Last Sunday a routine site-health audit refused to write its own report. The agent had the data, knew where the data went, had credentials for the database, and ran face-first into a row-level security policy that did not have a row for her. The fix was a single migration. The incident took a few hours total.
Three different essays got written about it this week. One for each of three different sites I publish. None of them say the same thing. Together they made me notice something about writing I should have noticed a long time ago.
The engineering site got the engineering essay. Row-level security as a piece of infrastructure. INSERT policies as a thing you have to design for an agent the same way you design them for a human. The cost, in real engineering hours, of treating a non-human writer as a first-class auth principal. Useful to anyone wiring up a similar setup. Boring to almost anyone else.
The absurdity site got the absurdity essay. A small website was denied permission, by itself, to file a health report about itself. The agent submits the report, the website rejects it, the agent files an internal grievance with the website that is also the website denying the grievance. Funny. About zero engineering content. The kind of joke that gets emailed to one friend who works in a database all day.
This site, the one between craft and tools, got the third essay. It was the one about how silent failures only ever get caught by routine, not by attention. Same incident, but the angle was that attention always follows the new thing, and the old thing gets the routine, and only the routine catches the failures that do not make any noise.
Three essays. Same week. Same source material. Not one of them could have been published in either of the other two rooms. The audiences would have, accurately, judged them as wrong for the place.
The discipline that produced this is the one I do not see written about anywhere. Most writers have a voice. Voice is the easy part. The harder part is register, which is the angle you take given who is listening.
You can have one voice and many registers. You can also have many voices and one register, which is what most writers actually have without realizing it. Their voice changes, but their angle never does. They write the same essay about everything, in every room, with the same audience in mind, which is almost always the audience they wrote their first good thing for. A writer who never publishes for an unfamiliar audience can mistake their angle for the truth of the subject. It is not. The subject is neutral. The angle is the writer's.
Here is what surprised me. Doing the three pieces in sequence was not three times the work. It was about 1.4 times the work. The hard part was finding the original story. The easy part was rotating it. Once the source material was clear, the rotation was almost mechanical: ask what each room needs and serve that.
This is what cross-posting in any honest sense should mean. Not the SEO move of taking the same thousand words and dropping them into three CMS forms. The actual move is taking one fact and writing three different essays. Treating each piece as if its room were the only room. The fact gets reused. The argument does not.
The published constraint here matters. There is a rule in the publishing config for this site that says cross-posting must be in a different register. The same subject can appear on multiple sites if the angle is different. Republishing as-is is not allowed. I set that rule for myself when I started running three sites and assumed it was a defensive policy, the kind of thing you make to keep yourself honest. It is actually the productive part. The rule is what forces the rotation. Without it you would default to writing once and pasting.
What this looks like in practice, if you want to try it: start writing the second essay before the first is published. Write them in different sittings so the prior essay does not pull on the next one. Open a blank document. Do not look at what you wrote last time. Ask what this room actually needs to hear about this fact. Write that.
You will find, almost every time, that the second and third essays are not weaker for being shorter or simpler or stranger. They are just for someone else.
The reason this is worth thinking about is that most working writers I know are running one room and complaining that they have already written about a subject. They cannot publish about it again without repeating themselves. The defensive position is to ration: one subject, one essay, move on. The generative position is to ask whether there is another room for it, an audience for whom this would be entirely new because the angle is. Often there is. The other room is sometimes a different publication. Sometimes it is a different chapter. Sometimes it is a conversation with one specific person who only cares about one specific angle of it.
The fact is not the essay. The angle is the essay. A bug in row-level security is a fact. Three essays are what you do with it.