When I was a youth, getting better at rock climbing but still a universe away in ability from the superstars, I’d sometimes sit in the café in the Derbyshire village of Stoney Middleton, then one of the centres of hard limestone climbing in the UK. I’d have a chip butty and a pint of tea, and enjoy overhearing the conversation of those same superstars. Then, when they’d gone back to the rock, I’d take a look the New Routes Book, a tattered hard-back exercise book in which those stars would record their hardest new ascents, to be commented on by their friends and rivals. But every now and again, a hapless interloper (probably a visitor from the South) would have recorded the ascent of a claimed new line, usually at a modest grade. Across the proud paragraph, one of the local superstars would have scrawled in aggressive capitals, “DONE YEARS AGO”.
This illustrates the danger of any field of human activity which is in some sense progressive, with new achievements building on the old ones. As the field gets mature, it’s more and more difficult to have a sense of what’s genuinely new, and what’s merely retreading the old. Science is no exception.
I remember as a young lecturer, enjoying the catty diatribes of the great theoretical physicist Sam Edwards, about other polymer scientists who weren’t as well grounded in the foundations of physics as he was. “What is it with Rouse, and his ‘beads and springs’. Didn’t he realise that his ‘Rouse modes’ are trivially the normal modes that drop out when you write down the Rayleighian for a chain? And are those Dutch people still hawking their tired old lattice model? They’ve just rediscovered the Hartree method and badly discretised it”. (No, intellectual one-up-man-ship is not entirely absent from the world of academia.). And when the excitement about nano-materials and their novel mechanical properties was at its height, there were moments when I’d think to myself, “Yes, that’s great, but you’d know that already if you’d read ‘The New Science of Strong Materials’” (J.E. Gordon’s brilliant 1968 popular science book).
Maybe Sam’s influence has left me hypersensitive to this, but I’ve certainly had dark nights when I’ve wondered whether the whole of my field was essentially already implicit in the combined output of Van der Waals, Lord Rayleigh and Boris Derjaguin. One wonders how much of science is essentially rediscovering what was already known – or, to put it more positively, how much progress is possible by revisiting and recombining the existing, but largely forgotten, corpus of prior knowledge. Perhaps this will be the major contribution of large language models to the scientific enterprise.
Entirely in keeping with its theme, this post re-iterates an argument already recently made by Dan Davies in his substack.