Pity poor Popper

I don't subscribe to the view that science is all about falsifying things. It's rather popular on the internet but seldom crops up much in day-to-day research, which tends to be a lot more free-wheeling. Not that it never comes up at all - people do worry if their model is hard to test - but it's not the dominant strain of scientific thinking, if there even is such a thing.

As discussed at length previously, real science tends to be messy. When it comes to theories and models, broadly they can be said to be scientific and/or useful if : they can be compared with observations; they have different strengths and weaknesses that can be compared with each other; they are at least testable in principle if not yet in current practise; they offer markedly different fundamental mechanisms which affect our intuitive understanding of a situation. A model which does not do this last point, which does not describe the actual physics at work, is not a scientific explanation - no matter how precise its predictions or how thoroughly they can be tested. Hence astrological predictions are not scientific because they lack any mechanistic explanation.

This is a lot more sophisticated than the classic notion of falsification, which essentially makes the process weirdly half-binary : keep testing things until their eventual but inevitable destruction, at which point, come up with something better. Instead, I believe it's entirely possible to both prove and disprove something with sufficient evidence - and given sensible constraining assumptions. If you don't have those, then you will say you never know anything at all, and may as well sod off and stop annoying everyone. But, while being able to do this is certainly very nice, and always desirable, it certainly isn't necessary to doing useful science. Not at all.

But this Aeon piece goes too far*. It leans towards a view of science in which everything is possible, where falsification can essentially never happen. That's not science in my book, it's philosophy. And philosophy is feckin' important, but it ain't science.

* Hypothesis : any article which uses the word "neoliberal" is likely to be awful. Except when the offending word is in quotation marks, obviously.

The process of science, wrote Popper, was to conjecture a hypothesis and then attempt to falsify it. You must set up an experiment to try to prove your hypothesis wrong. If it is disproved, you must renounce it. Herein, said Popper, lies the great distinction between science and pseudoscience: the latter will try to protect itself from disproof by massaging its theory. But in science it is all or nothing, do or die.

Three philosophers were pulling the rug away beneath the Popperians’ feet. They argued that, when an experiment fails to prove a hypothesis, any element of the physical or theoretical set-up could be to blame. Nor can any single disproof ever count against a theory, since we can always put in a good-faith auxiliary hypothesis to protect it: perhaps the lab mice weren’t sufficiently inbred to produce genetic consistency; perhaps the chemical reaction occurs only in the presence of a particular catalyst. Moreover, we have to protect some theories for the sake of getting on at all. Generally, we don’t conclude that we have disproved well-established laws of physics – rather, that our experiment was faulty.

A classic, "yeah, but" moment. Yes, we need to consider implicit assumptions, faulty measurements and the like - and yes, sometimes we can even have more confidence in a theory than an observation. And yes once more, most of the time all we can do is vary probabilities : this new data increases our confidence in theory A, this one favours theory B, this finding suggests something entirely new that we hadn't thought of before. But while it's very stupid to view science as a set of binary true/false choices, it's not much better to suppose that it can never falsify anything. If we found a truly superluminal object, that would instantly falsify relatively. If we found a giant space cat, that would instantly disprove... err, well, lots of things, probably - and simultaneously vindicate anyone who had proposed the existence of a giant space cat.

The notion that science is all about falsification has done incalculable damage not just to science but to human wellbeing. It has normalised distrust as the default condition for knowledge-making, while setting an unreachable and unrealistic standard for the scientific enterprise. Climate sceptics demand precise predictions of an impossible kind, yet seize upon a single anomalous piece of data to claim to have disproved the entire edifice of combined research; anti-vaxxers exploit the impossibility of any ultimate proof of safety to fuel their destructive activism. In this sense, Popperianism has a great deal to answer for.

I just think that's daft. First, anti-science loonies will always be anti-science loonies; they aren't just demanding impossibly high standards of proof - they are just not very good at thinking. This idea that such people are just demanding unreasonably high levels of evidence, and are actually just as rational as everyone else, is pernicious : on the contrary, if you do actually talk to such people, it becomes very obvious very quickly that they are neither especially interested in the truth nor in any way rational : they are already convinced of their own ideas and use that conviction as circular justification for their belief. Second, I'm just not seeing a scientific culture dominated by falsification; at most, reviewers can be overly-skeptical but that's about it. It hasn't "normalised distrust" - science is always and unavoidably more complex than that, relying on a careful, considered, moderated approach, neither too trusting nor too skeptical. Just because some philosopher described it as something else hasn't had that much effect. Popper, in the end, was just not that important : reshaping a culture is a very great deal harder than writing a book.

The abuses of Popper

A powerful cadre of scientists and economists sold Karl Popper’s ‘falsification’ idea to the world. They have much to answer for.