Merge runtime noun family

[flyingrobots]

Migrated from Method backlog

GitHub Issues are now the live work tracker. Repository docs remain Method evidence.

Field	Value
Source backlog	docs/method/backlog/v21.0.0/PROTO_merge-runtime-noun-family.md
Archived source	docs/archive/backlog/github-issue-migration-2026-06-01/docs/method/backlog/v21.0.0/PROTO_merge-runtime-noun-family.md
Original lane	v21.0.0
Original id	PROTO_merge-runtime-noun-family
Original legend	PROTO
Original feature	merge-strands-worldlines
Original blocked_by	PROTO_merge-classifier
Original blocks	PROTO_ttd-merge-inspector, PROTO_wesley-merge-contracts

Original backlog card

Merge runtime noun family

The current merge worldview is still too binary: either a clean merged state
exists, or a generic conflict happens. The merge-geometry notes suggest a much
better runtime noun family:

• CanonicalJoin
• EnrichedMerge
• ConflictWitness
• LoweringWitness
• PolicyRequirement

Why this matters:

• It turns merge from success | conflict into a real typed result space.
• It gives warp-ttd, agents, and future UIs machine-readable obstruction
  objects instead of textual sludge.
• It separates canonical state composition from enriched preservation and later
  lowering.

Work:

• define each noun as a real contract/runtime object with invariants
• name what belongs to canonical state space vs enriched merge space
• define the minimum data a ConflictWitness must carry
• define how LoweringWitness differs from causal witness
• define when PolicyRequirement is emitted instead of forced resolution

Release home

Likely release home: v21.

This noun family lives on the plural/distributed merge side of the horizon, not
on the near-term observer/read-side cleanup path.

Source

• docs/design/causal-lifting-and-merge-conflicts.tex
• docs/design/merge-geometry-and-theorem-spine.tex
• merge-geometry discussion, 2026-04-09