The Asymmetric Agency Framework (AAF)

A Structural Theory of Agency Constraints, Power, and Latent Instability

Christos Makiyama

January 2026

Version: AAF v1.0 (Canonical)

Canonical Abstract

Complex systems across human, organizational, and artificial domains often exhibit prolonged periods of apparent stability followed by rapid and irreversible failure. In many such cases, instability is detected only after observable degradation occurs, or explained retrospectively once collapse is unavoidable.

The Asymmetric Agency Framework (AAF) is a structural diagnostic theory of agency that explains why systems can already be on a path to failure while still appearing stable.

AAF does not model what agents want to do, optimize, or intend. Instead, it models what a system makes impossible, regardless of intelligence, motivation, or effort.

Agency is defined not by preference or rational choice, but by the constraints imposed by interacting capacities that scale, adapt, and decay asymmetrically over time. These capacities evolve with delayed and non-synchronous feedback and are shaped by historically accumulated constraint.

AAF’s core contribution is the explicit separation between Memory Capacity (R), defined as persistent, path-dependent historical constraint, and Interpretive Capacity (K), defined as active abstraction, understanding, and sense-making. By treating power and tooling as a distinct Instrumental Capacity (I), the framework explains why intelligence, optimization, and success frequently precede burnout, collapse, or systemic instability.

AAF is diagnostic, not prescriptive. It does not predict when failure will occur or propose corrective strategies. Its purpose is to identify latent structural instability before failure becomes observable.

What AAF Is — and Is Not

AAF is:

• A structural theory of agency

• A diagnostic framework for latent instability

• A model of constraint, not preference

• Applicable across individuals, organizations, artificial agents, and sociotechnical systems

AAF is not:

• A behavioral model

• A theory of intention or motivation

• An optimization framework

• A control or governance prescription

AAF operates at the level of possibility and impossibility, not decision-making or action selection.

Core Ontological Commitments

AAF is built on five foundational commitments:

1. Capacity-based agency

Agency emerges from interacting capacities, not from preferences or intentions.

2. Structural asymmetry

Capacities differ in scalability, responsiveness, and decay.

3. Temporal non-synchrony

Feedback between capacities is delayed and often exceeds regulatory horizons.

4. Path dependence

Historical constraints shape present agency independently of interpretation.

5. Non-reducibility

No single capacity explains agency in isolation.

The Five-Capacity Agent Model (P–M–R–K–I)

An agent is modeled as a configuration of five interacting capacities:

Physiological Capacity (P)

The energetic, biological, or operational ability to exist and function.This includes physical integrity, environmental viability, uptime, and basic sustainability.

Psychoregulatory Capacity (M)

The ability to regulate internal states such as stress, fear, motivation, attention, and impulse. It stabilizes behavior under uncertainty and sustained pressure. It is not intelligence or knowledge.

Memory Capacity (R)

Historically accumulated, path-dependent constraint formed through lived interaction with environments and systems.Memory Capacity is persistent, inertial, and largely non-erasable. It is not information, data, or knowledge.

Habits, trauma, institutional legacy, sunk costs, and ingrained routines belong here.

Interpretive Capacity (K)

The ability to form, manipulate, and apply abstractions, models, skills, and meaning. It enables understanding, learning, planning, and sense-making.

High Interpretive Capacity does not guarantee effective action when other constraints dominate.

Instrumental Capacity (I)

The ability to act upon the world through tools, resources, authority, and execution mechanisms. It includes money, infrastructure, automation, power, and coercive force.This capacity is highly scalable and frequently destabilizing.

Formal Definition of Agency

An agent A at time t is defined as:

A(t) = {P(t), M(t), R(t), K(t), I(t)}

Agency is the space of actions that remain possible given how these capacities constrain one another at a given moment.

Asymmetric Agency Condition

A system enters an asymmetric agency condition when at least two capacities scale or act faster than another capacity can regulate, absorb, or compensate.

Formally, an asymmetric condition exists when delayed inter-capacity differences exceed a system-specific stability threshold.

This condition identifies structural instability, even when no immediate failure is observable.

Latent Instability and Time

AAF treats time as a first-class variable.

Capacities evolve asynchronously.

Feedback arrives late.

Damage accumulates invisibly.

This explains why systems may appear stable while irreversible constraints are already forming, and why alarms often arrive too late to prevent collapse.

Structural Failure Modes (Illustrative)

AAF identifies recurring failure patterns, including:

• Interpretive Saturation

Understanding grows faster than memory and regulation can absorb.

• Instrumental Overreach

Power scales beyond physiological or psychoregulatory sustainability.

• Memory Lock-in

Historical constraint prevents adaptation despite high intelligence.

• Coercive Collapse

External pressure degrades core capacities faster than response is possible.

Scope and Epistemic Limits

AAF is intentionally diagnostic, not prescriptive.

It does not:

• prescribe interventions

• define optimal policies

• predict time-to-failure

• assume full observability or measurability

Several capacities are inherently difficult to quantify.

AAF adopts a pre-instrumentation stance: it defines which structural relations matter before specifying how they might be measured in particular domains.

Authorship, Usage, and Attribution

The Asymmetric Agency Framework (AAF) was introduced by Christos Makiyama in January 2026.

The framework is shared for open intellectual use, critique, and application under a Creative Commons Attribution 4.0 (CC BY 4.0) license.

You are free to use, extend, and apply AAF provided that:

• Proper attribution is maintained

• Core terminology is not redefined without explicit notice

• Derivative interpretations clearly distinguish original claims from extensions

Canonical Reference

This page is the authoritative canonical definition of AAF v1.0.

Subsequent essays, applications, and formal publications derive from this specification.