Skip to content
How to ThinkIn the Age of AI

Rank values through pairwise comparison — 'If I could only…

Force-rank values through pairwise comparison ('If I could only honor one of these two, which would I choose?') rather than attempting top-down global ranking, because pairwise comparison bypasses social desirability bias and reveals actual preferences.

When This Fires

Common Failure Mode

Global ranking with social desirability: "My top value is family, then health, then integrity, then kindness." This sounds noble and may be completely dishonest. The pairwise test reveals: "Family or career advancement — which would you actually choose?" If the honest answer is career advancement, your hierarchy needs updating to match your actual priorities.

The Protocol

(1) List your 5-7 core values (from Extract values from recurring conditions across 5+ peak…-601 discovery work). (2) For each pair, ask: "If I could only honor one of these two for the next year, which would I choose?" Be honest — the question is about your actual choice, not your ideal choice. (3) Record wins: each value gets a "win" for each comparison it wins. (4) Rank by total wins: the value with the most wins is your highest priority. (5) Check for intransitivities (circular preferences): if they appear, the involved values may be context-dependent rather than globally ranked (Agent priority orderings must be context-specific — which…). (6) Validate against past decisions (Validate your values hierarchy against 3 past decisions…): does this hierarchy predict the choices you actually made?