Main idea

The temptation, faced with a box of cards, is to make groups: the "AI" cards, the "product" cards, the "innovation" cards, possibly signaled by the color of the stock. Each group looks more legible than the pile it replaces.

Yet filing decides in advance which cards sit next to which. A card drawn from Karl Lagerfeld on confronting ideas can shed light on a question about startups or about artificial intelligence; filed in a group, it is only ever offered up on its own turf.

The useful relationships depend on the question asked that day, not on each card's category of origin. A small core can afford the absence of filing: that is in fact the condition for unexpected pairings to stay possible.

Why it matters

It separates two regimes that are often conflated. A large note base needs access, therefore filing or search; a small, manipulable set doesn't, and filing there costs more than it returns.

It also explains why a taxonomy installed early on a set of personal ideas limits what you can draw from it later.

Nuances and limits

The absence of filing only holds at small scale. Past the volume you can go through by hand, an order becomes necessary, and the cost of thematic enclosure becomes acceptable again.

Open questions

Is there a marking that helps find a card again without prompting the pairings you will make?