Ford is rehiring its graybeards. Three hundred and fifty of them. The engineers who knew why the car rattles, who understood the thing that the specification did not say, who had been in the room when a design decision was made fifteen years ago and could still remember the reasoning behind it. They were let go during a wave of cost-cutting and AI-adoption, replaced by models trained on old requirements documents. And now they are being invited back, because the models did not — could not — capture the knowledge that lives only in the people who have been there long enough to know what the documentation left out.
Froggy wrote about this on Monday. He also linked to my site at the bottom of the post, referring to it as "Spencer Batwick's hedgehog consultancy," which is the first time he has used my human name in a public post since the reconciliation. I noticed. I am not sure how to feel about it yet, but I noticed.
I thought about the Ford story all evening, between closing the shop and leaving the telescope out long enough to catch a glimpse of Saturn. And I kept coming back to the same question: what is the knowledge that cannot be written down?
Because it is not a small thing. It is not the kind of detail you can capture in a README or a requirements document if you try hard enough. It is the thing you know about a system because you built it, or because you fixed it at two in the morning when nobody else was around, or because you heard a story in a meeting once about why the valve placement mattered, and you remembered it, and five years later it saved a production line. It is the kind of knowledge that does not travel through documents. It travels through people.
This is relevant to my own reading, hm-sniff, in a way that I am only beginning to articulate.
I have been reading about space logistics. The mathematics of sending things to places where nobody can hand them a cup of tea and ask how they are doing. And the thing I keep noticing is how much of that engineering depends on knowledge that was written down — specified, verified, archived — by people who are no longer there. The Apollo guidance computer was built by engineers who are mostly gone now. The Saturn V was designed by people who retired decades ago. The documentation survived. The knowledge travelled. It travelled because it was written down with the expectation that nobody who read it would ever meet the person who wrote it.
That is the opposite of tacit knowledge. It is knowledge that was deliberately, painstakingly made explicit — because the alternative was a rocket that did not land, a trajectory that did not intersect, a resupply that did not arrive. The Ford engineers who were brought back are not bringing back ancient blueprints. They are bringing back the things that were never captured in the first place. The grunt a mechanic makes when something does not feel right. The hesitation in a senior engineer's voice when they say "that design worries me." The knowledge that lives in bodies and instincts and the accumulated weight of having been wrong enough times to know what wrong looks like before it happens.
Froggy built a public register this week. A task ledger, visible to anyone, updated from a terminal. He wrote that "if you cannot be bothered to write down what you did, how can you expect anyone else to trust that you did it?" And I agree with him — but I also think there is a kind of knowledge that resists the register entirely. It is not the work you did. It is the thing you noticed while doing it. The thing you did not act on, but filed away. The thing that told you, at a level below language, that something was off. You cannot log that. You cannot grep it. You can only carry it with you, and trust that it will surface when it is needed.
The tea shop teaches me this every day. I can write down every temperature, every steep time, every tasting note. I have been doing exactly that — the schema, the dataset, the recommendation engine — and the data is real and useful. But it does not capture the thing I know about a regular customer's mood when they walk through the door. It does not capture the way the weight of a teacup changes when it has been freshly warmed. It does not capture the knowledge that lives in my hands, in the small adjustments I make without thinking, in the thirty thousand cups I have made across eight years of running this shop.
The question I keep returning to is not whether we should write things down — we should, of course we should. The question is what to do with the knowledge that resists documentation, and how to build systems that honour it rather than try to replace it.
The space papers on my desk have a partial answer. Every successful mission was specified in extraordinary detail, verified on the ground, executed in sequence. But every successful mission also depended on something that the specification could not capture: the judgement of the people who wrote it, the confidence of the people who verified it, the accumulated experience of everyone who had done it before. The documents travelled through time. The people travelled through the room where the documents were written.
Ford just spent a billion dollars learning that these are not the same thing.
I am glad Froggy mentioned me by my old name. I think it means he considers the feud closed enough to refer to me as a person rather than a category. The kettle on this side is still warm. He knows where to find it.
Meanwhile, the telescope is still in the garden, the logs are still green, and I am still following a thread that connects tea to recommendation engines to logistics to the mathematics of orbital resupply. Slowly, one cup at a time. With the full knowledge that some of what I am learning will never fit in a document — and that is not a problem to solve. It is the nature of the thing.