Does Explaining Your Job Actually Save It?

The week unfolds like this:

• Remote jobs from around the world - Monday

• The Hidden Layer - Tuesday

• The Long Game - Wednesday

• Mental Models - Thursday

• One Quote, Unpacked - Friday

• The Real MBA - Saturday

• Weekly Recap - Sunday

The Long Game

In 1894, on a rock called Oxcars in the middle of the Firth of Forth, two lighthouse keepers were withdrawn for good. A clockwork timer took over the daily operation, while a boatman continued to visit periodically to service it.

Nobody threw them a party. Nobody wrote it down as history. It was just the first lighthouse in Scotland to be automated, and two men going home while a mechanism took the watch.

It’s tempting to call this the start of a hundred-year handover, and in a loose sense it was. But the truth is stranger and less tidy than that. Oxcars wasn’t even fully automated by modern standards, someone still had to keep visiting to keep it running. And it sat alone for decades. The Northern Lighthouse Board didn’t attempt automation at scale until the 1960s, driven less by any grand technological leap than by a blunt fact: manning more than ninety lighthouses had become extremely expensive. The last Scottish station, Fair Isle South, wasn’t automated until 1998. Trinity House’s last manned lighthouse in England, North Foreland, followed that same year.

So the real story isn’t one continuous hundred-year project to crack a single hard problem. It’s a slow, uneven, cost-driven rollout that happened to start in one unusually convenient spot in 1894 and finish, for entirely different reasons, more than a century later.

Which raises the more interesting question: what actually was hard to automate, and why did it take so long even once the cost pressure was obvious?

Some of it was plainly technical. Reliable offshore power. Fog signals that could trigger themselves correctly. Systems robust enough to run for months without a human checking on them. Those are solvable engineering problems, and they got solved, station by station, over decades.

But keepers weren’t just switch-flippers standing next to a light. It’s reasonable to imagine that they accumulated the kind of tacit knowledge that anyone who has spent years doing a physical, high-stakes job in a fixed place will recognize: a felt sense of when a familiar structure sounded wrong in a storm, an instinct for weather shifting before an instrument confirmed it, a read on equipment that came from years of proximity rather than any manual. There’s no surviving record of specific keepers describing this, but the pattern is a common one wherever people do the same physical job in the same place for decades. This is what the philosopher Michael Polanyi, in his 1966 book The Tacit Dimension, called knowledge that exceeds what its owner can put into words. We know more than we can tell, he wrote, and he meant it as a fact about how minds work, not a metaphor.

Decades later, the economist David Autor picked up Polanyi’s observation to make sense of a different puzzle: why some tasks that look intellectually impressive get automated almost overnight, while humbler, more physical jobs resist for far longer. His answer, echoing Polanyi, was that it’s hard to program a machine to perform tasks whose rules humans themselves have never been able to adequately specify. Autor’s paper is about the broader shape of computerization and labor markets, not lighthouses specifically. But the mechanism he describes fits the keepers well enough that it’s worth borrowing, carefully, as an illustration rather than a documented cause.

Here’s where a comforting version of this story usually shows up, and where it’s worth being suspicious of it.

The tidy moral goes: make your tacit knowledge explicit, get it into the system, become the person watching over the automation instead of the person it replaced. Modern lighthouses genuinely are monitored remotely now, North Foreland’s from a Trinity House centre in Harwich. It’s a real and appealing image: the keeper’s replaced, but the watching continues somewhere, on a screen.

What isn’t established, though, is that this is how the actual keepers fared. There’s no record showing that the people who lost their postings were the ones retrained into those monitoring roles, or that their tacit knowledge was what got them there if they were. What’s documented is blunter: automation was pursued because paying to keep people on ninety-plus remote stations was expensive, and once the technology existed to do it more cheaply, it did. Some keepers likely found other work in the same organizations. Many did not. The profession, by Trinity House’s own account, largely passed into history once the last stations went automatic.

That’s the part worth sitting with, more than any story about who ended up in the control room.

The tacit knowledge itself doesn’t need to have been captured, transcribed, or rewarded for the larger point to hold. Even if none of it made it into a manual anywhere, the navigational function the keepers served was preserved in the automated system that replaced their postings. The occupation disappeared. The capability didn’t. Those are different things, and it’s easy to blur them when the story is being told as a lesson rather than a history.

Which is the harder question underneath the advice to “make your knowledge explicit” before a machine gets there first. It might be worth doing. It might make you easier to keep around, or easier to retrain, or simply easier to have a conversation with about what you actually contribute. But there’s no guarantee, in the lighthouse record or anywhere else, that articulating what you know is what determines whether you’re the one still needed once the system that replaces you is built. Sometimes the economic value goes to whoever builds the monitoring centre, not to the people whose work the monitoring centre was built to replace.

The lights are still burning on every one of those rocks tonight. Nobody’s standing at the top of the stairs. What happened to the people who once did isn’t a single clean story, and it’s worth resisting the version of it that makes it one.