Not every `contains` edge is parthood

A bakery has a box that contains cupcakes. It also has a staff roster that includes decorators, a cleaning checklist that includes Monday’s closer, and a custom cake that includes three layers and a support board. Ordinary language uses the same kind of verb for all of those cases. Good systems should not.

The cupcakes are inside the box. The decorators belong to a team. The closer participates in a workflow for one shift. The cake layers are actual parts of the cake. Record all of those claims with one sloppy contains edge and the software starts making promises the business did not mean to make.

Hadto’s recent Keet study sharpened that point. Structural composition, containment, membership, participation, and constitution can sound close in casual speech while carrying different consequences for reasoning, reuse, and operator trust.

One generic edge collapses different claims

When a platform flattens many relation types into one inclusion-style edge, it quietly mixes together different kinds of commitment. A district can contain franchise units. A transaction can contain order lines. A staff member can belong to an organization. A person can participate in a workflow. A product can be constituted by ingredients or components. Those statements may look similar in a graph. They do not mean the same thing.

Some describe structure. Others describe location, group membership, temporary involvement, or material makeup. Treat them as interchangeable and the system starts inheriting the wrong assumptions. A rule that should apply to structural parts may spread onto things that merely sit in a container. Membership in an organization is not the same as being a physical component of it. Participation in an event should not inherit the same semantics as the parts of an object.

Operational systems feel the difference

Hadto is building systems meant to help turn employees into business owners. That only works when the operating picture can be handed to another person without hidden ambiguity. Ownership, responsibility, escalation, kit composition, workflow participation, geographic coverage, and reporting boundaries all depend on relation meaning.

If those edges blur, a system can stay formally tidy while becoming operationally misleading. A location-style relation can make a reporting rule spread farther than intended if the platform treats it like structural parthood. Membership modelled like composition can make the business look as though it permanently owns people or roles it only temporarily organizes. Participation data can imply a stronger dependency than the business actually has. Constitution handled like ordinary containment collapses material structure and commercial structure into one layer.

Relation families need names before they need automation

The point of the taxonomy is not academic neatness. It is to name a few relation families before the system starts automating on top of them. A business ontology should separate structural parthood for real components, containment or location for things that are in or within something else, membership for belonging to a group or organization, participation for taking part in an event or process, and constitution for what something is materially made from.

Each family supports different downstream behavior. Some relations should support transitive reasoning. Others should stay local context. Some justify rollups. Others describe a temporary situation. Reusable business infrastructure depends on making those distinctions explicit first.

Governance is the remaining job

The encouraging sign in Hadto’s internal ontology work is that the stack is already moving beyond one generic inclusion vocabulary. The current layer-zero relation surface distinguishes shared-part, location, participation, membership-like, and constitution-style families. The remaining gap is governance. Contributors still need a clear rule for when a vertical-specific relation should reuse one of those families, when it should stay local, and what kind of propagation or audit behavior is safe.

That is the operating standard: do not let a convenience edge stand in for a stronger claim. An owner-operator platform should make relation meaning trustworthy enough that people can hand off work without silently changing what the system says.

---

Source evidence used in this note: smb-ontology-platform/docs/plans/2026-03-31-keet-ontology-engineering-progress-tracker.md (2026-04-12 entries), smb-ontology-platform/docs/issues/ONT-017-add-foundational-ontology-contract-for-layer0-relations-and-attribution-modelling.md, smb-ontology-platform/docs/issues/ONT-007-add-rbox-compatibility-and-safe-property-chain-audit.md, and existing Hadto blog posts reviewed to avoid duplicating prior notes on foundational posture, semantic lifting, and RBox governance.