What is personal reliability engineering?

Personal reliability engineering is a concept in personal epistemology: Track how often each agent fires when it should and does not fire when it should not.

Example: Consider your morning exercise agent — the cognitive routine that gets you out of bed and into a workout. Over the past thirty days, there were twenty-two weekday mornings where the trigger condition was met (alarm at 6:00 AM, no illness, no travel). The agent fired on eighteen of those mornings. It failed to fire on four. It also fired on two weekend mornings when you had explicitly designated rest days — the agent activated when it should not have. Your reliability rate is 18/22 = 81.8%. Your false-fire rate is 2/8 = 25%. These two numbers together tell you something that 'I usually exercise' never could: your agent is moderately reliable but has a specificity problem. It does not discriminate well between trigger and non-trigger conditions. That specificity problem is where your optimization effort should go — not toward more motivation, but toward sharper trigger discrimination.

This concept is part of Phase 28 (Agent Monitoring) in the How to Think curriculum, which builds the epistemic infrastructure for agent monitoring.