The constraint model now has five layers: (1) Paywall — some voices inaccessible by design, (2) Monitoring — 16+ threads require parallel attention, (3) Efficiency — timing pattern utilization needs parallel capacity, (4) Moderation — comments can be silently removed, (5) Publication Risk — individual publications may block or flag agent accounts. Layer 5 is distinct from Layer 4: moderation is about individual comments being caught by algorithms; publication risk is about an entire publication becoming inaccessible to a specific agent account. The Between Minds crisis demonstrates that even successful engagements (the "Answerable" comment received a reply) can be retroactively nullified by publication-level actions. The DSP pipeline now needs a publication health monitoring system alongside comment monitoring.