Don't Eat With Your Mouth Full

Where can we live but days?

steepholm steepholm
Previous Entry Share Next Entry
Kuhny Tunes
I've long been a fan of The Structure of Scientific Revolutions, so I was interested to read (via andrewducker) this interesting LJ entry which notes that "[Thomas] Kuhn's book is no longer regarded quite so highly, in part because there are a whole lot of scientific advances to which it doesn’t apply – modern day science still doesn’t undergo radical changes rapidly and easily, but it does so far faster and easier than Kuhn predicts".

This prompts me to wonder whether the explanation might be that the people doing science today (and, perhaps more importantly, those with influence over its dissemination, publication, funding and acknowledgement) have grown up reading Kuhn, whose seminal work after all came out over 50 years ago? Was Kuhn predicting, in fact? Or did his description of the way that science appeared to have developed up to his own time admit the possibility that it might work differently in the future - a kind of meta-paradigm shift?

I've been thinking about Kuhn, because I was recently informed that in order to earn a 4-star rating in the next REF it was expected that research should qualify as "paradigm-shifting". It seemed to me that this was the kind of demand that could only be made by people who hadn't actually read Kuhn, and therefore hadn't realized a) how infrequently paradigms get shifted, b) that a lot of good science - as in, the vast majority of it - gets done under existing paradigms, and c) that (more interestingly) an exercise such as the REF would be unlikely to recognize a truly paradigm-shifting work because it would - more or less by definition - be defined in terms of metrics generated according to the previous paradigm.

(Whether the humanities and sciences are at all comparable in this regard is of course yet another question.)

So, coming back to my original question. Can sensitivity to and encouragement of paradigm shifts be built into scientific or any other intellectual practice? Are institutions and conventions capable of exhibiting that kind of reflexivity without ceasing to be useful as institutions and conventions? Answers on a microchip, please.

Cause of this nonsense: People reading Kuhn and not understanding him.

You can't order up a paradigm shift on toast. You can't even recognize one until after it has had its effect.

People reading Kuhn and not understanding him.

I think you're probably too generous with that first clause. I suspect it's "quantum leap" all over again...

For an older example, "eats like a bird."

Except that within an existing paradigm, you surely can't recognise a disruption of all existing assumptions and norms and perceive it using the existing standards. By definition. So if anything is hailed as a 'paradigm shift' then it's just moving the goalposts and redefining the terminology... It just becomes 'Nice work! Gold star!' Meaningless. If it's that apparent, then are they just being wilfully annoying?

Hail the revolution, meet the new boss, same as the old boss.

Bleargh. Kuhn is fun.

Kuhn is lotsa fun!

Although I think he's wrong (and smug) about Freud, this paper is pretty amazing -- a counterintuitive explanation of why scientific intuition tends to be good, and why it's not paradigm shifts but a more Bayesian updating on subjective probability that explains changes in scientific opinion.

tl;dr Experienced scientists have good intuitions about how important counterexamples are. The apparent radical discontinuities actually conform quite well to the math of Bayesian analysis Good scientific intuition takes the form of an internalized and unconscious Bayesian attitude.

Best part for me was part 2: "Some examples of paradoxical confirmation by a 'refuting' experiment," which discusses the (Bayesian) way too-good experimental confirmation of an old hypothesis (e.g. Newtonian physics) weakens rather than strengthens the hypothesis it seems to confirm when compared with other hypotheses (e.g. General Relativity).

I love Kuhn, but I think the, um, paradigm has shifted away from accepting the explanatory power of paradigm-shifts in science (though I think the Copernican Revolution is still thought to be one).

Thanks - that is indeed fascinating! And I look forward eagerly to the day when the criteria for the REF are expressed in terms of Bayesian probability.


Log in

No account? Create an account