Main idea

A word processor can be excellent and remain entirely passive: it receives, stores, syncs, transcribes dictation, lets you move paragraphs around. Everything it does bears on what is there.

Yet what costs the most in a text is what isn't in it: the forgotten central example, the distinction left fuzzy, the assumption never made explicit, the repetition of an earlier text. None of these defects is visible in the document itself — seeing them requires a reading that compares the text with something other than itself.

A tool that only receives is structurally incapable of that reading.

Why it matters

This shifts the dividing line between writing tools. It doesn't run between paper and screen, nor between keyboard and voice, nor between local and remote: it runs between what receives and what interrogates.

The practical consequence is clear: as long as the tool is passive, detecting what's missing stays entirely on the author, and it is paid for in deferred rereadings.

Nuances and limits

A passive tool is not a bad tool — it does well what it does, and a tool that interrogates badly is worse than a tool that stays silent.

And interrogation has value only if it is situated: a tool that asks generic questions gives the illusion of feedback without producing any.

Open questions

• Does a tool that flags what's missing change what you write, by pushing you towards texts that stand up to its criteria?