I'm taking drum lessons at 44 from the instructor I wish I had at 14. His name is Chuck James, and if you've spent any time around live music in Honolulu, you already know it. When I was 14, I had a different instructor — a working professional, at the time a musician with the Royal Hawaiian Band, which meant he could play anything you put in front of him. He just couldn't tell me why my hands wouldn't do what his did. He wasn't a bad man. He was a category error. We'd hired a swimmer to teach swimming, and nobody had asked whether he'd ever had to learn.

That's not an insult. That's the whole problem, and it took me thirty years and a flight instructor certificate to understand it.

Before I ran an IT company, I taught people to fly airplanes. The FAA has a rule about that which sounds bureaucratic and is actually profound: being able to fly doesn't qualify you to teach it. The commercial certificate proves you can do the thing. The flight instructor certificate proves something entirely different — that you know how the thing goes wrong. For every maneuver in the training syllabus, an instructor candidate has to know the common errors: where students fixate, what they do too early, what they do too late, what the wrong picture looks like from the other seat. The teaching credential isn't a higher grade of the doing credential. It's a catalog of failure modes.

The maneuver where this clicked for me was steep turns. Roll into a forty-five degree bank, hold your altitude, roll out on the heading you came from. Simple to describe, and every student makes the same mistake: they stare at the altimeter. It's right there. It's precise. It's authoritative. And it's the error — because the altimeter tells you what already happened, and a steep turn is flown by what's happening now. You hold the maneuver by pinning the nose to the horizon and feeling the airplane through your body. It's the one maneuver in primary training where staring at the instrument is itself the failure mode. I didn't learn to teach that by being good at steep turns. I learned it from a syllabus that had watched ten thousand students stare at the same gauge.

Doing and teaching are different bodies of knowledge. Almost everyone hires for the first when they need the second.

I'm relearning golf at 44, and my instructor went to school to learn how to teach golf — not to play it, to teach it. He's a good golfer. He's a better teacher. When my ball goes sideways, he doesn't show me his swing. He tells me the handful of reasons a ball goes sideways and which one is mine today. That's the failure-mode catalog again, wearing a polo shirt. Somebody handed him the errors so he wouldn't have to guess at them from his own talent — because his own talent is exactly the thing that can't see them.

Last year I wrote a book called Tuesday Person, and some of the people in it got interviewed by the author across the dinner table. My wife runs a household the way some people breathe — the schedule, the logistics, the quiet arrival of everything in its place. I once asked her how she's so good at it. She said, "Good at what?"

That's a frog. Not refusing to teach — there's nothing in there to hand over. The skill was never composed, so it can't be decomposed. She isn't being modest. The question doesn't parse. You cannot learn to swim from a frog, because the frog has never once thought about swimming.

But there's a more dangerous animal than the natural, and I'm it.

I taught myself this trade. Twenty-five years, no journeyman above me, nobody to watch. Everything I know about running networks and running a company I figured out alone, usually at night, usually the hard way. Which means I'm not a frog exactly — I did have to learn to swim. But I learned in a pond nobody else was in, and every stroke I know is a workaround that made sense at the time. The natural frog gives the apprentice nothing. The self-taught frog gives them the wrong things, with authority. My way works. My way is also 1999's way, refined in isolation, full of scar tissue presented as technique. And in this trade — in most trades — that is the standard failure mode of training: the senior guy teaches the way he does it, the way he does it is the way he survived it, and the apprentice inherits the scars without the wounds that explained them.

So here's what I tell my technicians now, and it's the closest thing I have to a teaching philosophy: here's the bones of the issue. Don't document how I would do it. Ask Claude the right way to do it.

Apprenticeship was always watch one, do one. That transfers procedure, and procedure matters — the journeyman is genuinely good at handing it over. What never had a source was the why. The theory underneath. The reason it's this order and not that one, and what failure looks like if you get it backwards. That layer traditionally came from formal education nobody in this trade has, or from a master trained in pedagogy, which nobody in this trade gets. So it just didn't come, for generations, and the trade reproduced procedure without theory.

A tool like Claude sits in exactly that gap, and it sits there for a specific reason: it is pedagogy without practice. It has never closed a ticket — the way a syllabus has never flown an airplane. But it's trained on the written record of this trade, and technical writing is unusually dense in failure modes, because that's what technical writing is for. The postmortems. The vendor documentation. The accumulated arguments about the right way. The journeyman supplies the doing. The tool supplies the why. The apprentice assembles both into something neither source has alone.

Watch one. Do one. Ask why one.

One caution, and it's going to be a through-line in everything we write about these tools: how you ask is the whole game. "Ask Claude the right way to do it" is not the same instruction as "ask Claude whether my way is right." The second question gets a yes. It always gets a yes — these tools amplify the frame you bring them, and a question shaped like a request for approval returns approval. Part of learning from the machine is teaching it how to teach you, and you do that with the shape of the question. Not "will this work?" — plenty of wrong things work. "Is this the right way to do it?" The first question defends your draft. The second one exposes it.

My wife can't tell you not to learn from her. She doesn't know there's anything to not-learn. That's the one advantage the self-taught frog has over the natural: he knows what he is. The most useful sentence a twenty-five-year veteran can say to an apprentice isn't here's how I do it. It's don't do it my way.

That's the only teaching move a frog gets.

I'm making it.