Reasoning reliably from knowledge under uncertainty — human, collective, and machine intelligence, and architectures that keep outputs calibrated to tier. Planned series; preview only.