Explore the Atlas

What kinds of alignment are there?

Think of these as lenses: different questions that can reveal different things about an apparently aligned system.

These forms are related, but none should be assumed from the others. The current view collects some especially useful distinctions for comparing how alignment works across systems.

States and targets

Different things about a system can be aligned.

Value alignment → Do components evaluate outcomes similarly?
Goal / intention alignment → Do components pursue the same outcome?
Epistemic / model alignment → Do components form compatible understandings?
Behavioral / plan alignment → Does behavior fit the relevant target or activity?
Functional / role alignment → Do differentiated activities contribute to a competent whole?
Representational alignment → Do components represent the right things?
Perceptual / relevance alignment → Do components notice what matters?

Relations and architecture

Alignment can live in relationships and structures, not just inside the parts.

Relational alignment → Does interaction make local activity answerable to the larger system?
Interoceptive / stress-sharing alignment → Can a problem in one part become part of what other parts must regulate?
Institutional/procedural alignment → Do rules, roles, rights, and procedures organize compatible activity?
Concordance → Do different actions or plans form mutually compatible trajectories?
Alignability → How readily can the system be redirected?

Time, scale, and change

Alignment also differs in what survives, changes, or emerges through time and across scales.

Achievement alignment → Can the system reach the desired state?
Maintenance alignment → Can it keep that state stable?
Corrective/robust alignment → Can it recover after deviation?
Regenerative alignment → Can it rebuild the machinery that produces alignment?
Developmental alignment → Does alignment change the participating parts?
Transformational alignment → Can the larger pattern itself change?
Compositional / multiscale alignment → Do aligned parts construct an aligned whole?
Ecological alignment → Does the agent–environment relation support viable activity?
Generative alignment → Can the system discover new targets over time?

The map stays open

This view is not meant to exhaust every possible kind of alignment. It is a set of useful lenses—enough to make some important differences visible and to make comparison possible. A distinction belongs here when it helps us notice something in the evidence. The point is not to fill out a zoo of types. It is to keep the distinctions that help us see.