When was the last time you were handed a problem with no instruction manual? Examples: A new job in which nobody explains how things work; your own body when it does something unwelcome.
My nominee for an acute problem: A baby who will not stop crying. How did you solve it?
I bet it was not by demanding proof. You tried and watched. The rocking helped a little; the singing helped a lot. You kept what seemed to work. If you had insisted on certainty before you acted, the baby would still be crying.
A messy, humble, incremental way of learning like that is not a failure. It is how learning works. But a large part of the scientific establishment has spent a century pretending otherwise.
The verdict habit
Most research is conducted this way: You assume that nothing is happening. The new treatment does nothing. No outside assumptions allowed; they will compromise the data. You assume you know nothing.
You gather data and ask: Is this result enough to force me to abandon that assumption if I assume nothing but chance is operating? If yes, the finding is declared real. If no, it goes in the bin.
This may be courtroom mentality. The defendant is presumed innocent, and the study is the trial. The verdict is binary and final. Every trial starts with the jury's memory wiped. What was learned last time doesn't count. The 40 studies that came before don't prove a thing; a treatment that helped some people in some circumstances gets filed away as ineffective.
Finding the submarine
In May 1968, the American nuclear submarine Scorpion vanished in the Atlantic with 99 people aboard.
The government handed the problem to U.S. Navy chief scientist John Craven. He had one search ship. He did what you do with the crying baby, only more carefully.
He divided the ocean into squares and numbered each square. He asked veteran submarine commanders to guess which number contained the sub. Where might it have gone down, and why? He let them bet on their scenarios, sometimes even with bottles of scotch.
Every time the ship swept a square and found nothing, that empty square was not a failure. It lowered the chance for that square and raised it for the others. The map moved and the votes changed. He searched again where the map now said to look.
In late October, the ship found the submarine about 460 miles southwest of the Azores.
The same idea worked on the Air France 447 wreck, lost between Rio and Paris in 2009. Two years of searching turned up nothing. A team of mathematicians took those failures, treated every empty pass as a clue, not a dead end. They redrew the map. The next search found the plane in a week.
These searchers knew the submarine was somewhere. They did not throw away a search because it was empty. They started with their best guess, let reality push the guess around, and got a little less wrong every day.
Just a little bit of science
Statisticians have a long and sometimes bitter argument about whether science should work like the courtroom or the search. I won't drag you into it or use geeky words; that is exactly what has kept people from realizing they already know the answer.
The search says: Knowledge is a map of how confident we are in our guesses; old evidence stays on the map and new evidence moves it. A study that finds nothing still tells you something. Disagreement is allowed, and two people who disagree can still share their findings.
The verdict produces winners and losers. An idea gets convicted or acquitted, and a field spends decades relitigating instead of learning.
We are not clones. What helps your neighbor may not help you, and what helped you last year may not help you now. A verdict about the average person cannot tell you that. A map that you keep updating for this person, in this life, can. The way you figured out your baby is the way a good clinician figures out a client. If we are wise, we will make progress on the rest.
Where this lands
We are living through a period of large, moving, and unfamiliar challenges—climate, fractured communities, and changing technologies. Nobody has the manual. Waiting for proof before we act is a recipe for standing still while the water rises. Pretending to be certain when we aren’t is like driving off a cliff with confidence.
The alternative is humbler. Make your best guess. Say it out loud. Act on it, and let what happens change the guess. Treat the things that did not work as clues. Pool all the maps.
That is how you soothe a crying baby and find a lost submarine. In one square of ocean at a time, we might find our way through what we are facing. You already know how to do this; you've been doing it your whole life. The only truly new step is to stop apologizing for it.