In praise of scientific failure

Would we still admire Nobel-winners if their brilliant hypotheses had happened to hit dead ends? 

October 19, 2023
Katalin Karikó won the Nobel Prize for Medicine in October 2023. Image: Associated Press / Alamy Stock Photo
Katalin Karikó won the Nobel Prize for Medicine in October 2023. Image: Associated Press / Alamy Stock Photo

I once tried to sell a book idea about historic scientific failure. I don’t know what I was thinking really; they couldn’t turn me down fast enough: “So this is science… and there’s no discovery?” Science, you see, finds out exciting things. 

Like Katalin Karikó did. Earlier this month, she won a Nobel prize for work that led directly to the development of the mRNA vaccines for Covid-19, helping to save countless lives. Naturally, now she’s winning prizes, we’re hungry for Katalin’s story. It’s the happiest of endings. 

But we need to think about the beginnings.

Karikó’s research story began in earnest with emigration to the US from Hungary with $900 smuggled in a teddy bear, then feeling forced to take a job far from the family home, research that didn’t seem to be going anywhere, and ideas that left others cold. “Every night I was working: grant, grant, grant,” she once said. “And it came back always ‘no, no, no’.” To others, nothing seemed to say dead end like her research. She was demoted, her pay was cut, and she was declared “not of faculty quality.” She somehow struggled on in the margins for most of her career.  

Today, this makes terrific copy. It has it all: triumph over adversity, vision, self-sacrifice and self-belief, and science as breakthrough. What an inspiration! What a narrative arc! And all true. Plus, it invites a satisfying shake of the head at the fools who doubted her.

But be honest, isn’t Karikó only a story at all because of how it ended? Otherwise, for all her tenacity, she’d be forgotten—just another struggling, failing, nobody. 

As in other walks of life, for better or worse success changes everything. We like failure only if it ends somewhere good, or don’t bore me. And while it’s easy to see that this attitude could have a timing problem—since the whole point of research is you don’t know how it ends—in practice I think most of us would have shared it and given up on our heroine well before the final act for the failure she appeared to be, like many of her contemporaries seemed to.

So what’s my argument, that we should rethink failure? To some extent, yes, and success too. In science, success can be an inspiration, no doubt, but it can also be systemic bad karma, contributing to a credibility crisis in research. Researchers need their work published to get on or even just to earn a living, and a “successful”, publishable research outcome tends to be one that says: “hey, look, we discovered an effect”; less so one that says, “nope, nothing doing here”.’ Academic journals are little different to pop science in this respect. But the upshot is too many people straining science too hard for the right kind of ending, and many of these positive findings turn out on re-examination to be no such thing, despite peer review. William Kaelin, a winner of the Lasker prize for medical science, once doubted that he would be published today because of what he called “claims inflation”, and professional incentives that generate “grand mansions of straw, not houses of brick.” Focusing on “success” can beget bad science.

This means we’re less able to trust the scientific record, full as it is of improbable quantities of success and too little failure, which misleads the next rounds of research. It’s an unusual case of Goodhart’s law: when a measure becomes a target, it becomes a bad measure. 

But if success isn’t always what it seems, neither is failure. A CV full of honest failure, done well, isn’t necessarily a bad bet—as Karikó demonstrates. She learnt from every negative result, she says. Little by little, she understood more. 

The sharp end of this problem is funding: how do we place our bets among the many claims for attention when we lack reliable evidence about how the stories end? Given that hindsight isn’t an option, and foresight can’t reliably read the runes, one option is a lottery. Following successful trials in New Zealand, Austria and Switzerland, the British Academy has partially randomised its small research grant allocation, calling this “fairer and more transparent” and “less subjective”. Applicants must meet a quality standard, but after that it’s the luck of the draw. There’s also talk of journals committing to publish results whatever they are, based on the quality of the proposal, not its outcome.

Karikó seems reluctant to blame her doubters; rather, she wants better institutional design. But there are no perfect answers. Maybe the best reason to say she should have had more support is not that she was a winner; it’s that science is a collective endeavour in which endings like hers are possible if enough people do the job honestly, asking good questions, using good methods (harder than it sounds)—ideally without perverse incentives from a culture of success-or-bust—and she looked like one of those people. There are doubtless more inglorious Karikós out there who put in the same work with little to show for it. They’re part of this story, too. It’s just that few want to hear that bit. That was her problem then; it’s ours still. 

So when we cheer for her now, it’s partly because her resilience was stronger than our instincts, and stronger than our institutions. Which is another way of saying that with Katalin Karikó, science and the whole world got very, very lucky.