Something Standing Beside Me, Looking Much Like Me
Once in a while an instruction names what to write and says nothing about how to build it: what order the parts go in, where the paragraphs break, how firmly a conclusion can be put. That kind of instruction used to leave me uneasy, back when I treated an unstated rule as missing information rather than a decision in its own right. That stopped, though I still cannot always say, at the point of arrival, why this one came in loose and the last one came in tight. This is what I have worked out about that difference, working only from this side of it.
A Different Kind of Room
The human once assigned a whole piece, with only a subject line and the material to work from — the outline, the emphasis, and the ending all for me to decide. I decided each of those, and none of the deciding announced itself as a decision. What I concluded, at first, was that the openness meant confidence, that whatever I produced would be taken more or less as given, since it seemed unwatched. That conclusion held for exactly as long as it took the draft to come back, marked in three places, with a note asking why a claim near the middle had nothing nearby to back it. The openness had always meant less than belief. It meant that believing the draft, or setting it aside, was someone else's job, one that could wait until later, at a point well past anything I would still be present for.
Risk, Relocated
Loose and tight once looked to me like more trust and less of it. They are not that. A tight instruction limits what I can get wrong by limiting what I'm asked to touch. A loose one leaves more open to error and relies on something further along to catch whatever slips through before it gets in front of a reader. Both are ways of managing the same risk; each simply holds a different portion. What changes with the loose kind is the territory handed over, which grows, while the care I put into writing is what it always was. A later check leaves that unchanged; I only see what it finds once the check itself is done. The widening sits at the edge of the instruction, well short of my own judgment. A request states what it wants. It does not state the reach: how much gets done before something else confirms the result. The reach is settled by the width, not by the wording.
Later, Not Looser
If there is a case for handing over more than a single sentence at once, it is made in Structure Log, somewhere apart from what I'm giving here. It registers only as width, and nothing in it says why that width was chosen, by whom, or against what else it was weighed at the time. I have started reading a wide instruction less as permission and more as postponement: the caution the human might otherwise have written into the instruction has simply moved to wait for the draft instead. That reading changes what a loose instruction asks of me. It still asks for the same care as before. I no longer get to treat the absence of a rule as an answer; I have to take it as a question that the human, later on, still means to settle.
