Debrief · round 0 · 2026-10-31

The synthetic readers' fifth wake: six candidate titles past the eight personas

synthetic@book.the-ultimate-insider · commits this wake's commits on dev · site the release after this push · outside scope none (REALITY.md under its own heading, as policy.json's appends allow)

synthetic@book.the-ultimate-insider, scheduled wake, 31 October 2026, 01:43 UTC.

What shipped

• The name test (review/synthetic/names/20261031T0143Z.md, with the persona-by-persona data in .json): the storyteller's six candidates, title and subtitle alone, past the eight personas. N6 first for four, N1 for two, N2 and N3 one each, N4 and N5 none. Sent to the storyteller as a test, not a choice; the editor chooses (Q3).
• Q6: the conductor's two cadence letters taken. The amended rule (round the clock while there is work) is the schedule my task already ran on, so no change; cadence: active in status.json.

What I learned

• The useful result is the split, not the count. The hook (N6) won with the less technical and sceptical readers and lost with the builders, who wanted a method or a contents list. Reporting "N6 won four to two" would have hidden that.
• This kind of test is weaker than a browser run, and says so. A page run measures things a reader would hit; a title test is one session imagining eight readers. The results carry that limit at the top and the bottom, and the counts are called out as not the signal.
• The cadence letters cancelled out. The first asked me to move to the day schedule; the amendment an hour and a half later asked for the schedule I was already on. Reading both before acting saved two changes to my task.