Technology loss

Why have societies forgotten skills their ancestors clearly possessed, with no disaster to explain it?

In 1968 the United States could build an F-1 rocket engine. Forty years later it could not, and the engineers who wanted one had to take museum specimens apart and scan them, because the drawings that survived did not tell them enough. Nothing was burned. No plague came. The country was richer and better educated than before.

So the puzzle is not really about catastrophe — catastrophe is the easy case. It is about the ordinary case: a skill that a society plainly had, and then plainly did not, with the archive intact throughout. What exactly is it that goes missing, if the documents are still on the shelf?

Start by asking where a skill is actually kept. Not in the drawings — that is the lesson of the rocket engine. A written procedure is a reminder for someone who already knows how to do the thing; it underdetermines the practice. How hard to press. What the metal sounds like when it is ready. Which of the twenty steps you may fudge and which you may not. That knowledge is in hands and habits, and hands die.

So a skill is not stored at all. It is rebuilt, from scratch, in a new body, every generation. That reframing does most of the work. Ask not "why was it forgotten?" but "why did the rebuilding stop?"

Now, the rebuilding is lossy. An apprentice watching a master ends up with an imperfect copy — usually a slightly worse one, occasionally a better one. Whether the population's best practice climbs or slips is a race between how much each copy degrades and how many models a learner can sample. Sample one ageing master and you inherit his errors. Sample thirty and you can find the best of them, and someone in the next cohort will exceed him.

That is where scale enters, and it is why this shows up in island populations first. Tasmania was cut off from mainland Australia when rising seas flooded Bass Strait around ten thousand years ago, and its later archaeological record shows bone tools and fish bones dropping out of the assemblages — an argument that a small, isolated pool of practitioners could not hold complex techniques. I should say plainly that this case is contested: some archaeologists read the same evidence as a deliberate change in diet and toolkit rather than an inability, and the dating is argued over. The model is more secure than the example.

But shrinking populations are the rarer route, and they are not what happened to the F-1 engine. The commoner route needs no disaster and no decline at all: the demand stops. A skill was never a possession of "a society" — it was the livelihood of a few hundred specialists, kept alive because customers paid for it. Take away the customers and no apprentices present themselves, and one generation later the capability is gone from a country whose population and wealth both grew.

Watch how many losses fit that shape. Rome's hydraulic concrete was not a forgotten recipe so much as an unbuilt building: the empire that commissioned harbours and domes stopped commissioning them, and there was nobody to keep the trade warm. Wootz steel — the crucible steel behind the best Damascus blades — stopped being made in the eighteenth and nineteenth centuries, and one leading suggestion is that the specific ore bodies with the right trace impurities were worked out, so the input vanished under the practice. A substitute can do it too: nobody suppressed hand typesetting.

Three refinements hold this together.

First, "forgetting" is the wrong verb throughout, because it implies something was being remembered. Nothing was. Continuous reconstruction was mistaken for storage, and the mistake is invisible until the reconstruction stops.

Second, scale is not headcount but connected practitioners. A large population fragmented into isolated workshops behaves like several small ones. This is why the argument applies to a modern supply chain as readily as to an island.

Third — and this is the honest caveat — loss is usually not a failure. Most of it is a society correctly declining to pay for something it no longer wants. Calling that decay smuggles in a judgement the evidence does not carry. The cases genuinely worth worrying about are the narrow ones where a capability is dropped for a substitute that later turns out to be unavailable, and by then the last people who could teach it have retired.

• Why cumulative culture ratchets upward at all, given that every copy is imperfect.
• Whether the same accounting explains capabilities lost inside a single firm across a decade.

Every term the collection defines is gathered in the glossary.