The Style Isn't the Skill

Chris Hillman — Mon, 31 Aug 2026 09:00:00 +1000

I was a kid when I first watched Bruce Lee, soon after and my friend signed up for kung fu lessons.

What hooked me wasn’t the fighting. It was the speed at which Bruce moved. The more you watched, though, the more you noticed something sitting behind the speed. He was obviously smart, fast, cool, and clearly thinking several moves ahead of everyone else in the room. Even as a kid you could tell the fighting was the visible bit of something much bigger.

Looking back, the thing I admired was the confidence. Not swagger. This was different. It was the kind of confidence that comes off someone who has poured a frightening amount of themselves into one craft. You can see it on people in all areas of life. I’m sure you’ve seen it in your industry. It could be a senior data engineer walking a junior through a query plan without looking at the screen. A modeller who hears the grain of a new fact table and already knows which join is going to hurt. It radiates, and it can’t be faked for long.

Then people will say…

Yeah, but he’s Bruce Lee.

You’ve probably said the same about someone like him. It sounds like humility, but it isn’t. It’s a permission slip — if he’s a different species, then the work he did to become that good is not available to me, and I’m excused from trying.

The problem with this is that it never was easy for Bruce Lee, and I’m sure it’s the same for others as well. The three moments that made Bruce Lee happened before anyone outside a few gyms and a cancelled TV show knew who he was. He wasn’t Bruce Lee yet. He was a Wing Chun kid from Hong Kong teaching out of a friend’s house in Oakland, then a sidekick on a show that lasted one season, then a man in a back brace. Enter the Dragon came out weeks after he died. Everything that matters here is from before.

1964: the messy win

In late 1964 Lee fought Wong Jack Man in Oakland, behind closed doors, in front of a handful of people. There are disagreements as to how it went. Lee’s side says he finished it in about three minutes. Wong’s side says it dragged past twenty and ended with both men exhausted. What everyone agrees on is that it was ugly, and that Lee came out of it winded and rattled. For a man who had spent the previous year telling anyone who’d listen that his technique was superior to the “classical” styles, it was a humiliating result. He’d won, by his own account. It didn’t feel like winning.

What he did next is the interesting part. He didn’t defend the technique. He tore it up, and hit the reset switch. That fight is the generally accepted origin of Jeet Kune Do, and the first thing he changed wasn’t his kung fu. It was his conditioning. He went looking for far more extreme fitness training than any traditional school would have asked of him.

The line he used to explain the whole shift:

“The proper method is the one that works.”

As with most things that Bruce Lee said, there is so much in those few words. It’s easy to read it as “method doesn’t matter,” and that’s the opposite of what it says.

It says there is a proper method. It says the test of whether a method is proper is a single question: does it work? And it says, quietly, that the answer can change. A method that worked last year against last year’s opponent gets re-tested, and if it fails, it isn’t proper anymore. Loyalty to it is not a virtue.

He made the point a second time, in a way only Bruce Lee would. He put a mock tombstone in his studio. The inscription:

“In memory of a once fluid man, crammed and distorted by the classical mess.”

He wasn’t mourning kung fu. He was mourning himself. The “classical mess” was the ceremony of the art mistaken for the art: the numbered forms, the fixed stances, the correct sequence of steps that everyone agreed was correct because everyone had always agreed. He’d drilled it for years. And in a real fight, against an opponent who didn’t know the choreography, it had crammed and distorted him.

On one side of the spectrum there might be a template or process that was set up. Does it work? Or is the only answer “it’s how we’ve always done it”? Sometimes we don’t spend the time reflecting — it might have worked perfectly initially, but is it as effective now? There’s also the other side, where we make exceptions to the process far too often, to the point it becomes a mess and we no longer know what the technique even is.

Bruce was ruthless about fixed form. He was obsessive about fundamentals. His response to a bad fight was more discipline, not less. The freedom he’s famous for was built on top of conditioning most people wouldn’t tolerate.

The style isn’t the skill. That’s the whole thing.

One kick, ten thousand times

One of Bruce’s quotes was:

“I fear not the man who has practiced ten thousand kicks once, but I fear the man who has practiced one kick ten thousand times.”

Ten thousand kicks once is a career spent collecting — a new tool every quarter, every certification going, every architecture pattern from every conference talk. You can talk about all of them. You’ve done each of them once, in a tutorial, on a dataset that was designed to make the tutorial work.

One kick ten thousand times is mastery of the skill. It isn’t knowing the technique. It’s having met every way the technique fails. I see this a lot meeting people with various titles: have they actually dealt with quality incidents, and how did they protect the data from it? Have they worked with a source system that offers deltas, and tested them thoroughly? The engineer who’s built them fifty times knows what happens when the source system back-dates a change, when two updates land in the same batch with the same timestamp, when the current flag and the high date disagree, and when a backfill quietly rewrites six months of history because someone ran it day by day. (I wrote a whole post about that last one, because it took me an embarrassing number of reps to see it coming.)

So pick the kick. Dimensional modelling. Streaming ingestion. Query performance on the warehouse you actually run. Pipeline testing. Not all of them. One, then maybe a second. The industry will keep offering you ten thousand shiny new ones, and most of them will be gone in three years. However, honing these skills is what shines out in interviews I conduct — I can see the difference between someone who knows how to kick, versus someone who has done that kick thousands of times.

Absorb, discard, add

The quote that describes how Lee built Jeet Kune Do out of Wing Chun, Western boxing, fencing, and whatever else he could get his hands on:

“Absorb what is useful, discard what is not, add what is uniquely your own.”

Absorb what is useful. Not “learn everything.” Absorb implies digestion. You take the thing in and it becomes part of how you move, without needing to consult it. Kimball’s grain-first modelling is worth absorbing. So are idempotent loads, and the habit of writing the test before the transformation. Once absorbed, you stop noticing you’re doing them.

Discard what is not. Harder than it sounds, because discarding means admitting you spent time on something that didn’t pay off. The pattern that made sense at your last company and doesn’t fit this one. The tool you got certified in that your current platform doesn’t run. Lee walked away from the style his own master taught him. Most of us can’t walk away from a Confluence page.

Add what is uniquely your own. What have you built that isn’t in any documentation? The reconciliation check you wrote because the standard tests kept missing one specific failure. The way you explain a lineage problem to finance so they stop asking the same question every month. That’s yours. It’s the part nobody can hire off the shelf. If your self-evaluation lists things you absorbed and nothing you added, you’ve described a good student, not a good engineer.

Water and the shape of the cup

“You put water in a cup, it becomes the cup.”

Water has no form of its own; it takes the form of whatever holds it. Bruce also points out that water can flow or it can crash. Formless isn’t the same as weak.

For a data modeller this is close to a job description. The model takes the shape of the question, not the shape of the framework diagram. A star schema is right until the business hands you a many-to-many relationship that every consumer needs, and then it’s wrong, and the modeller who is loyal to the star instead of the question will build something the analysts route around within a month. I’ve argued before that there’s no single right way to do data engineering, only the right way for the world you’re in. Lee got there fifty years earlier with the motto he put on the Jeet Kune Do emblem:

“Using no way as way, having no limitation as limitation.”

No way as way doesn’t mean no method. It means the method is chosen fresh for each fight. Kimball for the reporting layer, something flatter and wider for the feature store, streaming where latency is the product and batch everywhere else, with no shame in any of it. The limitation you refuse to accept is the one that says “we’re a Kimball shop” or “we don’t do streaming here” before anyone has looked at the problem.

1969: the letter

January 1969. Lee is twenty-eight. The Green Hornet had come and gone after one season. He’s picking up bit parts and teaching private lessons to actors to pay the bills. Linda is pregnant with their second child. Bruce’s Hollywood career has stalled.

He sits down and writes himself a letter. It’s on a sheet of notepaper headed “Secret,” in red and blue ink, and he titles it “My Definite Chief Aim.” The whole thing is four sentences. In the first he declares that he will be the first highest paid Oriental superstar in the United States (his words, and the era’s). The last two set a dated target and a personal one: world fame from 1970, ten million dollars by the end of 1980, and a life lived the way he pleases.

The sentence in the middle is the one that matters most:

“In return I will give the most exciting performances…”

This isn’t a wish list. It’s a contract — here’s what I will be, here’s what I will give in return.

I’ve read enough self-evaluations to notice a pattern. Most of them are the first sentence without the second. “I want to grow into a senior role.” “I’d like more exposure to the streaming work.” “I’m keen to lead a project next year.” All fine. All wishes. The “in return” clause is missing, and without it there’s nothing for a manager to hold onto and nothing for you to hold yourself to.

Try writing it Lee’s way. Something like: By the end of June I will be the person this team hands streaming ingestion to without a second thought. In return I will have built late-arrival handling on both event pipelines, documented the watermark strategy so someone else can run it, and carried the on-call for it the first three times it breaks.

That’s a sentence a manager can act on. It’s also a sentence that scares you slightly to write, which is how you know it’s the real one. (He achieved everything in the letter, incidentally, and then died at thirty-two.)

In the review room

When I sit with someone to go through their self-evaluation, I’ve started asking myself a slightly ridiculous question: what would Bruce Lee do with this? Ridiculous or not, it changes what I look for.

If you’re writing the evaluation.

Write the “in return” clause. Every aim gets a consideration.

Name your kick. What’s the one thing you’ve done enough times that you know how it fails? If you can’t name one, that’s the finding, and it’s a useful one.

Include something you added, not just something you absorbed. The thing you built that wasn’t in the docs.

If you’re reading it.

Judge what works, not adherence to form. The engineer who follows every ceremony and ships fragile pipelines has the classical mess. Say so, kindly and specifically.

But check the reps. “Judge outcomes” drifts very easily into “reward whoever got lucky.” Someone who has practised one kick ten thousand times has a track record of predictable outcomes, and that’s the difference between skill and a good quarter.

Look for the gap between knowing and doing. Lee was fond of a line he borrowed from Goethe: knowing is not enough, we must apply; willing is not enough, we must do. Most people in a review know exactly what they should be doing. The evaluation is a measure of the distance between that and what happened.

If you want a structured way to run that kind of exercise with your own team, I’ve written before about the Reflective Best Self process.

The tradeoff

Judging by what works is harder than judging by ceremony. Ceremony is easy to audit: the template was filled in, the points were estimated, the meeting was attended. Skill is hard to audit, because you have to understand the work well enough to know whether the outcome was earned or lucky, and that takes time managers don’t have and technical depth some of them stopped maintaining years ago. Take the Lee approach and you sign up for that cost.

There’s a cost on the other side too. A self-evaluation written without the mask, with a real “in return” clause and a named weakness, is a sharper tool for your own growth and a sharper tool in the wrong hands. In a healthy team it gets you coached. In an unhealthy one it gets quoted back at you.

Self Evaluation on Ghost in the data