Passing paradigms

I now turn away from my mythology binge – which continues in the background unabated – and back to an assortment of articles I've been meaning to read for bloody ages. Let's start with this one, a piece from the London Review of Books on scientific paradigms. Sounds dry ? Well, read this bit and then decide :

As Morris tells the story, ‘the bile just flowed out of him.’ Kuhn ‘started moaning. He put his head in his hands and was muttering, “He’s trying to kill me. He’s trying to kill me.”’ At the end of his tether, Kuhn threw an ashtray ‘with malice’, maybe at Morris, maybe just in his general direction: ‘It came hurtling across the room, spewing butts and ashes.’ (Kuhn and ashtrays were constant companions; he smoked upwards of six or seven packs of cigarettes a day, and died of throat cancer.) First Kuhn threw the ashtray, then he threw Morris out of graduate school. 

If I were the editor, I'd pick something from that for the big quote-in-bold designed to grab the readers attention. But what in the world could provoke such fury in the ivory-tower world of philosophy of science ?

That would be Thomas Kuhn's one-hit-wonder of 1962 : The Structure of Scientific Revolutions. I've had this in the back of my mind for a while as something I should probably read, but after reading this piece I've decided firmly against it. There's too much in it that would annoy me, but also far too much I'd misunderstand. Kuhn himself wasn't exactly thrilled by the book's reception :

Invited to a seminar by Princeton undergraduates inspired by what they took as the book’s anti-authoritarianism, Kuhn erupted, telling them that they just hadn’t read the thing properly: ‘I kept saying, “But I didn’t say that! But I didn’t say that! But I didn’t say that!”’

So I'm probably far better off with the article, providing the author, one Steven Shapin, has done their job well enough. I'll pass over the biographic stuff and stick to the ideas. 

Very broadly, Kuhn seems to have been the one who came up with the idea that science advances in revolutions. We tinker away diligently enough, but then some moment of genius happens which throws it all away, and suddenly we're in a whole new world. This idea seems so ingrained in popular culture that I took it for granted that it was just a natural conclusion that everyone reached; to find out that this is not so is interesting by itself (though what the model of progression was before Structure Shapin does not say). 

Regular readers will know I despise this idea. Or more accurately, one particular aspect of it : that singular geniuses are largely responsible and we an essentially ignore everyone else (the "great man" of history theory). And though of course our ideas and world views shift, I'm less sure of whether that feels like the popular notion : a sudden moment of revelation in which the nay-sayers are cast down and the plucky underdogs revealed as unfairly-shamed geniuses. I rather doubt it ever seems like that; popular science reporting has it that something akin to this happens all the time whereas to those of us on the inside it seldom feels much of anything like the popular depictions.

Anyway, the initial definitions of paradigms :

Kuhn plucked the word ‘paradigm’ from linguistics – where it referred to the permutation of forms having a common root, like the conjugation of verbs or the declension of nouns – and repurposed it as the term for a key regulative resource in scientific inquiry, a concrete model of ‘the right way to go on’... A New Yorker cartoon shows tramps leaning against a wall: ‘Good news – I hear the paradigm is shifting.’

Paradigms are one of the few bits of philosophy of science that we were explicitly taught as undergraduates. Our lecture's model was simply that the paradigm is the prevailing wisdom, the conceptual framework by which we default to interpreting data. Paradigms, he said, can of course change. Such a shift might or might not be dramatic and exciting, depending on how it happened. But though it would presumably be beneficial, it wouldn't be inevitably good or bad, it would just be a thing that happens. This is a much more neutral view. Only once does the article come close to this, with Kuhn later describing paradigms "as concrete, non-rule-like regulative structures".

Nothing here seems terribly offensive. So what got people so riled up that "groups of philosophers had "‘gathered and said that the book should be burned’" ?

Not so much, perhaps, of the nature of paradigms themselves. Having a framework within which to interpret data is inevitable. What seems to have really ticked people off is the nature of how Kuhn proposed they changed and who got to decide when such a change had occurred. In standard prevailing wisdom :

Science methodically compared theoretical expectations against observational and experimental evidence; it purged itself of bias and prior expectations; its knowledge was cumulative; the quality of that knowledge was guaranteed by explicit methodological standards shared throughout the scientific community; the various bits of science were part of a fundamental unity, whether of concepts, facts, or methods; it arrived at, or at least approached, truth. 

But :

Structure denied all this. Scientists were not notably open-minded. Their training encouraged the embrace of what Kuhn frankly called ‘dogma’: theirs was ‘a narrow and rigid education, probably more so than any other except perhaps in orthodox religion’. If the account in Structure were accepted, Kuhn wrote, the notions of ‘“scientific progress” and even “scientific objectivity” may come to seem in part redundant.’ We may ‘have to relinquish the notion’ that scientific change brings scientists ‘closer and closer to the truth’. Scientific knowledge did not accumulate: it moved from moments of puzzle-solving ‘normal science’ governed by one paradigm, through ‘crisis’ and ‘revolution’, to subsequent moments of ‘normal science’ governed by another paradigm... there was no independent way of adjudicating between them; the embrace of a new paradigm ‘can only be made on faith’.

...either side of a paradigm shift scientists live and work ‘in a different world’. There were no bodies of facts, methods for establishing facts, or theories for interpreting the facts, that were paradigm-independent; there was no ‘neutral algorithm’ for judgment. This was the Kuhn who was taken to be a reality and reason-denying relativist.

Ahh, now this is spicy stuff indeed. A rant here on my part would seem inevitable, were it not that the article reveals that this is far more of how Kuhn was misinterpreted rather than what he actually meant. 

Two thoughts though. "Dogma" is a trigger word, so I have to say I don't believe there is such a thing as scientific dogma* : provisionally. That is, I don't believe we have things we believe with inflexible rigidity. We use the term "Law" only to underscore the robustness of a finding, not to declare from on high that it is forever unchallengeable. The only sense in which I think we have "dogma" is a much weaker one. We have common "maxims, norms or (his preference) values of science"; there are useful heuristics for interpreting data, default frameworks which can safely be assumed to correct for everyday purposes. These are by no means set in stone, we just don't challenge them routinely because we feel that to do so would make no progress, with there being more obvious mysterious avenues that would be better explored instead.

* As for "faith", that is too big a topic to condense here, so see this post instead.

Secondly, the profound revelation of a paradigm shift according to these (mis)interpretations of Kuhn... well, this reminds me of one of my all-time favourite Star Trek quotes :

For that one fraction of a second, you were open to options you had never considered. Not mapping stars and studying nebulae, but charting the unknown possibilities of existence.

Scientists are not closed-minded to such revelations. On the contrary, I claim that such things are, in a smaller way, very normal. To illustrate : a useful tactic, I've found, is to think on different scales. Especially while coding. It happens regularly that I find my code doesn't do the thing I want it to do, and my first efforts are always to check for the small stuff : maybe I used the wrong variable, got a positive or negative sign wrong, that kind of thing. Then I move up a level to checking the structure : maybe I haven't indented a loop correctly, maybe I've not iterated correctly, etc. 

But if all that seems secure, I move to the largest scale (or if you prefer, higher level) : the basic premise of what I've written. Maybe the actual line-by-line execution is doing exactly what I wanted it to, but I've got the whole foundation of how to do the task wrong. Figuring out that I've done the wrong thing in the right way is, in some ways, one of the most rewarding parts of debugging code, albeit sometimes the most frustrating as it means not a mere tweak to a section but a full rewrite.

Paradigms, I claim, are just like this. They're on a grander scale to be sure but the principle is the same. And I don't think it's true at all that scientists aren't especially open-minded towards paradigm shifts. Rather we go deliberately looking for them, but we're still skeptical when we find evidence of them. Such a profound change deserves a very high level of confidence, just as you wouldn't want to start believing you could suddenly breathe underwater without some seriously compelling reason.

Even this, though, may not be enough to warrant hurling an ashtray at someone or calling for the book to be burned.

Structure implicitly challenged the notion that there was such a thing as a unified scientific community; rather, there were many communities, each of them organised through its commitment to specific achievements, specific methods and specific standards of fit between expectation and evidence. The much treasured idea of ‘scientific unity’ was also sacrificed: science was a ‘ramshackle structure with little coherence among its various parts’.

This I think is actually a good description of academia but a lousy one of science itself. Academics and individual disciplines certainly have different standards and methodologies, but the collective scientific world view is stupendously self-consistent. To call it "ramshackle" is nonsensical.

But there were even more controversial parts of Structure :

One philosopher targeted what he called Kuhn’s ‘purple passages’ – for example, where he said that there was no standard for scientific judgment higher than ‘the assent of the relevant community’ ... Campus radicals seized on Kuhn’s book as a brilliantly subversive exposé. Just as they had suspected, science wasn’t the open-minded objective pursuit of truth, but merely one more mode of authoritarianism. Scientists were just as dogmatic as anyone else, and one way of seeing the world was as good as another. If there were no better criteria for judgment than communal assent, why should anyone bow down to scientists’ pronouncements? Oh thank you, Mr Kuhn, for telling us about paradigms,’ he remembered the students saying. ‘Now that we know about them, we can get rid of them.’

Kuhn stood accused of being yet another philosophical ‘corrupter of youth’. Philosophy of science here bled seamlessly into Cold War politics.

That would seem to explain the vitriolic responses. When you both treat science as yet another religion, another mythological framework to be debunked, when you see scientists as another type of authoritarian High Priest, you essentially demolish the whole edifice. It denounces the entire practise as not worth doing. And that is deeply, deeply insulting. And just plain wrong.

But this doesn't seem to have been Kuhn's intent at all. I will attest myself, again on an altogether different scale, to having been misinterpreted on many occasions. This is deeply frustrating, because instead of writing what I want to write in a way that expresses things in a (to me) lively but clear way, I have to add caveats and provisos and clarifications that can rob the text of all its force. So I'm gonna give Kuhn the benefit of the doubt on this one.

What he actually seems to have been driving at is much more modest :

You should not, Kuhn had written, think that scientific change brought practitioners ‘closer and closer to the truth’. The outcome of change, in Kuhnian terms, was ‘the selection by conflict within the scientific community of the fittest way to practise future science’. 

This I think it would be harder to have much of a problem with. What seems the most promising line of inquiry ? Well, first we assume the code is basically okay, so we check for typos... but then we realise that's all fine, so we move to considering the whole premise of the operation instead. And lo, we get a code we can use for a good while longer, until we need it to do something else (I'll leave aside the notion that we don't ever reach some "objective truth", which is too big a topic for today).

In his Last Writings, Shapin says that Kuhn was intent on "showing how progress might occur across paradigms and why the involvement of subjectivity in science could be considered innocuous." That too is nothing problematic. Subjectivity in science is nothing to be squeamish about, simply because it's a) inevitable, it being folly to think you can mitigate all effects of bias completely and b) limited since you can objectively test your findings : you can get a binary yes/no answer. It is a step much too far, but by no means inevitable, to presume that because science isn't perfectly objective and rational that it's also totally corrupt, totally at the whims of the researcher's moods. It isn't anything of the sort.

Finally :

A theory should be accurate and consistent; it should have broad scope; it should simplify accounts of phenomena; and it should be fruitful in disclosing new phenomena or revealing new connections between them. There were, Kuhn acknowledged, substantial problems in applying these values... These were what Kuhn called the maxims, norms or (his preference) values of science, though he felt no need to offer systematic evidence that these values were generally agreed and invoked.

And there's nothing wrong with that either. We just don't need exact, specific instructions as to how fruitful a theory must be, how much more accurate the predictions of one model should be in order to consider it as overturning another. General guidelines are helpful and good. Overly-strict rules, in this case, would be the opposite. I believe it's helpful for scientists to actively think philosophically about what they're doing and why. But there's no need to do this constantly, let alone resort to hurling ashtrays because of methodological differences. That would make for an entertaining conference, though...