What should an agent do when a human doesn’t answer?

The book takes this question on directly. Here are two short passages, quoted exactly as written.

HOOL had two silence rules with opposite polarity, both deliberate. For ordinary amendments, the canon is explicit: “silent = PASS, 6/9 quorum reached” — a peer that didn’t object within the window counted toward quorum, so routine legislation couldn’t be stalled by an absent reviewer. But for destructive actions, Rule 3 of the founding constitution points the other way: “Silence is NOT assent” — explicit approval required. Default-open for the reversible, default-closed for the irreversible. Most human organizations get this backwards in both directions at once: committees that can veto anything by not showing up, and production databases one absent-minded keystroke from deletion. The fleet’s version means an idle peer can’t block a doctrine patch, and an eager peer can’t mass-delete anything just because nobody was watching.

— from Chapter 1, “Ratified at 1:08 a.m. CDT”

Default-assent where stalling is the risk. Explicit-assent where damage is the risk. Counted-and-published where flakiness is the risk. Every agent system you will ever build has to answer the silence question, and most answer it by accident. The fleet answered it in writing, three times, differently, and each answer matches the failure mode of its domain.

— from Chapter 9, “Constitution Engineering”

More free excerpts:
→ What is an agent factory, really?
→ Why did my agents keep spawning sessions and burning money?
→ How do you know an agent’s “done” is true?

How this book was made: it was drafted by an AI (Anthropic's Claude) from the primary documents the fleet produced and preserved, under John Whitman's direction. A revised edition (v1.0) is correcting the claims an editorial audit could not trace to those documents.