Glossary¶
This glossary defines initial terms for Because Drift Happens. Definitions should become more precise as the framework evolves.
Drift¶
The movement of system behaviour, state, decisions, interpretation, or outcomes away from intended operating bounds.
Coherence¶
Regulated alignment across system intent, execution, feedback, and adaptation.
Operational Coherence¶
The ability of a running system to remain understandable, governable, accountable, and effective as conditions change.
Dynamic Stability¶
The condition where a system appears stable because corrective mechanisms continuously adjust, rebalance, compensate, or recalibrate it.
Autonomous System¶
A system that can make or execute decisions with some degree of independence from direct human instruction at each step.
Autonomous Drift¶
The master term for drift caused by autonomous or semi-autonomous actors diverging from original intent through local optimisation, partial context, changing incentives, tool effects, stale memory, delegation, or unobserved adaptation.
Agentic Drift¶
An AI-specific subterm beneath autonomous drift. Use agentic drift when the context is specifically about AI agents or trend-facing AI discourse.
Agent¶
An autonomous or semi-autonomous actor that interprets context, selects actions, uses tools, delegates tasks, or changes state in pursuit of an objective.
Multi-Agent System¶
A system composed of multiple agents that act independently or semi-independently while contributing to a larger process or outcome.
Federation¶
An operating model where multiple agents, services, teams, or systems retain local autonomy while participating in a shared governance structure.
Federated Drift¶
The compounding drift that emerges when multiple autonomous systems interact and local misalignments become systemic risk.
Control Plane¶
The layer that observes, coordinates, constrains, and intervenes above individual agents, workflows, and execution paths.
Governance Plane¶
The part of the control plane concerned with policy, authority, accountability, auditability, escalation, and constraint management.
Intervention¶
A deliberate action that pauses, redirects, rejects, rolls back, contains, or escalates an autonomous process.
Veto¶
A decision that prevents an action from proceeding because it violates policy, exceeds authority, lacks sufficient context, or threatens systemic integrity.
Runtime Governance¶
Governance that is evaluated and enforced while the system is operating, not only during design, review, or documentation.
Execution Entropy¶
The tendency for execution paths to become less predictable, less aligned, or less understandable as autonomy, delegation, and environmental change increase.
Alignment Decay¶
The gradual weakening of alignment between declared intent and actual system behaviour.
System Trajectory¶
The direction a system is moving over time, inferred from repeated observations rather than a single state snapshot.
Correction Latency¶
The time between drift emerging, being detected, being evaluated, and being corrected or escalated.
Hidden Micro Drift¶
Small internal deterioration that accumulates beneath acceptable high-level signals.
Coherence Boundary¶
A boundary around a system, process, agent group, domain, or organisation where coherence expectations, authority, and control responsibilities are defined.
Feedback Loop¶
A mechanism that returns information about actions, state changes, outcomes, or policy checks so the system can adjust.
Policy Checkpoint¶
A point in execution where policy, authority, risk, context, or constraints are evaluated before an action continues.
Systemic Integrity¶
The condition in which a system remains consistent with its purpose, constraints, authority model, and accountability requirements under change.