previous post, my impression of the book from secondary sources was that Feyerabend was a madman and the book is sacrilege. Now, having read the book myself, I find myself impressed by the depth and clarity of his arguments and by his insight.
His key claim is that a successful (general) method of science is impossible and that trying to impose a (general) method is harmful.
In Feyerabend’s terminology, a method must “contain firm, unchanging, and absolutely binding principles for conducting the business of science” (p. 7 of the Fourth Edition, Verso 2010). To be counted as a success, such a method must “remain valid under all circumstances and [… be an] agency to which appeal can always be made” (p. 161).
I agree that all such general methods in the literature that I have been exposed to are failures, by Feyerabend’s standard. Neither of the theories I adopt in my doctoral dissertation (pending a public defense), the Bayesian approach to epistemology and Imre Lakatos’s theory of research programs, satisfy this test, and I freely admit this; both are very permissive and neither give objective and precise decision rules for considering the merit of a scientific hypothesis or theory, and thus do not count as methods under Feyerabend. And Feyerabend is quite correct (assuming his historical research is sound, which I am not qualified to judge) in his conclusion that no existing method (as the term is here defined) could have allowed certain key historic developments, and therefore none of them succeed.
For example, Popper’s falsificationism fails for two alternative reasons. If we suppose that it is a method (under Feyerabend’s definition of a method), then it must be followed literally in all cases, but in that case it fails the test case of Galileo, as discussed extensively in the book. But Popper can also be read metaphorically, or as general guidelines not to be taken as a literal method, in which case it can be understood to be consistent with the Galileo case; but then, it is (by assumption) not a method. In either case, it is not a successful method.
I also agree that it is probably impossible to come up with a successful method, by that standard. The history of philosophy is full of expounded theories, all of which seem to fail for some reason or other. It is very easy to move from this to a general scepticism: there can be no such successful method. It seems to me that it is also the correct (though defeasible) conclusion.
Further, if (as I have conceded) it is impossible to devise a successful method, then trying to impose a method is certainly harmful. I accept this.
The catch is here: this conclusion must be read with Feyerabend’s definitions firmly in mind. It is a misunderstanding of Feyerabend to further conclude that he denies the value of scientific methods. The singular in the title is a conscious choice, and very significant: Feyerabend does not oppose methods; he opposes a unified, one-size-fits-all method, singular.
Where Kuhn talks about paradigms and Lakatos about research programs, Feyerabend talks about traditions. Within a tradition, Feyerabend acknowledges there to be quite a bit of value in binding rules, and within a tradition there can be a successful method. Feyerabend’s “anything goes” is not a license to forget consistency requirements in a single piece of work or when working within a tradition:
Admitting velocities larger than the velocity of light into relativity and leaving everything else unchanged gives us some rather puzzling results such as imaginary masses and velocities. […] Admitting contradictions into a system of ideas allegedly connected by the laws of standard logic and leaving everything else unchanged makes us assert every statement. Obviously we shall have to make some further changes [… which] remove the problems and research can proceed as planned. (p. 246)
One of Feyerabend’s key conclusions is that traditions can only be evaluated from within a tradition, whether itself or some other one; an objective, meta-traditional evaluation is impossible. I will concede that his argument looks plausible (I will not review it here). Once this is accepted, it easily follows that no tradition is objectively better than all others. (I note that it is theoretically possible that some tradition appears superior to all other traditions viewed from any tradition; but certainly, no existing tradition has that rather remarkable property. Usually all other traditions appear weaker than the one being used as the vantage point.)
Feyerabend goes even further. He claims that traditions are incommensurable: a tradition involves a whole world-view, and there is no lossless conversion of ideas, thoughts and claims from a tradition to another. The only way to truly understand a tradition is to first become its adherent.
This conclusion seems rather absurd to anyone who has been educated in one tradition and has never made a leap from one tradition to another. However, the truth of this claim seems quite plausible from my own experience trying to read both quantitative and qualitative methodological literature: the former typically dismisses the latter as unscientific, and the latter typically dismisses the former as “positivistic” (that this label is a misnomer makes no difference). It is even more plausible to me having discussed methodology with both quantitative and qualitative researchers, and having observed discussions of methodology between quantitative and qualitative researchers. Usually, they talk past each other, each hearing nonsense from the other. It’s like they’re using different languages even though all are using ordinary scientific English.
Yet, I cannot accept incommensurability as a binding constraint. I hope to be able to transcend several traditions, to be able to work in them and hopefully function as a bridge of sorts. Maybe I am a lunatic in that hope; I do not know.
Finally, Feyerabend claims that because traditions are incommensurable and an objective comparison of them is impossible, there is no good reason why science should have priority in politics over any other set of traditions. Feyerabend died in 1994, before the evidence-based movement became the force it is today, but I suspect he would have loudly protested ideas like evidence-based policy (or more commonly nowadays, evidence-informed policy). He makes a forceful claim that basing policy on science is a form of slavery.
I can see his point but I am also in violent opposition. A lot of scientific activity is truly traditional, where things are done in a particular way just because they’ve always been done that way (though the “always” often can be as short a period as a couple of years), and when one goes to examine the history of that particular way, it turns out it was an accident, with no good rational reason for its adoption, and sometimes even was adopted despite good reason to abandon it. In these cases, adopting a scientific consensus position just because it is one is folly. And in general, where a decision one makes only affects oneself, it is better left to individual freedom and not impose any sort of an outside rule, whether scientific or otherwise. But there are quite many problems where a decision has to be made collectively, and I will vote for the decision to be evidence-informed; if I prevail, the only form of slavery involved is that of the tyranny of the majority.
In conclusion, it seems to me Feyerabend is more right than not, and that he has been mostly misunderstood (like he claims himself). I would recommend this book as a guide for anyone interested in multitradition (or mixed methods, as it is often called) work in the sciences. I would not recommend it as a methodology itself – and neither would Feyerabend:
‘anything goes’ is not a ‘principle’ I hold – I do not think that ‘principles’ can be used and fruitfully discussed outside the concrete research situation they are supposed to affect – but the terrified exclamation of a rationalist who takes a closer look at history. (p. xvii)
Let me repeat this: There is no Feyerabend method, and conducting research following Feyerabend is to misunderstand him. (At least to the extent one only considers Against Method; I have not read Feyerabend’s other work.) He preaches tolerance, but one should look for methodological guidance elsewhere.