Boundary

A boundary is an operational distinction between a system and environment that matters because causal relations differ across that distinction.

A boundary is not necessarily one geometric line, and it should not be understood as a sealed wall.

Connection Theory does not assume that every system has one uniquely privileged boundary. The analysis declares a system and a class of relations; whether a corresponding causal inside–outside differentiation exists, where it lies, and how strongly it matters are empirical questions.

The frame is declared by the analysis; the causal structure is constrained by reality.

Frame-relative therefore does not mean observer-arbitrary. A useful analysis cannot move the boundary after seeing the result merely to rescue an attribution.

Different processes can define partially overlapping boundaries for:

matter and energy;
information and signals;
control and authority;
membership and access;
resources and capital;
regulation, responsibility, and observation.

The important question is not simply whether something crosses the boundary, but what crosses, through which interfaces, under what conditions, and with what causal consequences for the specified system.

Thus:

open system ≠ boundaryless system

Causal boundary, regulated boundary, and boundary sovereignty

These should not be collapsed.

Causal boundary — an empirically distinguishable inside–outside difference for a declared relation class.

Regulated boundary — a boundary across which relevant flows are selectively regulated.

Organizational boundary sovereignty — the stronger capacity of an organization to determine what may cross, under what conditions, and with what consequences for internal state, membership, authority, or access.

A general causal boundary does not have to possess organizational sovereignty. In persistent biological, artificial, and social organizations, however, selective regulation of important cross-boundary flows is often part of what allows the organization to persist.

A simple artificial-system example

Suppose an analyzed AI organization is declared to include several models, a shared working state, and a specified set of permitted tools. Users, unrestricted external services, and databases not included in that declaration remain outside the analyzed system. API credentials, tool permissions, authenticated interfaces, and data-access rules can then implement regulated cross-boundary paths: they determine which information or actions may cross and under what conditions. If the organization can itself select, route, accept, reject, or constrain some of those crossings within its externally granted authority, that is a stronger boundary-control property.

This is a causal-role comparison, not a claim that an API permission is physically the same mechanism as a cell membrane. The invariant question is whether a declared inside–outside distinction is real for the relation being studied and whether crossing it changes the system through regulated interfaces.

Legal or institutional boundaries should be treated the same way. A legal title, ownership label, or contractual category matters to causal organization only insofar as it is instantiated through effective authority, enforcement, access, control, obligation, or constrained flows. Legal description alone is not automatically a causal boundary.

Boundary and interface remain distinct

A boundary identifies the causal inside–outside differentiation being studied.

An interface is a material, informational, institutional, or computational pathway through which exchange across that distinction is implemented.

A single boundary function may therefore be distributed across several interfaces, and one interface can participate in several boundary relations.

Boundary control and organizational persistence

For a persistent organization, boundary strength is not measured by impermeability. It is measured by whether cross-boundary flows remain sufficiently selective and regulated for the organization to preserve its relevant internal relations.

A company can remain highly open to customers, employees, suppliers, information, and capital while retaining strong organizational boundary sovereignty. It loses boundary integrity when external flows can arbitrarily rewrite internal roles, authority, assets, or information pathways without passing through the organization's own control processes.

Boundary failure can become organizational failure

A possible failure sequence is:

loss of selective boundary control → loss of internal state regulation → disruption of internal relations → organizational collapse

This does not imply that every organizational failure begins at a spatial surface. The relevant boundary may concern information, authority, membership, matter, or another causal flow.

A useful boundary claim should therefore specify:

the system / individuation frame;
the relation class being separated;
the relevant cross-boundary flows;
whether and how those flows are regulated;
the interfaces through which crossing occurs;
the observable causal difference or failure expected if the boundary relation changes.