Perfectly Confident

If the errors are so varied, why do they persist? Part of the answer, Moore argues, is that we get almost no useful feedback on our confidence, and the feedback we do get arrives too late and too tangled to learn from. You feel sure a deal will close; it closes; you conclude you have good instincts. But the deal might have closed for reasons that had nothing to do with your read, and you'll never run the counterfactual. Success and failure are noisy teachers.

Then there's the way we reach for round, comforting numbers. Ask people how confident they are and they'll say ninety percent, ninety-nine percent, a hundred — figures that sound decisive but are almost never earned. Moore is fond of pushing on that last one: if you're truly a hundred percent sure, you should be willing to accept any wager against being wrong, and almost nobody actually is. The certainty is a feeling dressed up as a measurement. The number was never checked against anything.

The heart of the book is the repair kit, and Moore's tools share a family resemblance: each one drags a private feeling out into the open where it can be tested. The first is the wager. Whenever you catch yourself feeling sure, imagine betting real money on it. Would you take four-to-one odds that the project ships on time? Ten-to-one? The moment you have to price your belief, the mushy "pretty confident" resolves into an actual number, and the number is usually humbler than the feeling. Betting is a lie detector for certainty.

The second tool is the confidence interval instead of the point estimate. Rather than guessing that the meeting will draw forty people, guess a range wide enough that you're ninety percent sure the truth falls inside it. Most of us set those ranges far too narrow — that's overprecision showing up in the wild — and the discipline of widening them until they honestly hold ninety percent of cases is, Moore argues, one of the fastest ways to calibrate. It forces us to own our uncertainty rather than paper over it.

Step back and the argument is quietly radical. The culture treats confidence as a character trait — something you either radiate or lack, something to project in interviews and manufacture before a big talk. Moore treats it as an estimate. And once confidence is an estimate rather than an identity, the whole moral weather around it changes. Being wrong about your odds isn't a failure of nerve or a personality flaw. It's a measurement error, and measurement errors can be corrected without anyone feeling small.

This reframing dissolves a stale debate. We tend to argue about whether it's better to be bold or humble, as if those were rival temperaments and we had to pick a team. Moore's answer is that both are wrong when they detach from evidence. Overconfidence and underconfidence are the same disease — miscalibration — pointing in opposite directions. The virtue isn't boldness and it isn't humility. It's accuracy, which sometimes asks you to press harder and sometimes asks you to wait, and doesn't much care which feels better.

We started with the familiar pep talk — trust yourself, back yourself, feel the certainty and act on it — and Moore spends the book quietly taking it apart. Not because self-belief is worthless, but because confidence turned out to be three different things, each capable of running too hot or too cold, and none of them improved by simply willing more of it. The fix he offers is almost unglamorous: bet on your beliefs, widen your ranges, hunt for the reasons you're wrong, and keep score. Calibration over conviction.