Exit criteria must be state+date pairs (observable condition + time horizon) — feelings in the moment are what the criteria exist to override

Why This Is a Rule

Exit criteria exist to override in-the-moment feelings — to tell you when to stop, pivot, or quit when your feelings would tell you otherwise. Feeling-based criteria ("I'll quit when it stops being fun" / "I'll exit if I feel like it's not working") fail precisely because the feelings the criteria are supposed to override are the same feelings evaluating the criteria. Sunk cost makes you feel invested. Status quo bias makes change feel risky. Optimism bias makes continuation feel promising. Under these feelings, no feeling-based criterion ever triggers.

State-plus-date pairs bypass feelings entirely: State (observable condition): a measurable situation that either holds or doesn't. "Revenue below $X" or "milestone not reached" or "three consecutive months of declining metrics." Date (time horizon): a specific date by which the state is evaluated. "If revenue is below $X by March 31" or "if the milestone isn't reached by Q2."

The combination creates a pre-committed, falsifiable trigger: on the specified date, check the specified condition. If the condition holds → the exit criterion is met, and the pre-committed response executes. In-the-moment feelings don't get a vote because the criterion was designed during calm deliberation (Design pre-commitments when calm to constrain behavior when stressed — never make rules in hot states) to override exactly those feelings.

When This Fires

• When designing exit criteria for any commitment, project, or strategy
• When a feeling-based exit criterion keeps not triggering despite deteriorating conditions
• When designing the "kill conditions" in a decision framework (Every decision framework needs five explicit components: criteria, sequence, time budget, kill conditions, and decision rights)
• Complements After irreversible commitments, schedule external reviews with pre-defined criteria — escalation of commitment corrupts self-assessment (external review with pre-defined criteria) with the state+date format specification

Common Failure Mode

Feeling-based exit criteria: "I'll quit the project if it doesn't feel right." It never "doesn't feel right" because sunk cost and identity attachment make continuation always feel more comfortable than quitting. Observable alternative: "If the project hasn't reached 100 users by September 30, I'll transition to the next project." On September 30, you check the number. 100 or not. Feelings are irrelevant.

The Protocol

(1) For any commitment with significant potential downside, define exit criteria as state+date pairs: State: "If [observable, measurable condition]..." — not "if it feels wrong" but "if [metric] is below [threshold]." Date: "...by [specific date]." The date creates the evaluation trigger. (2) Write both components and commit to them before entering the commitment. (3) On the specified date: evaluate the state condition. If the condition holds → the exit criterion is met. Execute the pre-committed response (exit, pivot, reduce). (4) Do not negotiate with the criterion at evaluation time. The criterion was designed to override your in-the-moment feelings. Trust the cold-state design (Design pre-commitments when calm to constrain behavior when stressed — never make rules in hot states). (5) If you genuinely believe the criterion is wrong → update it for the next evaluation period during a calm planning session. Never update criteria during the emotional pressure of the evaluation moment itself.