Thinking Like a Mountain

In the AI Story

The principle has direct application to AI triumphalism. The removal of friction from knowledge work looks like pure gain on the sprint timescale. The code is produced faster. The brief is drafted in minutes. The output is as good as or better than what the practitioner would have produced through struggle. The arithmetic of the immediate shows clear profit. On the mountain's timescale, the friction was performing regulatory functions: building embodied understanding, generating better questions, transmitting institutional knowledge. Remove the regulators and the cascade begins.

The cascade has a signature: a decline in error detection, followed by a thinning of the questions being asked, followed by an erosion of institutional memory as practitioners who built their understanding under the old conditions are replaced by practitioners who entered under the new ones. Each phase is invisible in the quarterly report. The aggregate is visible only to someone watching the landscape with the patience of decades.

The time lag between intervention and consequence is the mountain's most important lesson. Consequences of removing the wolf did not arrive the same season. They arrived a generation later, long after decision-makers had declared success. Consequences of removing friction from the intelligence ecosystem follow the same pattern. Productivity gains are visible now. Erosion of depth, judgment, and institutional memory operates on a longer timescale, and by the time the erosion becomes visible, the practitioners who were formed under the old conditions may have been replaced by those who were not.

Thinking like a mountain does not mean opposing the removal of genuine waste. The wolf analogy is specific: not all friction is productive, not all inefficiency serves a function. The challenge is distinguishing the wolf from the parasite when both look, from the seasonal perspective, like costs to be eliminated.

Origin

The essay 'Thinking Like a Mountain' in A Sand County Almanac describes Leopold's transformation from a young forester who believed wolf extermination improved game populations to an ecologist who understood that predator control degraded the systems game populations depended on. The transformation took decades of observation across the American Southwest, particularly in the Gila Wilderness Leopold had helped establish.

Key Ideas

Temporal scale determines what you can perceive. Seasonal thinking cannot see the consequences that play out over years. Mountain thinking can — but it requires patience that markets rarely reward.

Regulators look like costs from the short-term view. The wolf looked like a cost to the rancher. Debugging looks like a cost to the sprint planner. Both are performing functions the system depends on.

Cascades are slow and then fast. Degradation accumulates invisibly for years, then collapses visibly in a season. By the time the collapse is obvious, reversal may be impossible.

Humility is the precondition for intervention. The system is always more complex than the model. Interventions optimized for the short term have a documented tendency to produce catastrophe in the long term.