On Being a Generalist in an Age of Specialists

Every few years someone publishes an article explaining that the generalist is dead — that in a world of hyper-specialisation, the only people who matter are those who go ten thousand hours deep into a single domain. These articles are always well-argued, always slightly wrong, and always written by specialists.

I’ve spent most of my career being deliberately, stubbornly broad. I’ve worked in backend infrastructure, built ML pipelines, written product strategy documents, done customer interviews, and spent more time than I’d like to admit thinking about typography. This has made me, by specialist standards, mediocre at most things. It has also made me unusually useful in specific situations.

The Coordination Tax

Here’s something that rarely gets discussed in conversations about specialisation: coordination is expensive.

When a team consists of specialists who each own a narrow domain, everything that crosses domain boundaries becomes a negotiation. The backend engineer and the ML engineer have to agree on an interface. The product manager has to translate between the engineer’s constraints and the designer’s vision. Every handoff is a potential loss of fidelity.

A generalist can reduce this tax. Not because they’re better than any specialist at their specialty, but because they can speak multiple technical dialects well enough to translate without distortion. They can see the whole board.

The Synthesis Advantage

There’s a second advantage that’s harder to articulate but more valuable: generalists are often better at synthesis.

The best insights don’t come from going deeper inside a domain — they come from noticing that two domains are secretly the same problem. Information theory and genetics. Market microstructure and evolutionary biology. Compiler design and organisational structure. The people who make these connections tend to be people who’ve spent time in both fields without fully committing to either.

This is the dirty secret of interdisciplinary breakthroughs: they require someone who’s simultaneously an outsider in both fields and literate in both.

The Honest Trade-off

I’m not going to pretend there’s no cost. There are things I can’t do that specialists can. I will never write a novel distributed systems paper. I will never design a state-of-the-art training loop. If you need the world’s leading expert in a narrow problem, I’m not it.

But if you need someone who can identify which problem is actually worth solving, talk to the people building it and the people using it, and hold the whole thing in their head long enough to make a decision — that’s a different job description, and breadth is a genuine advantage there.

The question isn’t “specialist or generalist?” It’s “which problems actually benefit from each?” For most organisations, the answer is: more generalists than you think, placed at the right points in the system.