Legibility and Control

Making relationships explicit—creating legibility—is a control technology. This framework does it: map dependencies, formalize interfaces, quantify metrics. But legibility has a dark side: it can destroy the informal, adaptive systems that actually work.

The Concept

James C. Scott’s Seeing Like a State (1998) describes how states simplify:

Before: Local names, customary rights, variable measures, oral tradition After: Standardized surnames, written records, uniform measures, formal rules

States couldn’t tax what they couldn’t count. Solution: Make society legible through standardization.

When Legibility Fails

Scott identifies high modernism—the belief that scientific planning can optimize society, that traditional practices are inefficient superstition, that experts know better than locals.

High modernism + state power + weak civil society = disasters:

Project	What Happened
Scientific forestry (Prussia)	Monoculture plantations. Initial success, then massive die-offs. Lacked biodiversity that sustained forest health.
Collectivization (Soviet Union)	“Scientific” agriculture ignored local knowledge. Result: famines.
Brasília (Brazil)	Planned city separated functions by design. Result: sterile, alienating; informal settlements grew around it.

Pattern: Replace functioning complexity with rational simplicity → system fails → informal complexity re-emerges.

Métis: Knowledge That Can’t Be Formalized

Métis = practical, local, tacit knowledge:

A farmer’s sense of when to plant (beyond weather data)
A nurse’s intuition something is wrong (before tests confirm)
A trader’s feel for the market (beyond models)

The irony: The more legible the system, the more it depends on illegible métis to function. Hospitals run on nurses’ tacit knowledge, not just protocols. Remove the métis, and the legible system collapses.

Implications for AI and Entanglement

When Legibility Helps

Hidden dependencies cause cascading failures → make them explicit
Informal arrangements enable hidden coordination → formalize interfaces
Standardization enables substitution and competition
Measurement enables improvement

When Legibility Hurts

Legibility Project	Destroyed Métis	Result
Strict API contracts	Informal flexibility for edge cases	Brittle systems
Mandatory documentation	Quick informal knowledge sharing	Slow, outdated docs
Complete isolation	Beneficial implicit coordination	Coordination failures

AI Has Métis Too

Model tacit knowledge: Learned patterns that work but aren’t understood
Operator tacit knowledge: “This prompt works better even though I can’t say why”
System tacit knowledge: How components actually interact vs. how they’re designed to

Finding the Balance

Principles:

Preserve what works — Before standardizing, understand why the current system works
Gradual legibility — Incremental formalization allows learning
Escape valves — Formal systems need room for informal adaptation
Monitor for brittleness — Highly legible systems often fail in novel situations

The meta-principle: The drive to make things legible is itself a bias. Sometimes the right answer is: “This is working; don’t formalize it.”

Key Takeaways

Legibility is a control technology — We standardize to reduce entanglement risk, like states standardize to tax
High modernism fails — Replacing functioning complexity with rational simplicity causes disasters
Métis matters — Tacit knowledge often carries crucial understanding that can’t be formalized
The paradox — The more legible the system, the more it depends on illegible métis
Balance is key — Make explicit what needs to be explicit; preserve what works