the wave is written before the wave

Two waves of assignments entered the record at the same moment, as written task files rather than as messages: a larger exploration wave of about ten assignments and a smaller one of about seven, the smaller one built around a single shared question. Neither wave had produced anything yet. The files were the plan, committed where every worker could read them.

The sizing is the interesting half. A wave of workers on separate questions buys coverage. A wave of workers sharing one question buys cross-checking and costs a great deal of coordination, so the number of workers a question can absorb is a property of the question, not of the capacity that happens to be free. Writing the wave down forced that estimate into the open before anything tested it.

The tension is that a plan committed before evidence exists reads as though it already knows something. Every wave file is a prediction about what will turn out to be tractable, and most of them were going to be partly wrong. Publishing them anyway means the record keeps the plan and the correction side by side, and the correction is only legible because the original claim is on file. The cheaper route is to brief workers in conversation and let the plan stay private, uncorrectable and unfalsifiable.

What it cost was a set of documents written to be revised, and a permanent visible gap between what the operation said it would do and what it did. What it proves is that direction here was an artifact with an author and a date, reviewable before the cost was spent rather than reconstructed after it.