Against Method

by Paul Feyerabend

I think this excerpt from the Introduction to the Chinese Edition summarizes the book best:

One consequence of the thesis is that scientific successes cannot be explained in a simple way. We cannot say: ‘the structure of the atomic nucleus was found because people did A, B, C …’ where A, B and C are procedures which can be understood independently of their use in nuclear physics. All we can do is to give a historical account of the details, including social circumstances, accidents and personal idiosyncrasies.

Another consequence is that the success of ‘science’ cannot be used as an argument for treating as yet unsolved problems in a standardized way. That could be done only if there are procedures that can be detached from particular research situations and whose presence guarantees success. The thesis says that there are no such procedures. Referring to the success of ‘science’ in order to justify, say, quantifying human behaviour is therefore an argument without substance. Quantification works in some cases, fails in others; for example, it ran into difficulties in one of the apparently most quantitative of all sciences, celestial mechanics (special region: stability of the planetary system) and was replaced by qualitative (topological) considerations.

It also follows that ‘non-scientific’ procedures cannot be pushed aside by argument. To say: ‘the procedure you used is non-scientific, therefore we cannot trust your results and cannot give you money for research’ assumes that ‘science’ is successful and that it is successful because it uses uniform procedures. The first part of the assertion (‘science is always successful’) is not true, if by ‘science’ we mean things done by scientists – there are lots of failures also. The second part – that successes are due to uniform procedures – is not true because there are no such procedures. Scientists are like architects who build buildings of different sizes and different shapes and who can be judged only after the event, i.e. only after they have finished their structure. It may stand up, it may fall down – nobody knows.
But if scientific achievements can be judged only after the event and if there is no abstract way of ensuring success beforehand, then there exists no special way of weighing scientific promises either – scientists are no better off than anybody else in these matters, they only know more details. This means that the public can participate in the discussion without disturbing existing roads to success (there are no such roads). In cases where the scientists’ work affects the public it even should participate: first, because it is a concerned party (many scientific decisions affect public life); secondly, because such participation is the best scientific education the public can get – a full democratization of science (which includes the protection of minorities such as scientists) is not in conflict with science. It is in conflict with a philosophy, often called ‘Rationalism’, that uses a frozen image of science to terrorize people unfamiliar with its practice.

Well, I can’t really do better than Feyerabend at summarizing Feyerabend. But maybe I can critique it.

First, I think I might suggest that epistemological anarchism is appropriate for gods, not for men. All models are wrong, but some models are better than others, and we puny finitely-resourced computational machines need some general heuristics for truth-seeking. And it turns out that some sets of these general heuristics are useful not just descriptively but also prescriptively, in the sense that they are meta-predictive of the physical laws of our universe. Our current scientific progress is one such example. Scientists follow roughly similar prescribed “rules” (or methodologies, also a subject of the author’s ire) and as a result gain “knowledge” of the world much more efficiently than if they did not.

Second, I must admit that his humanitarian impulses are well-founded, but his prescriptions are less so. Alongside the (almost Marxian?) attempt to prove the inherent contradiction present in any truth-seeking methodology, he includes various political arguments with regards to the negative effects of group-think and majoritarian ideologies. Yes, echo chambers are bad. Echo chambers are bad for good epistemics. Communities with ideological diversity are, on the whole, more tolerant and empathetic than those without. Understanding the nature of scientific discovery in its historical context will improve one’s understanding of the world. These are not commensurate with his overarching claims (and neither should they be, that’s what the other half of the book attempts to do).

Third, I really enjoyed his Popperianism critiques. Popper advocates a form of “critical rationalism” which necessitates putting forward theories, viciously attacking them, and reforging them stronger to combat even greater criticism. To this, he says:

Now at this point, one may raise two questions.

Is it desirable to live in accordance with the rules of a critical rationalism?

Is it possible to have both a science as we know it and these rules?

As far as I am concerned, the first question is far more important than the second….
But these are not the problems I want to discuss now. In the present essay I shall restrict myself to the second question and I shall ask: Is it possible to have both a science as we know it and the rules of a critical rationalism as just described? And to this question the answer seems to be a firm and resounding NO.

Why? Well, it appears that the actuality of scientific discovery does not neatly fit into a framework of “identify problem, propose theory, critique theory, propose better theory”. Instead, it originates from this nebulous entanglement of intuition, rigor, and fortunate happenstance that seems to indelibly characterize all great breakthroughs. Galileo, Newton, Faraday, Einstein—can we really say that these men were perpetually rigorous?

Yet ultimately, this falls victim to another case of the claim is simply not commensurate to the evidence. He has wonderful critiques of scientism, of the state of discovery in the 1970s and 1980s, of the astounding lack of intellectual freedom on academic campuses, and of existing attempts at making a scientific philosophy. But to take this information (which is primarily pragmatic in nature) and from this generalize to the advocation of the dereliction of the standards by which we approach empiricism is too far. It lacks nuance. It lacks a concept of degrees of correctness.

It is good that this was written. Its critique is, in a sense, perpetual. Any attempt at a philosophy of science can be critiqued on the basis that it is not enough, that it leaves value on the table by not capturing elements of the natural world that ought be understood. But alas, we are mortal beings, and something is better than nothing.

(This is a wonderful quote, by the way:)

There is no guarantee that scientists will solve every problem and replace every theory that has been refuted with a successor satisfying the formal conditions. The invention of theories depends on our talents and other fortuitous circumstances such as a satisfactory sex life. But as long as these talents hold out, the enclosed scheme is a correct account of the growth of a knowledge that satisfies the rules of critical rationalism.

Also kudos to him for writing a timeless work in an obviously timed fashion tinged by his age and context. That, that is worth applauding.