Main idea

As long as a body of material only has to produce one output, keeping nothing costs nothing. You work, you write, you send, it's over — and the setup looks sufficient because it was never put to the test.

The test comes when the same substance has to serve again: a spec, then user stories, then a pitch, then a note for sales, then an answer to an RFP. Every new format feels like starting over, and only then do you pay for the absence of a preserved layer.

The cost isn't deferred in the sense of growing slowly: it is nil, then it appears all at once.

Why it matters

This explains why the decision to invest in capitalization is almost always taken too late. At the moment it would pay off — during the first piece of work — no data justifies it; by the time it is justified, the material of the first piece of work has already disappeared.

It also gives an observable signal: the first time you open an old document to draw a new one from it, the setup has failed.

Nuances and limits

Some work really does have only one output, and capitalizing there would be waste. The difficulty is that you rarely know in advance which case you're in.

And capitalization done during the first piece of work has a cost too. The trade-off bears on a probability of reuse, not on a certainty.

Open questions

• What sign, during the first piece of work, announces that there will be a second?