The Monitoring Principle

In the AI Story

The intelligence commons monitoring challenge is structurally different. The resource flows are abstract, their degradation subtle, and the characteristic failure modes of AI-augmented work — fluent fabrication, concealed judgment failure — require deep domain expertise to detect. Invisible degradation is the monitoring problem specific to the intelligence commons.

Ostrom's framework suggests several directions. First, community-based monitoring must be prioritized over external monitoring, because the practitioners who work within the resource daily are the ones best positioned to assess its health. Code review processes that assess not just correctness but comprehension. Peer evaluation mechanisms that distinguish output reflecting genuine engagement from output that passes AI text through without critical examination. Mentoring relationships in which experienced practitioners systematically evaluate whether junior colleagues are developing genuine capability.

Second, organizations must invest in the time monitoring requires. The Berkeley study of AI-augmented workplaces found that production intensifies rather than reduces under AI adoption, with every available moment filled by additional tasks. Under these conditions, the time required for monitoring is the first casualty. The technology that most urgently requires monitoring also creates the conditions under which monitoring is least likely to occur.

Third, the gap between formal monitoring rules and actual monitoring practice — rules-in-form versus rules-in-use — must be tracked and addressed. An organization may have a formal policy requiring human review of AI-generated output; in practice, time pressure may reduce review to a cursory glance.

A 2025 study published in Artificial Intelligence formalized monitoring rules as part of governance architecture itself — embedded in the structure of agent interactions through what the researchers call an "Action Situation Language" — demonstrating that monitoring need not be an afterthought to governance but a constitutive feature of it.

Origin

The principle emerged from Ostrom's empirical finding that visibility of the resource condition was the single most reliable predictor of governance success across her comparative database. Its adaptation to AI governance requires rethinking what visibility means when the resource is abstract and its degradation invisible to surface inspection.

Key Ideas

Visibility as precondition. Without the capacity to observe the resource's condition, governance chains break at every subsequent link.

Community-based prioritization. Practitioners embedded in the resource have informational advantages no external monitor can replicate.

Time investment requirement. Monitoring requires the time that AI-augmented workplaces systematically eliminate.

Rules-in-use gap. The distinction between formal monitoring rules and actual monitoring practice must be tracked, because the gap erodes the entire governance system.