Measure behavioral agent progress by displacement rate, not perfection — replacement is gradual, not binary
Track agent displacement by measuring the percentage of times your designed agent fires instead of the default, not by whether you execute perfectly every time, because replacement is gradual and competes against thousands of prior reinforcements.
Why This Is a Rule
When you design a new behavioral agent to replace a default behavior — "When I feel the urge to check email, I'll write for 5 minutes instead" — you're competing against thousands of prior reinforcements of the default. The email-checking behavior has been rewarded with novelty and social validation thousands of times. Your new writing agent has been reinforced zero times. Expecting perfect execution from day one is asking a new recruit to outperform a veteran.
The displacement rate — what percentage of trigger instances fire the new agent versus the default — is the correct progress metric. Going from 0% to 20% in week one, then to 40% by week three, then to 70% by month two represents genuine progress even though you're "failing" 30% of the time. The trajectory matters, not the current score.
Perfection-based tracking ("did I do it every time?") produces binary failure: you succeeded perfectly or you failed. This creates a demoralization spiral where any slip "ruins" the streak, destroying motivation. Displacement-based tracking produces gradient progress: every firing of the new agent is a data point of improvement, and slips reduce the percentage but don't erase the trend.
When This Fires
- When evaluating whether a new behavioral agent is working
- When you feel discouraged because a habit "isn't sticking" — check displacement rate rather than perfection
- During weekly agent reviews (Review new agents weekly, established ones monthly, and all agents after major context changes) when assessing progress
- When coaching others on behavior change and they report "failure"
Common Failure Mode
Using streak-based tracking: "I did it 12 days in a row, then missed a day — back to zero." Streaks create fragile motivation structures that collapse on first failure. Displacement rate is resilient: a missed day changes the percentage slightly but doesn't erase progress. It also reveals trends: 60% this week vs. 40% last week vs. 20% the week before shows acceleration even though you're far from 100%.
The Protocol
(1) For each active behavioral agent, track two numbers per week: total trigger instances (how many times the trigger condition occurred) and new agent firings (how many times you executed the designed response). (2) Calculate displacement rate: firings ÷ trigger instances. (3) Track the trend over weeks. Rising displacement rate = the agent is installing. Flat rate = something needs adjustment (see Diagnose failing behavioral agents by component — trigger salience, condition scope, or action effort each require different fixes for diagnosis). Declining rate = the agent is losing to the default. (4) Celebrate displacement improvement, not perfection. Going from 30% to 50% is significant progress that deserves reinforcement. (5) Target 80%+ displacement before considering the agent "installed" — 100% is aspirational, not required.