Undervalued on Paper, Irreplaceable in Practice
The Generalist Series, Part 2
The generalist’s LinkedIn profile looks like they couldn’t make up their mind. Their performance review says “meets expectations.” Their manager isn’t sure how to level them.
Then they leave. And everything gets weird.
Nobody can quite explain why things feel slower, messier, more siloed. The work that used to just get picked up stops getting picked up. Decisions that used to get made start needing meetings. Nobody can point to the gap because nobody had a word for what filled it.
The system wasn’t built for them
In the first post in this series, I described the generalist: the person on an engineering team who goes where the problem is, regardless of whose territory that is. The connective tissue. The one who holds both sides of a boundary problem.
This post is about why that person is almost always undervalued.
Levelling frameworks reward depth. You get promoted by going deeper into your domain: more expertise, more ownership, more mastery within a defined scope. The generalist who spent the quarter solving three cross-functional problems that nobody else could see looks, on paper, like a distracted specialist. They touched too many things. They didn’t go deep enough in any of them. The rubric has no category for “prevented two incidents and connected three teams.”
Performance reviews compound the problem. They ask: did you deliver on what you were hired to do? The generalist’s honest answer is often “well, it depends on how you define my job.” Most review systems have no rubric for this person. The role was never written down in the first place.
Job descriptions close the loop. If you can’t title the person, you probably never wrote a JD for them. You didn’t deliberately hire for them. Their presence on your team was accidental. You got lucky. And luck is not a strategy.
The Tiger model
David Epstein opens Range with a contrast worth sitting with.
Tiger Woods started practising golf before he could walk. By three, he was already on the path that would make him the most dominant player the sport had ever seen. His career is the story we tell about mastery: specialise early, go deep, compound over decades.
Roger Federer tried badminton, basketball, skiing, wrestling, swimming, and tennis before eventually settling on tennis relatively late. For years, his path looked like a failure to commit.
Epstein’s argument is that most domains work more like Federer’s than Tiger’s. Tiger’s model is optimised for what Epstein calls a “kind” learning environment: fixed rules, clear feedback, patterns that repeat. Golf qualifies. Chess qualifies. Most engineering work doesn’t.
Engineering is a wicked environment. The problems are cross-domain, context-dependent, and always shifting. Yesterday’s best practice is today’s technical debt. The thing that worked on your last team doesn’t transfer to this one because the codebase is different, the people are different, and the business context has changed.
Companies built their career frameworks on the Tiger model. Specialise early. Go deep. Demonstrate mastery in a defined scope. The levelling rubric rewards this. The performance review measures this. The promotion criteria select for this.
It’s a rational system. It’s just optimised for the wrong environment.
The breadth penalty
There’s a study Epstein cites that should unsettle anyone who designs career systems.
Scientists who maintained creative pursuits alongside their technical work, who painted, played instruments, wrote, built things outside their field, were far more likely to produce Nobel-level contributions. Those who focused exclusively on their discipline didn’t keep up.
Breadth of exposure didn’t dilute their contribution. It predicted it.
The implication is uncomfortable: the career systems most engineering organisations run aren’t just neutral toward generalists. They’re filtering them out. Every time a screening process skips a pivoting CV, or a rubric penalises “lack of specialisation focus,” the system isn’t neutral. It’s selecting against the people most likely to produce original thinking.
This isn’t an abstract concern. If your promotion criteria say “deep expertise in X” and your most creative problem-solver has moderate expertise in X, Y, and Z, they won’t get promoted. They’ll plateau. And eventually, they’ll leave.
The work that doesn’t show up
The third reason companies get this wrong is subtler than the first two.
Generalists do a lot of work that prevents problems from occurring. They’re the person who noticed the API contract mismatch before it became an incident. The person who raised the question about the third-party dependency before the team was mid-sprint. The person who caught the assumption gap in a handoff before it caused a month of rework.
Prevention is invisible. Incidents are visible. The generalist’s contribution often shows up most clearly in what doesn’t go wrong and that almost never appears in a performance review.
There’s a concept called Chesterton’s Fence: before you remove a gate or a process step that seems pointless, you should first understand why it was put there. We remove the step nobody likes, streamline the process, cut the approvals and only discover what we lost when something breaks that the old, annoying step used to catch.
The generalist is often that step. The invisible check. The person who asked the awkward question in the planning meeting that saved the team two weeks of wasted work. You don’t notice what they prevent. You only notice when they stop preventing it.
The quiet exit
Generalists rarely leave loudly. They don’t storm out over a bad review or a pay dispute.
They leave quietly. At some point they realise the organisation has no language for what they do and no path for where they could go. They were never told they were indispensable. Nobody had the vocabulary to say it. The manager liked them, valued them even, but couldn’t articulate “why” in a way that showed up in the system.
So they go somewhere that needs them more obviously. A smaller company. An early-stage startup. A consulting role. Somewhere the boundaries are messier and the value of crossing them is harder to ignore.
And the team spends the next three months wondering why everything feels slightly harder than it used to.
Generalists aren’t usually hard to manage. The problem is that we built the system before we had words for what they do. And systems without vocabulary for something tend to treat it as if it doesn’t exist.
So if the system is the problem, can we fix the system at the point of entry?
The next post is about hiring: what it looks like to find this person, and how our interview processes filter them out before they reach the table.
Have you ever lost someone like this and only understood what you’d lost after they were gone? Let me know in the comments.