# ADDENDUM to PREREG_rushmore_round2 — the R7 network sampler (29 September 2026) Run 1 of rushmore2_crowds hung before printing anything (output was buffered), and no result exists from it. **Cause, a defect of mine.** My sampler redrew each network until it was STRONGLY connected. That needs every mind to be listened to by someone. With each of N minds listening to 4 others, about N·e⁻⁴ minds go unheard, so at N = 800 a strongly connected draw has probability about 4e-7, and the loop never ended. Strong connectivity is not what DeGroot needs. Consensus needs ONE closed class, aperiodic. Self-weights make every class aperiodic, and a mind nobody hears simply has influence π = 0. **The sampler now redraws only when eigenvalue 1 of W has multiplicity other than one.** No leg, quantity or criterion of R7–R10 changes.