The Wrong Unit

The "10x engineer" is a familiar idea: some people produce ten times the output of an average peer. They ship faster, break fewer things, see around corners, lasso the moon on occasion¹. The concept is compelling enough that it's become shorthand for a certain kind of hiring ambition: find the 10x people, and profit somehow. Everyone wins through the power of magical thinking.

It's not a bad observation. Some people are genuinely and dramatically more productive than others. The problem isn't that the 10x engineer doesn't exist: the problem is that individual output is the wrong unit to optimize for. Hire 10x engineers if you can - they're occasionally wonderful to work with - but what you want is more likely the exponential engineer, someone who makes working better, not someone who works better.

What You're Actually Measuring

When someone produces ten times the code, or ten times the designs, or ten times the written analysis - what have you got? More output from one person in the system. The system itself hasn't changed. The other people are doing what they were doing before, at the rate they were doing it, with the same information flows and the same decision-making patterns.

Individual multipliers are linear. You hired a 10x engineer; you got 10x from that engineer. If you have p team members and the median production is x, what you have is px+9x - where one x generates ten times what the other x's do. That's addition. That's good, and it's better than just px, but...

Some people don't add to a team. They change the team's exponent. They affect how everyone else operates - what information flows where, how decisions get made, where uncertainty resolves faster. They change the team's character. The value they generate isn't necessarily in their own output; it's in what the team becomes capable of that it wasn't before.

That's not addition. A team of five doesn't become six and a bit. It becomes something else entirely: not a larger machine, but a different kind of machine. This is better understood as an exponent - the right person doesn't just add output. They change the rate at which the whole team can resolve uncertainty, make decisions, and move.

The AI Counterexample

AI coding tools have made this distinction impossible to ignore - and yet, a lot more confusing. It is now genuinely possible to produce ten times the code in a given day, even for a below-average coder.

The problem is that we're still measuring velocity, as though speed were the only thing that mattered. But velocity without direction isn't progress; it's drift. A below-average coder with AI can now move very quickly in the wrong direction - producing ten times the code and a fraction of the value, because the system doesn't cohere, the architecture wasn't considered, the edge cases weren't examined, and the team now owns something nobody fully understands.

This is vibe-coding at its worst: output inflation without judgment. The multiplier is real, but the system value is negative. High velocity in the wrong direction costs more than going slowly in the right one.

The same tools in the hands of someone who understands what "good" looks like produce something different entirely. ByteCode.News' UI itself was built from scratch in a few weeks, with significant AI assistance - but the direction was never delegated. What it needed to be, what "quality" meant, what was acceptable - those decisions were made continuously, not generated.

The result isn't "a lot of code got written." It's that BCN exists, works, has architectural coherence, and has no passwords to expose - because the heading was correct the entire time.

Output is visible. Direction is structural. The multiplier model measures the first and misses the second entirely.

What It Looks Like

A senior engineer who writes the document that prevents six people from going down the wrong architectural path for three months doesn't produce more code. They change the shape of the next quarter.

An editor who gives a writer the precise note that unlocks a stuck draft doesn't write the piece. They change what the piece becomes.

A project manager who addresses the assumptions that their teams are making about requirements. They change what the delivery looks like, from "we're still working on it" to "it's working."

None of these people show up well in individual output metrics. All of them are doing something more valuable than the person who ships twice as fast - or even four times as fast - in isolation.

The Mechanism Is Structural

What these roles share isn't a skill exactly - it's a position in the system. They sit at junctions, perhaps implicitly, hopefully explicitly. They're not faster nodes; they're better-connected ones, operating as magnifiers on team topology rather than on their own throughput.

They're the people asking the right questions, making the right suggestions, measuring the mood and mode of the team and making everyone happier and more productive, looking at what matters rather than what's flagged.

This is why the multiplier framing misses. A multiplier is a property of a node. An exponent is a property of a system. The question isn't "how much does this person produce?" It's "what does the team become when this person is in it?"

Synergy - real synergy, not the word drained of meaning by its role in helping win a thousand games of buzzword bingo - isn't linear in the optimal case. It compounds. The goal isn't addition; it's a higher exponent.

Not Just Engineering

This pattern shows up anywhere complex work happens in groups. The teacher whose framing of a concept allows thirty students to build on it rather than just repeating the concept word for word, the journalist whose beat knowledge lets an entire newsroom know which calls to make - or the editor who knows which words to cut and tells the journalists why.

None of these are an example of "10x." Yet all of them are changing the exponent of the system around them.

The Better Question

When evaluating someone's contribution - including your own - the useful question isn't "how much do they produce?" It's "what does the system do differently because they are in it?"

That question is harder to answer. It doesn't fit neatly in a performance review. It often shows up as absence - the crisis that doesn't happen, the bad decision that gets caught early, the month that doesn't get wasted.

But it's the right unit. Once you start measuring for it, you start seeing it everywhere. And you can try to be one of those exponential people - not a "10x," but a power.