Revised Framework Center

Coherent agency under constraint

Alignment is the preservation of coherent agency under constraint. Healthy systems preserve the capacities required for meaningful participation, judgment, adaptation, refusal, responsibility, and action.

Adaptive systems depend on capacity-forming functions. Participation helps maintain those capacities. Support can scaffold participation or replace it; constraints can preserve usable agency or capture it. Authority determines who may legitimately bind action. Continuity concerns whether meaning and authority survive handoffs and time.

Misalignment can occur when participation becomes symbolic, capacity is substituted away, authority is detached from consequence, or capability and scale outrun the coherence required to govern them. The design problem can be compressed as: Alignment is the architecture that prevents scale from outrunning coherence.

The theoretical spine

1. Capacity
2. Participation
3. Support/substitution
4. Constraints
5. Agency
6. Authority
7. Continuity
8. Governance
9. Internalization

This sequence explains the concepts in reading order. It is a working account of their relationships, not a proven universal causal chain. The Current Framework Map places reality and constraints around the functions, participation, and authority they shape.

Capacity, participation, and support

A capacity-forming function is an activity or process that a person, group, or system must continue to perform, at least to some meaningful degree, in order to retain the ability to carry that function. Participatory capacity is the retained ability to remain meaningfully involved in the functions that shape judgment, responsibility, adaptation, learning, and action.

The support/substitution mechanism remains central: a relation can preserve immediate function while eroding the participation that maintains longer-term capacity. The same tool may support one person or task and substitute in another. Capacity, task design, reversibility, frequency, and authority affect the relation.

Constitutive co-regulation

Support that belongs to healthy functioning itself and preserves participation through shared regulation.

Developmental scaffolding

Temporary support directed toward stronger future participation.

Stable distributed competence

Durable shared competence across people, tools, and environments that preserves the participant's role.

Substitutive dependence

Support that preserves output while reducing participation in a capacity-forming function over time.

Continuing support can be healthy. The framework does not require isolated self-sufficiency. See the four modes of support and participation, co-regulation, and substitution.

Constraints and agency

Physical limits, cognitive limits, timing, rules, permissions, and institutions shape the action space. Legitimate boundaries can make that space safer and more intelligible. Excessive, deceptive, arbitrary, or illegitimate constraints can capture agency.

Agency is the usable capacity to understand, judge, refuse, revise, choose, direct action, learn from consequence, and remain responsible within a system that affects real outcomes. Nominal options or an approval click do not establish that these capacities remain usable.

Constraint, Agency, and Alignment sets out the questions for examining reality-grounded purpose, proportionate restriction, valid authority, challenge, revision, and capacity development. These questions cannot automatically determine moral legitimacy.

Authority and true gates

Authority determines who may legitimately impose constraints, delegate scope, approve risk, or bind action. It should be inspectable: who may approve what, under which conditions, for how long, with what revocation path?

A true gate is a consequential decision boundary at which rightful authority can govern passage by halting, revising, refusing, approving, or escalating before consequence. A false gate blocks or permits without rightful authority, meaningful power to change the outcome, or reality-grounded conditions. An effective block can still be illegitimate; a human signature can still be powerless.

Continuity across handoffs and time

Semantic continuity preserves meaning, source lineage, context, uncertainty, and transformation history. Downstream actors should be able to distinguish a source from an interpretation of it.

Runtime continuity preserves authorization from permit to execution: the action that executes remains the authorized action. Fluency does not establish source fidelity, and permission for one branch does not authorize a changed branch.

The cross-cutting Handoff Integrity requirement carries source lineage, validated context, authority scope, transformation history, expiration conditions, and proof across transfers. Alignment is therefore a trajectory problem as well as a question about a system at one moment.

Governance, internalization, and maturation

Healthy governance preserves or restores the conditions of meaningful agency while keeping action within legitimate boundaries. External checks can be necessary during learning, uncertainty, or high risk. Internalization is the maturation process by which initially external constraints become increasingly carried through integrated understanding, skill, judgment, and governed habit.

The proposed direction is to detect disorder, restrain unsafe action, explain patterns, improve proposals and judgment, internalize governance, lighten unnecessary gating, and preserve proof. Reduced friction is justified where maturity increases; it is not an instruction to remove legitimate permanent constraints or necessary co-regulation.

HAPI's Agency Maturity Model describes this human and institutional trajectory. In AGS, Governance Memory analyzes receipt history and recommends changes for human review. It cannot silently change policy, boundaries, or authority.

HAPI and AGS inherit from Alignment Theory

HAPI applies the theory to human systems and institutions. It asks whether constraints preserve understanding, meaningful refusal, revision, live authority, and accountability, and whether governance restores capacity.

AGS applies the theory to delegated operational agency: consequential action under borrowed human or organizational authority. AI agents are not treated as moral persons. Responsibility remains with the humans and organizations delegating action.

The ten-node trajectory-governance spine locates PGDL, AAG, Runtime Binding, Receipts, and Governance Memory within the authority, deliberative, enforcement, evidence, and operator planes. The Continuity Console is the operator surface.

Working claims and limits

This synthesis is a working conceptual and design framework, not a peer-reviewed or universally established result. Similar patterns across domains do not make those systems equivalent. The mechanisms and evidence must be examined in each domain.

Support without capacity loss, reduced participation with durable gains, or the inability to distinguish scaffolding from substitution can count against an application of the mechanism. The boundary conditions and stress tests retain the earlier research record. New claims about constraint legitimacy, continuity, and maturation need their own operational criteria and evidence.

Independent convergence supplies comparisons to test, not proof of ownership, empirical validation, or technical validation of AGS.

Biblical and philosophical interpretation

The biblical/Babel layer is a philosophical and theological mapping. In this interpretation, Babel represents capability, coordination, and scale detached from an order adequate to govern them. The proposed anti-Babel relation puts power under discernment, agency under authority, delegation under accountability, technology under stewardship, and execution under proof.

The restoration theme relates external law to the formation of internal capacity. These readings do not supply scientific proof of the framework or its software. Readers can evaluate the conceptual theory without accepting theological commitments. See where reasoning and theological commitment differ and the historical Biblical Grammar.

Development and provenance

The earlier internal/external model examined the difference between integrated regulation and imposed compliance. The May 6, 2026 framework center refined participation and the four modes of support. This September 14 synthesis retains that mechanism within the broader account of coherent agency under constraint.

Source: Michael Nathan Bower, Alignment Theory: From Support/Substitution to Coherent Agency Under Constraint, canonical conceptual handoff, September 14, 2026. Earlier papers and diagrams remain in the Archive. See current definitions and the canonical Markdown summary.