Skip to main content

DPCH10 — Addressable

“Provides Well-Defined Output Interfaces”

What DPCH10 is really asserting​

DPCH10 is not asserting that:

“There is an API somewhere” or “users can download data.”

It is asserting that:

A Data Product is addressable and consumable through explicit, well-defined output ports that match the needs of different consumers — human and machine — with experience layers built on top of those ports, not instead of them.

The product is accessed through ports, not through ad-hoc paths.

The Essence (HDIP + MSDH Interpretation)​

A Data Product satisfies DPCH10 if and only if:

  1. It exposes explicit output ports (APIs, query endpoints, streams, files)
  2. Those ports are consumer-appropriate, not one-size-fits-all
  3. Any UI or experience layer is a consumer of the ports, not the product itself

The product must be addressable independently of any UI.

Ports vs Interfaces (Critical Distinction)​

Output Ports (Product Contract)​

These are part of the Data Product:

  • REST / GraphQL APIs
  • SQL / federated query endpoints
  • Event or stream interfaces
  • File/object outputs (where appropriate)

They are:

  • versioned
  • documented
  • governed
  • policy-enforced

Experience Interfaces (Consumers)​

These are not the product:

  • dashboards
  • visual explorers
  • analytics notebooks
  • MSDH visual interfaces
  • marketplace viewers

They consume the ports to create experiences.

Comic World Analogy (Viz Landscape)​

  • Comic Product = Data Product
  • Asset Access APIs = Output Ports
  • Comic Reader UI = Consumer experience

The comic reader:

  • does not own the comic
  • does not embed the assets directly
  • consumes the comic via asset APIs

Likewise, a Data Product must not be “trapped” inside a UI.

Positive Criteria — When DPCH10 is met​

DPCH10 is met when all of the following are true:

1. Output ports are explicit and documented​

The product declares:

  • what ports it exposes
  • what each port is for
  • who it is intended for (humans, machines, both)

Examples:

  • API for applications and agents
  • SQL port for analysts
  • Search/query API for visual explorers

2. Ports are aligned to consumption needs​

Different consumers can:

  • choose the right port for their purpose
  • consume without adapting the product itself

Examples:

  • ML model consumes via API
  • analyst uses query interface
  • executive uses visual interface (via API)

This is multi-modal consumption done right.

3. Experience layers are decoupled​

UI tools:

  • do not contain business logic of the product
  • do not bypass governance
  • can be replaced without changing the product

The product remains stable even if experiences change.

Negative Criteria — When DPCH10 is not met​

DPCH10 is not met if any of the following are true:

❌ Product is accessible only through a UI​

Examples:

  • “Use this dashboard to access the data”
  • UI is the only access path
  • No API or query interface exists

This is an application, not a product.

❌ Consumers must reverse-engineer access​

Examples:

  • undocumented endpoints
  • ad-hoc credentials
  • access instructions via email or chat

This breaks self-service and governance.

❌ One port forced for all consumers​

Examples:

  • only SQL for everything
  • only file dumps
  • only REST regardless of use case

This leads to misuse and duplication.

Edge Cases (Important Guidance for Agents)​

Case 1: “UI exists and APIs exist behind it”​

⚠️ Partial

Rationale:

  • APIs exist, but may not be documented or supported as first-class ports
  • Often transitional

Case 2: “Multiple ports with clear contracts”​

✅ Met

Rationale:

  • product is addressable
  • consumers choose appropriate interface
  • experience layers remain optional

Case 3: “APIs exist but bypass governance”​

❌ Not met

Rationale:

  • ports must be policy-enforced (ties back to DPCH09)
  • raw access ≠ productized access

Evidence Signals an Agent Should Look For​

Authoritative evidence:

  • Product output port definitions
  • API specs / query contracts
  • Port documentation in catalog/marketplace

Supporting evidence:

  • Usage telemetry per port
  • Multiple consumption archetypes (CSxx)
  • Visual tools consuming the same ports

Red flags:

  • UI screenshots as “documentation”
  • ports known only to internal teams
  • direct DB access credentials

How an AI Agent Should Decide​

Decision rule (simplified):

If the Data Product cannot be consumed through explicit, documented output ports independently of any specific UI or tool, DPCH10 is not met.

Why DPCH10 Is Non-Negotiable​

Without DPCH10:

  • reuse becomes fragile
  • automation is impossible
  • AI agents cannot consume products safely
  • MSDH experiences become tightly coupled silos

DPCH10 is what turns a Data Product into a platform-grade asset.

Canonical Statement​

DPCH10 is satisfied only when a Data Product exposes explicit, governed output ports aligned to diverse consumer needs, with experience interfaces acting solely as consumers of those ports rather than as the product itself.