Main idea

Describing a product's objects can be done in a formal language, with the tools that derive conclusions from it and the standards that go with them; or in one text card per object, a title, a few fields, some links — whatever a text editor opens.

Choosing between the two is not a question of seriousness, it is a question of available information. The rigour required follows from the use, and the use has not started. Writing in a heavy formalism before knowing what it will be used for means paying a constraint whose counterpart has not arrived yet, and shutting out the contributors who cannot read it.

The asymmetry settles it: starting simple and getting more complex later costs less than starting complex and having to deflate everything. A rich format can be derived from a poor one that has been kept properly; the reverse means going back through every card.

Why it matters

It lifts the most banal blocker of this kind of undertaking: waiting until the right tooling has been chosen before writing the first card. The question of the tool becomes one you are entitled to postpone, and that will be settled on facts.

It also works as a review criterion: a formatting constraint that no identified use calls for is to be removed, not defended.

Nuances and limits

The rule does not excuse you from keeping the form: what has to stay rigorous from day one is traceability — where each piece of information came from — and the stability of file names, because those are what will allow conversion later without rereading anything.

And postponement has a limit: the larger the base grows, the more a change of format costs. The window in which "we'll see later" stays true is not infinite.

Open questions

• What signal, in the use, indicates that the poor format has stopped being enough — and that you are not simply adding formalism for comfort?