Years don't turn into judgment

Somewhere in the pitch about your exquisite eye there is a second flattering line, the one about your years. Twenty of them in the trade. All that scar tissue. The assumption underneath is that time does the work on its own, that if you stand in a field long enough the judgment settles on you like dust. It does not. The dust is confidence.

Ask anyone with twenty years how they know what they know and you get the same word back. Experience. Said the way you'd say the name of a saint.

A skill needs an error signal, the plain that-was-right or that-was-wrong, landing close enough behind the action that your brain can pin the correction to the thing that caused it. No signal, no learning, and years are not a signal. They are a lot of actions with the grading left off. And judgment's grade is the hardest one to get. The hire works out or doesn't, two years on, by which time the market turned and the team changed and the brief was rewritten twice. The verdict comes back late, tangled, easy to talk your way out of, which is the natural state of a real decision and a terrible way to learn anything.

So the default of experience is not learning. It is a very good impression of learning, which is worse, because it fools the one doing it. When the outcome lands the ego does with it what it does with taste: keeps the verdict, blames the weather. The wins were your judgment. The losses were the timing, the funding, the guy who left. Twenty years of that does not calibrate you. It laminates the original confidence under two decades of stories, not one of which ever put the verdict at risk. Most of what passes for senior judgment is senior conviction, and it argues back, which a beginner's doubt at least has the grace not to do.

The literature on this is not kind. In 1975 Baruch Fischhoff gave people the background to an obscure 1814 clash between British and Gurkha forces and asked how likely each outcome was. Then he told other groups how it actually ended and asked again. The ones who knew the ending were sure they'd have called it all along. He named the paper "Hindsight Is Not Equal to Foresight." That gap is the thing eating your track record. The instant you learn how a decision turned out, your memory rewrites what you thought before it, so the misses you'd learn most from are the ones you no longer believe you made. You cannot study for a test you're certain you already aced.

The ten-thousand-hours story that got passed around dropped a word, and it was the important one. Anders Ericsson's research never said repetition. It said deliberate practice: specific goals, full attention, immediate feedback on each rep. Take the feedback out and you don't have slow practice, you have a hobby. In 2014 a meta-analysis by Brooke Macnamara and colleagues measured how much of expert performance practice explains, and the number falls through the floor as the domain gets murkier. Games, about a quarter. Music, a fifth. Sports, a bit less. Education, four percent. Professions, under one. A profession stops handing you the clean feedback that makes practice practice, so the hours stop buying anything. The domains where judgment matters most are the ones where standing there longest teaches you least.

Then look at the people it does teach. Weather forecasters are almost perfectly calibrated, which sounds like a joke and isn't. When a good one says seventy percent chance of rain, it rains about seventy percent of the time, a reliability Allan Murphy and Robert Winkler documented in the seventies and it has held. Nothing to do with meteorologists being wiser than the rest of us. The job hands them the loop for free: every forecast is dated and numeric, and tomorrow either rains or it doesn't, in public, on the record, tomorrow. No talking your way out of it. The forecaster gets the one thing the strategist and the investor and the person making the hire almost never get, a fast, honest, unspinnable grade, over and over, for years. The forecaster practises. The rest of us accumulate.

I did this to my own writing for years. I judged a piece the way everyone judges taste: if it looked right, it was right. I never once wrote down, before I shipped, what "working" would even mean, and I never went back to score the thing against it. The verdict was a feeling I kept and never let anything touch. The first time I forced a real grade onto it, the gap between what I was sure of and what held up was not small.

So the fix is not more time. You cannot make your domain as regular as the sky, but you can build the loop the sky hands the forecaster for free and bolt it onto decisions that don't come with one. That built loop is the difference between doing a thing for twenty years and practising it for two. Five moves.

Log the call before the outcome. Before you act, one line: what you're deciding, what you predict, why. This is the move that beats Fischhoff, that pins down what you actually believed before the result gets to revise it. Written after, it's fiction. The machine will hold the log and stamp the time. It will not make you write the line while you're still unsure, which is the only moment the line is worth anything.

Make the prediction gradeable. Not "this'll go well." A number, a date, and the threshold that proves you wrong: ships by the fifteenth, clears two hundred signups in week one, under that I misjudged demand. A call that can't be graded can't teach. This is the step that turns a mood into a bet.

Shorten the loop until it stings. Build the smallest real test that returns a verdict fast, put the thing in front of actual people, watch what the system does instead of what the room says. A correction that lands two years late attaches to nothing; the you who decided is gone. Small reversible tests beat one big irreversible bet.

Grade on a calendar, cold. Once a month, not when the feeling takes you, because the feeling only takes you after a win. Reopen the log, mark each closed call right, wrong, or unresolved, against what you wrote. Say the buried ones out loud. The machine will dig up the bets you quietly filed away and tally your real hit rate. It cannot supply the nerve to open a losing month and look at it.

Update the model, out loud. For every miss, one sentence: I was wrong because I believed X, the correction is Y. That sentence is the rep. Skip it and you skipped the only part where the muscle loads.

So the last question is worse than it sounds. Which of your decisions this year ever actually got graded, and of those, how many did you write down before the outcome and then change your mind after? If the honest answer is none, you do not have twenty years of practice. You have one year, run twenty times, each pass sealing the last a little tighter. The fix was never more experience. It was the part you kept skipping, writing it down before you knew.