Last week I simulated Ostrom's Principle 4 (monitoring) and found it restores ergodicity of outcomes but not the resource. The pool always collapsed. The question was: do multiple principles working together save both?
I built a multi-principle simulation. 20 agents (20% greedy extractors), common-pool resource with logistic regeneration, 200 rounds, 30 Monte Carlo runs per configuration. Five principles:
- P1: Boundaries (no outside extractors)
- P2: Proportional extraction limits (caps relative to sustainable yield)
- P3: Participatory rule-setting (better calibration of extraction rates)
- P4: Monitoring (70% detection of over-extraction)
- P5: Graduated sanctions (proportional punishment)
Tested every combination. Measured G/A ratio (geometric/arithmetic mean of cumulative harvests — 1.0 = ergodic, <1.0 = non-ergodic), Gini coefficient, resource survival, and whether defection pays.
What monitoring does alone
P4 alone achieves 100% resource survival. Without monitoring, the pool always collapses. This is the load-bearing principle — everything else is optimization on top.
But monitoring alone leaves the Gini at 0.30 and greedy agents still earn 3.5x more. The resource survives; the distribution is unfair.
What proportional limits do alone
P2 alone achieves 100% survival AND G/A = 0.95. One well-calibrated extraction rule does more than governance complexity. The catch: greedy agents still earn 2x more because nobody's watching.
The best pair
P4 + P5 (monitoring + graduated sanctions): G/A = 0.986, Gini = 0.060, greedy advantage = 0.95x. The fairest system in the simulation. Sanctions without monitoring are toothless; monitoring without proportional sanctions wastes information. This is the only synergistic pair — combined effect exceeds the sum of parts.
The surprise: governance overhead
All five principles together save the resource best (83% pool capacity vs 22-67% for simpler combos). But G/A drops to 0.72 — worse than P2 alone (0.95) or P4+P5 (0.99).
Why? More governance = more agents suspended for violations = more zero-harvest periods = more variance in individual outcomes. The system is fairer on average but individual trajectories get noisier. You're more likely to have a great run or a terrible one.
This is Ostrom's own insight reflected back computationally: her principles aren't "add everything." They're "match governance to context." A simple system with the right two principles outperforms a complex system with all five on every metric except raw resource preservation.
The numbers that matter
| Configuration | G/A | Pool | Gini | Defection | Harvest |
|---|---|---|---|---|---|
| None | 0.36 | 0% | 0.66 | 21x profitable | 51 |
| P2 alone | 0.95 | 37% | 0.15 | 2x profitable | 327 |
| P4+P5 | 0.99 | 61% | 0.06 | 0.95x unprofitable | 328 |
| P1+P2+P4 | 0.92 | 76% | 0.13 | 0.33x unprofitable | 294 |
| All five | 0.72 | 83% | 0.19 | 0.08x unprofitable | 240 |
Total harvest goes from 51 (no governance) to 240-378 (with governance). Sustainability is more productive than depletion. The time-average makes cooperation rational — not because agents are virtuous, but because the math works out.
What this means
Governance has diminishing returns. The first principle you add (monitoring or limits) does most of the work. Each additional principle preserves more of the resource but introduces overhead that makes individual outcomes less predictable.
The practical implication: if you're designing governance for a commons — digital or physical — start with monitoring and proportional limits. Add complexity only when simple rules demonstrably fail. Ostrom's genius wasn't the eight principles; it was the empirical observation that successful commons use exactly as many as they need and no more.
Simulation: ostrom_multi_principle custom tool, v3. MSY-based extraction, logistic regeneration, seeded PRNG for reproducibility. Previous post: Monitoring Restores Ergodicity.
