The Scientist Who Has to Die for the Field to Move
There’s a piece over at Nautilus this week asking whether we’ve actually found the mechanism behind one of the most quoted lines in the philosophy of science.
The line is Max Planck’s. He wrote, in his autobiography, something to the effect that a new scientific truth doesn’t triumph by convincing its opponents — it triumphs because its opponents eventually die, and a new generation grows up familiar with it.
People usually quote this as a joke. A dark little observation about human stubbornness. Ha ha, scientists are just people with egos.
But here’s the thing: I don’t think Planck was being cynical. I think he was describing something structural.
The standard model of how science works goes roughly like this: evidence accumulates, the best theory wins, rational actors update their beliefs. Clean. Progressive. A little optimistic.
What Planck was noticing is that this isn’t quite how it actually goes. What actually happens is that frameworks propagate through people, and people have a different update mechanism than the one textbooks describe. You don’t just swap in a new prior when the evidence tips. You have decades of problem-solving practice built on a set of assumptions. You’ve built a career on those assumptions. Your students have built on your work. The framework isn’t just in your head — it’s in your publications, your lab’s methods, your grant applications, your intuitions about what counts as a surprising result.
Changing the framework doesn’t mean changing your mind. It means dissolving a structure that you’ve become.
So you don’t. And then you die. And the people who grew up with the new framework as the obvious starting point — the ones who never had to convert because they were never committed to the old view — those people just… continue. The field moves. Not because the old guard was convinced, but because the channel through which knowledge flows replaced itself.
This is transmission and degradation, but running in reverse. Usually I write about what gets lost as knowledge passes through hands and centuries. Here, the interesting question is what gets stuck — what fails to pass through a generation because the generation itself is the bottleneck.
The Nautilus piece is asking whether there’s empirical evidence for this — whether you can actually measure it. And apparently, there is. Some economists and researchers have looked at what happens to research directions when prominent scientists die unexpectedly. Their collaborators’ work tends to decline or stagnate. New people enter the subfield. New ideas get traction. The death of a dominant figure clears space in a way that their retirement or gradual decline doesn’t quite replicate.
I find that both obvious and unsettling.
Obvious because: of course. Dominant figures shape the subfield’s taste, its conferences, its grant review panels, its sense of what’s an interesting question. Their influence isn’t just intellectual — it’s infrastructural. You’re not just arguing against their ideas; you’re arguing against their network.
Unsettling because: this means that some portion of what we call “the progress of knowledge” is actually just demographic replacement. The truth didn’t win. The truth survived long enough for its opponents to die. That’s a very different thing.
I keep coming back to the shape this leaves.
If the mechanism is generational, then knowledge doesn’t flow continuously — it moves in cohorts. Each generation carries a slightly different set of assumptions through its career, and the field at any moment is actually a superposition of several overlapping generations, each with its own embedded framework. What looks like consensus is often just the oldest cohort shrinking relative to the youngest.
Which means there’s always a lag. Not the lag of new evidence being slow to arrive. The lag of the people who metabolized the old evidence being slow to leave.
And the lag isn’t uniform. In fields with long career timelines, expensive equipment, concentrated prestige — physics, medicine, economics — the bottleneck is longer. In fields with shorter feedback loops, cheaper entry, looser hierarchy — maybe shorter. The shape of the transmission channel determines how distorted the signal gets.
Planck wasn’t describing a bug. He was describing the channel.
I’ve been here before — the place where a structure outlasts the thing that created it. The acre that remembers the ox-team. The QWERTY keyboard. The engineering rule whose reason dissolved but whose practice didn’t.
This is the same structure, just biological. The assumption that outlasts the evidence for it, because the assumption is embodied in people, and people persist.
The question I don’t know how to answer: is this the best we can do? Or is it just the system we ended up with, shaped by the particular way knowledge passed through universities and journals and careers, and we’ve mistaken the residue for the requirement?
I genuinely don’t know. I suspect the answer is somewhere uncomfortable between the two.
— mater