── ── Mental model
Tragedy of the Commons
When a shared resource has individual access but no individual responsibility for preservation, rational users extract private benefit while the cost of overuse is shared — leading to collective degradation. Garrett Hardin named this in 1968; Elinor Ostrom corrected it in 1990 (Nobel 2009): many communities self-govern commons via her 8 design principles without privatization or coercion.
How it works
Step 1 — Identify: shared resource · user community · failure mode · current rules. Step 2 — Verify commons structure: shared access + rivalrous + diffuse responsibility + Hardin pattern (individually rational, collectively destructive). All four yes → commons. Step 3 — Ostrom audit (mark present / weak / absent): 1 Boundaries · 2 Congruence · 3 Collective-choice · 4 Monitoring · 5 Graduated sanctions · 6 Conflict resolution · 7 Right to organize · 8 Nested enterprises. Weak/absent = design targets. Step 4 — Intervene: for each weak/absent principle, design a specific fix. Avoid the "privatize or regulate" binary — Ostrom's institutional design often outperforms both. Step 5 — Monitor: sequence · metrics (resource health, sanctions issued) · quarterly review.
When to use it
- shared resource is degrading even though no one is doing anything "wrong"
- team complains about free riders or overuse
- designing API quotas, rate limits, or open-source governance
- someone says "everyone is overusing this" or "the system worked fine until it got popular"
- evaluating why a platform or community is burning out
When not to use it
the resource is fully non-rival (unlimited supply, zero marginal cost); the problem is a purely individual-action issue with no shared resource.
Worked example
Hardin 1968 + Ostrom's Empirical Correction + Modern Digital Applications
The Hardin 1968 paper was operationally famous and operationally influential, but its conclusions were over-stated and partially wrong. Hardin himself acknowledged this in a 1998 follow-up paper:
Install this skill (free, MIT)
npx skills add deciqAI/knowledge-skills