Representational Topology in MCIR

MCIR representation is constitutional because structure determines which containment, sharing, authority, dependency, transformation, and lineage relations can be expressed and checked. This paper evaluates a specific candidate profile: a scoped tree for human authoring, a typed directed acyclic graph for canonical sharing and dependency, and a separately protected receipt-lineage graph for proposals, decisions, executions, and revisions. The comparison is not a contest between two unrelated data structures. A tree is a restricted DAG. The relevant differences arise from single-parent containment, reference and sharing law, edge typing, cycle policy, canonical identity, and mutation semantics. An AST does not automatically confer scoped authority or deterministic evaluation; a DAG does not automatically confer provenance, semantic equivalence, or institutional realism. Those properties require explicit law and evidence. The three-representation profile is promising because it separates local authorial structure, canonical reuse, and historical custody. It also creates new risks: compilation can lose scope or order, shared nodes can widen authority, canonical graph edits can diverge from the authoring source, and mutable lineage can rewrite the explanation of past effects. The architecture is therefore defined through correspondence, ownership, projection, identity, and receipt obligations. This work is narrower than the adjacent MCIR Topologies design-space paper. It develops one candidate configuration and its tests. Whether it remains a standalone companion, becomes a profile within the broader paper, or is integrated into the canonical MCIR work is an author decision; no merger is performed here.