Software

Software as a living system — and a pressure test of the Atlas ontology

Retained, not resolved

Ontology contradictions

The separated ontology arrived after some relations had already been declared with a single owned-by kind. Those edges are retained deliberately: reclassifying one would assert a relationship type nobody has declared, which is the exact inference the ontology forbids.

Contradictions

1

Needing attention

1

Legacy kinds

owned-by

Question status

under-test

Each edge, and why it stays

Flagged legacy edges

Endless-Technologies/Atlasowned-byLovable build agentattention

`owned-by` links the Atlas repository to the Lovable agent, but the separated ontology has no `owned-by` role: the edge is doing an unknown one of four different jobs.

Candidate kinds (none established)

legal-owner-of-assetparent-provenancegoverned-bydepends-on

Cannot reclassify because

The original assertion never recorded which sense of ownership it meant. Choosing `legal-owner-of-asset` would assert a legal fact about a build agent; choosing `parent-provenance` would assert that Atlas's origin chain runs through its toolchain. Both are inferences the ontology forbids, and the edge's own provenance cannot settle which was intended.

Would need

  • · A declared statement of what the original `owned-by` edge was asserting (asset ownership, provenance, governance, or dependency).
  • · A re-ontology decision on whether `owned-by` is retired, split, or retained as an explicitly ambiguous legacy kind.
  • · If legal ownership is intended: an assertion record naming the asset, the owner, and the evidence.

Risk if left: Readers may take `owned-by` as ownership of Atlas's identity by its build tooling — precisely the inference ONTOLOGY_WITHHELD refuses.

Governed by q.ownership-stewardship →

Append-only state history

Does Ownership differ fundamentally from Stewardship in software?

ownershipunder-testq.ownership-stewardship

Construction distinguished them clearly; software conflates them in Git.

  1. openinitial2026-08-08 00:00 UTC

    OPEN: ownership and stewardship were declared as one unresolved question; Git conflates them under repository ownership.

    Software Hub ontology pass — original question declaration.

    src/modules/software/registry.ts QUESTIONS (Release 1 compilation).

  2. under-testanswer-candidate2026-08-25 16:40 UTC

    ANSWER CANDIDATE: identity, parentage, stewardship and legal ownership are four distinct relationship types. The prior OPEN state is preserved, not replaced.

    Steward statement (Jarrod Cobb), ontology/provenance refinement request.

    src/modules/software/identity-ontology.ts · mem/doctrine/identity-parentage-stewardship-ownership.md

Answer candidate — unresolved items

  • · Git itself still exposes only one ownership notion; the four-way distinction is Atlas-side and cannot yet be read back from the host platform.
  • · `owned-by` remains on repo.atlas → agent.lovable from the earlier model; it is retained rather than silently rewritten, and its correct kind is not yet established.
  • · Whether stewardship should carry its own lifecycle ladder (assumed → active → handed over → withdrawn) is not decided.

Limits

  • · A contradiction listed here is retained deliberately; it is not a defect awaiting a quick fix.
  • · Reclassifying a legacy edge requires a declared statement of original intent, not a best guess.
  • · The governing question stays OPEN in its history even after an answer candidate is recorded.