Skip to content

Evidence Consumer Boundary

The evidence-book uses runtime outputs, but it is not the runtime itself.

That separation helps readers avoid a common mistake: using runtime usage docs to answer trust questions, or using study dossiers to answer execution questions.

flowchart TD
    A[What are you trying to learn?] --> B[How do I run this?]
    A --> C[How much should I trust this claim?]
    B --> D[Runtime handbook]
    C --> E[Evidence Book]

Use the runtime handbook when your question is operational. Use the evidence-book public guide when your question is about study reproduction, parity, freshness, or bounded trust.

Practical Reading Rule

Ask which question you are actually trying to answer:

  • runtime question What can the package do, how do I run it, and what are the public API and CLI contracts.
  • evidence question Which claims are benchmarked, parity-reviewed, study-backed, stale, bounded, or still open.

Keeping those questions separate is part of the public honesty model. The runtime can be broader than the currently closed evidence surface without that difference being hidden.

Why The Split Improves Clarity

  • runtime docs can stay practical instead of trying to embed every dossier detail
  • evidence pages can stay trust-focused instead of pretending to be the main product manual
  • readers can move between usage, validation, and bounded trust without one page trying to answer every question badly