Why Delegate a Newsroom
Decomposition did not discover four equal parts — it discovered three the agent could carry and one the practitioner could not hand over.
1. Where this sits
This is the first Field Note under The Newsroom, Delegated — the experiment that ran while the Lab was being built, and the instrument the Lab itself runs on. It covers the opening two steps of the method: isolating the act and decomposing it with predictions recorded before anything ran.
—
2. What this step set out to do
Name the act of work in one sentence. Then split it into its working parts and write down, before building, which parts the agent was expected to hold and which it would drop.
—
3. What was predicted
No prediction was recorded for isolation and decomposition themselves — these were the steps that produced the predictions, not the steps predictions were made about. The four predictions that emerged, recorded before building:
> Scouting: Delegable. The criteria could be made explicit enough for an agent to apply without correction.
> Research: Assisted. The agent would assemble evidence competently but would need oversight on whether the thesis survived genuine challenge.
> Production: Assisted. The agent would produce publishable drafts that preserved the thesis and conclusions, requiring minor corrections but not rewriting.
> Publication judgment: Human-only, by design. The decision to attach a name to a piece of analysis was not delegable regardless of output quality.
Overall prediction: Assisted.
—
4. What happened
The act isolated in one sentence: an editorial operation that scouts signals across six domains, researches them adversarially, produces written analysis, and selects what to publish — running continuously, delegated to agents, governed by stage-gated discipline.
The case for choosing this operation as the first experiment was time. The practitioner’s hour was the constraint. The operation consumed attention daily, and that attention could not scale. If agents could handle intake, research, and drafting to a standard worth signing, the practitioner’s attention could concentrate where it could not be replaced: the decision to publish.
Decomposition split the work into four parts — scouting, research, production, and publication judgment — with gates between every stage. Two gates required human authority. The fourth part, publication judgment, was excluded from delegation before the experiment began. Not because the agent could not attempt it, but because the decision to attach a name to a piece of analysis was not delegable regardless of output quality.
No significant difficulty arose during isolation or decomposition. The boundaries followed the practitioner’s existing workflow. The four-part structure held through every subsequent version of the pipeline.
—
5. Verdict against prediction
Isolation and decomposition performed as working steps of the method. The act was testable in one sentence, the parts were distinct, and the predictions survived as honest priors against which the verdicts in the parent experiment were later measured. The overall prediction of Assisted proved correct.
—
6. Lesson
> Decomposition did not discover four equal parts — it discovered three the agent could carry and one the practitioner could not hand over.