Originally written in French. Translated by AI — the meaning has been preserved, not the prose.
Angle
A capture setup isn't tuned by what it retains, but by the delay it imposes between meeting a source and deciding to work on it: while it's hot, the filter must stay weak, because the excitement of the find makes judgment unavailable and a severe sort at that instant loses sources that would have counted; a few days later, the same list sorts itself almost unaided, because the captures that held only by their novelty have stopped holding and the accumulated mass finally makes visible the processing time they all demand together. These are two distinct regimes of judgment, and conflating them is the single cause of the two failures ordinarily blamed on tools — the To read folder that swells without end, and the system so demanding at the entrance that it no longer receives anything.
Synthesis
The starting point is a gap in value that the gesture conceals. Saving an article and drawing a reformulated idea from it both produce one more file; only the second produces knowledge. Three levels separate one from the other — capturing keeps, distilling transforms, acting produces — and a setup that stops at the first hasn't capitalized, it has accumulated.
From that gap follows the rule about time. Since a capture is worth only what its resumption makes of it, it is the decision to resume that must be well made, and it is precisely the one you cannot make at the moment of the encounter. What is decided while it's hot is not a source's value, it is its promise: can it be attached to a project, a question, an ongoing line of thought. Nothing more, and above all no complete justification — capture is a placing in waiting, not a validation.
The delay then does two things you don't ask of it. It exhausts the seduction of novelty, and it lets the stock build up. The second effect is the most counter-intuitive: examined alone, a capture always defends itself, because its cost lies not in it but in the processing time it shares with the others. It is mass, and mass alone, that makes that cost visible — a list of twenty or thirty items says at once what no single item said. The right reaction to an overflowing list is therefore not to file it better, a gesture that destroys exactly the signal it was emitting, but to cut.
Cutting requires lifting an inhibition and setting aside a false way out. The inhibition is attachment to what has been written: the cost already paid for entering it passes itself off as future value, when deleting a capture destroys no knowledge — it protects processing capacity, the only genuinely scarce resource. The false way out is the second space, the "maybe later" folder, which has never deleted anything: it moves the stock to a less visible place, and the less visible place is precisely where mass stops doing its work.
What remains is not to mistake the scope, and that is where the thesis stops. The distance between discovery and decision assumes an external source, which waits intact while you cool down. A personal thought does not wait: what is precious in it — the momentum, the association, the example that made it tangible — fades during the very delay used to sort the others. Hence two spaces and not one, and hence the fact that the subject was never the tool: a setup is judged by what it makes you produce, not by the completeness of its coverage, and an elegant system that absorbs more upkeep than it returns is simply too heavy.
Tensions / contradictions
The head-on contradiction bears on the delay itself. Here it is what makes sorting possible; elsewhere it is what empties an idea of its substance before you could use it. Both hold, and the dividing line is where the material lives: in the source, which doesn't move, or in the head of the person who had it.
A second tension, on mass. An unbounded queue cancels arbitration when it is collective and never reread; yet it is accumulation, in an individual practice reread as a block each week, that provokes arbitration. The same stock therefore does the opposite depending on whether it is looked at — which shifts the question from size to frequency of rereading.
Questions
- Can a setup shared by several people hold two regimes of judgment, when the second assumes the same person captured and then decides?
- Would what cooling makes disappear from a source — the mental context of the encounter — deserve to be captured too, or is its loss part of the sorting?