Main idea

Linking notes in Obsidian or building a graph in Miro works, but every relationship goes through an interface: selecting a tool to draw an arrow, choosing its direction, zooming out to recover the overview, selecting an element before moving it. Between intention and gesture, an operation that doesn't belong to thinking is inserted every time.

On a table, picking up a card and putting it somewhere else is enough. All the cards stay visible at once, with no zoom level; a layout can be tried, destroyed in a second and replayed differently, with nothing recorded.

The table is then not a memory support: it's a space of composition, where the layout follows a thought in the process of forming instead of recording relationships already taken for granted.

Why it matters

It gives a criterion for choosing a thinking support: not what it lets you preserve, but how many operations it inserts between an intention and its execution.

It also explains why an already rich digital graph is of little help in thinking something new: it offers existing relationships where composition requires starting from nothing.

Nuances and limits

The advantage depends on small numbers: a few dozen cards fit on a table, a corpus of thousands of notes doesn't. Beyond that, the digital tool becomes the only workable one again.

And what is composed this way isn't preserved: the layout disappears when the cards are put away.

Open questions

Could a digital tool remove the interface to the point of matching that gesture, or does physical manipulation bring something other than the absence of an interface?