Overture: The Man Who Was Never Wrong

Overture: The Man Who Was Never Wrong

Picture the trick you have seen a hundred times and never once questioned.

A stranger walks into a room. Sherlock Holmes glances up — bored, always bored — and in the time it takes you to say good afternoon he has read the man down to his bank balance and his broken engagement. The tan line. The frayed cuff. The particular mud on the left boot but not the right. Each detail a thread; each thread pulled taut; and at the end of them, hanging in the air like a conjuror's dove, the whole secret life of a person who has not yet sat down.

It is the most seductive promise in all of literature: that the world is legible. That if you were only clever enough, attentive enough, cold enough, you could walk through your days reading every soul around you like large clear print. Arthur Conan Doyle made that promise in 1887, and we have never stopped wanting it kept.

Here is the first uncomfortable thing this book will tell you, and I am going to tell you the uncomfortable things early, because you paid for honesty and not for comfort:

It was never deduction.

Holmes says the word constantly — deduction, deduction, he wears it like a monogram, he even titles his little manifesto "The Science of Deduction and Analysis." But deduction is the one thing he almost never does. Deduction is the airtight kind of reasoning, the kind where the conclusion is trapped inside the premises: all men are mortal, Socrates is a man, and so — with no wiggle room, no luck, no escape — Socrates is mortal. Deduction cannot be wrong if you feed it true premises. It also cannot tell you anything genuinely new. It only unpacks what you already smuggled in.

What Holmes actually does is the opposite kind of thing. He sees a tan line and leaps to a returned soldier. He sees mud and leaps to a particular street. That leap has a name — philosophers call it abduction, inference to the best explanation — and its defining feature, the feature Doyle spends sixty stories carefully hiding from you, is that it can be wrong. A tan line could be a tanning bed, a holiday, a skin condition, a lie. The leap is a bet. Holmes is a gambler who has been told, by the only power that matters in his universe, that the house will never let him lose.

That power is Arthur Conan Doyle.

This is the whole magic trick, and once you see it you cannot unsee it. In the real world, a hundred detectives make Holmes's exact leap and eighty of them land in a ditch, and you never hear about those eighty because nobody writes a thrilling story about the inspector who confidently arrested the wrong man over a misread bootprint. In Holmes's world there are no ditches. Doyle builds the crime backwards from the solution, so that every clue Holmes notices was placed there, by God, to mean exactly what Holmes says it means. The tan line always is the soldier. The mud always is that street. Holmes is a marksman in a world where the author paints the bullseye around the arrow after it lands.

I want to be precise about the accusation, because I love this man and I intend to be fair to him even as I take him apart. Doyle's sin is not that he wrote a genius. Doyle's sin is that he wrote a genius who never has to be uncertain — and in doing so he sold three or four generations of readers a fantasy of thinking that, if you ever tried it at home, would quietly wreck your life. It would make you overconfident. It would make you fall in love with your first theory and bend every later fact to fit it. It would make you mistake a vivid story for a true one. We even have laboratory names for each of these injuries, and Part I of this book is essentially a tour of the wards.

And yet.

Here is the second uncomfortable thing, the one that turns this from a hit-piece into a resurrection. Everything Holmes only pretended to have — we can now actually build. Not the flourish. The real thing underneath the flourish.

There is a mathematics of the honest guess; it is called Bayes' theorem, and it does, rigorously, what Holmes only mimed: it tells you exactly how much a clue should move your belief, and — crucially — how much it shouldn't. There is a real science of reading the physical world, forged in the decades after Holmes by men who admired him and then did the hard, unglamorous work he skipped: Locard, who built the first true crime lab in two borrowed attic rooms in Lyon; Galton, who proved a fingerprint was worth something with actual statistics. There are people alive today who out-Holmes Holmes for a living — not with flashes of genius but with humility, base rates, and the radical act of keeping score — and we can measure exactly how good they are, which is a courtesy Holmes never once extended to us. And there are machines, right now, humming in server farms, performing industrialized abduction on a scale that would have stopped the great detective's cold heart: reading your traces, guessing your secrets, and — this is the part that should raise the hair on your arms — occasionally getting an innocent person arrested, because they have inherited Holmes's overconfidence without his author to save them.

So this book has a strange shape. It is going to spend its first act taking your hero apart, screw by screw, until you are slightly annoyed with me for it. And then it is going to spend its second and third acts showing you that the thing you loved him for is real after all — realer than Doyle dared to make it, because the real version can lose, and knows it can lose, and is therefore the only version worth trusting.

The game, as the man said, is afoot. He just had no idea what the game actually was.

Let us begin, as all good autopsies do, with the body.