Science

If They Say They Know, They Don’t Know

A principle for understanding which experts to trust, including the CDC.

Exterior of the Centers for Disease Control and Prevention’s Edward R. Roybal campus
The Kiesenhofer principle says you can trust them. Tami Chappell/Reuters

The other day I came across a piece of profound wisdom in Cycling News:

Now I’m old, I’m 30, and I started to realise that all those people who say they know, they actually don’t know. Many of them don’t know, and especially those who say that they know, don’t know, because those who do know say that they don’t know.

The sage speaking here is Anna Kiesenhofer, the Austrian mathematician specializing in partial differential equations and symplectic geometry who stunned world cycling by winning the gold in the Olympic road race on Sunday. She jumped out to a huge early lead, well out of sight of the peloton; apparently Kiesenhofer was considered such a noncontender that the higher-ranked riders didn’t notice she wasn’t with them in the group. When Dutch rider Annemiek van Vleuten crossed the line at the front of the pack, she thought she’d come in first—but Kiesenhofer had finished more than a minute earlier.

Advertisement
Advertisement
Advertisement
Advertisement

That is not the way you’re supposed to win an Olympic race. But as the self-coached Kiesenhofer says, she doesn’t always do things the way the experts say you ought to.

Nondeference to experts is the spirit of the moment. Who are those pinheads, those elites, those people who say they know, to demand our obedience? This commitment to irreverent questioning is part of the spirit of mathematics, too. Math gives us tools and wherewithal to pick apart an argument, to look at a confidently expressed conclusion and see the unstated and unexamined premises it’s leaning on. (Even the style of Kiesenhofer’s slogan recalls math: “Those who do know say that they don’t know” is, in logic parlance, the contrapositive of the mathematically equivalent “those who say that they know, don’t know.”)

Advertisement

The problem is too many would-be skeptics adopt the first part of Kiesenhofer’s insight and ignore the crucial final clause. Those who do know say that they don’t know. If you pooh-pooh the credentialed authorities, but then hop into obedience to someone else who says they know, you’re totally missing the point of the Kiesenhofer principle.

Advertisement

I was thinking about this when I read a series of heated tweets from former Food and Drug Administration Commissioner Scott Gottlieb. He was mad about how the Centers for Disease Control and Prevention was trying to model the course of the COVID pandemic now that the delta variant has become dominant in the viral population. The CDC reported on models from a wide range of labs that incorporated a wide range of built-in assumptions, which means they wound up with a widely divergent spray of predictions, some foreseeing a rapid decay of the pandemic, others a full-on exploding fourth wave.

Advertisement
Advertisement

Not good enough, says Gottlieb! “The huge variance in the estimates shows CDC doesn’t know how to model this wave, and has little practical idea whether we’re at beginning, middle, or end … For the week ending August 14, CDC estimates there will be either an average of 10K infections a day, or more than 100K. Either the infection wave will be largely subsiding, or will be raging out of control. The CDC isn’t sure.” He calls the CDC’s information “deeply disappointing and not actionable.”

What Gottlieb is complaining about is that the CDC is telling the truth. They are, in Kiesenhofer’s formulation, the people who say they don’t know. In other words, they’re the people you can trust. Gottlieb complains that the CDC or some alternative agency ought to be “action-oriented.” Fair enough—we all like action. But it’s not like the CDC doesn’t give advice!

Advertisement
Advertisement

And if you want good advice about what to do, Kiesenhofer would tell you, listen to the person who tells you they don’t know what’s going to happen if you take their advice.  Because that person is telling you the truth.

I don’t want to make it sound like we live in an utterly opaque fog. We know some things reasonably well. We know vaccines prevent most symptomatic COVID and we know almost certainly that the death tolls from future COVID waves in heavily vaccinated populations are going to be much lower than what we’ve seen before. But we definitely don’t know whether 10,000 or 100,000 Americans a day are going to be getting COVID next month.

Advertisement

Mathematical models of disease aren’t—can’t be!—fortune-telling oracles. Math can’t tell you exactly what’s going to happen in the future, because the dynamics of a pandemic aren’t math—they’re math plus people’s reactions to math. (We can already see this happening: Vaccination rates seem to be rising again in the U.S. after months of decline, maybe in response to new mandates, maybe in response to news about the surge of delta cases, maybe because of some other aggregate whim. Whatever the motivation, this affects what happens next.)

Advertisement
Advertisement

People want to know what’s going to happen next and they want to know what to do right now. That’s natural. And they look to scientists for answers to those questions, either because they want good advice or because they already have a course of action in mind and want to put a “FOLLOW THE SCIENCE DUMMY” sticker on it.

And so we’re attracted to the sound of uncomplicated confidence.  But it’s an attraction to be resisted.  Take it from the mathematician who rides a bike really fast.  There are the people who say they know and there are the people who know. When the stakes are high, which do you want to listen to?

Advertisement