Factimonious®

Software Evidence in a Minute

Evidence Chains

Link claim → observation → artifact → verification so engineering conclusions stay traceable end to end.

In one sentence

An evidence chain is an ordered link from a claim to the observations and artifacts that support it, and to the verification steps that can re-check those links.

Why it matters

Single artifacts mislead. A green CI job, a persuasive agent summary, or a commit message each look convincing alone. Chains force you to show how pieces connect—and where the chain breaks.

How it works

  • State the claim in one sentence.
  • List the observations that would have to be true for the claim to hold.
  • Attach concrete artifacts to each observation: commit SHA, diff hunk, test name, metric query, deploy record.
  • Name the verification step that re-runs or re-reads each link.
  • Stop the chain where a link is missing; do not paper over the gap with confidence language.

Example

Claim: “v2.4 fixes the checkout timeout.” Chain: failing production span → reproducing test on commit A → fix hunk in `checkout.ts` on commit B → test passes on B → canary deploy of B shows timeout rate drop. If the canary metric is missing, the chain ends at “fix exists and test passes,” not at “production is fixed.”

What this proves

A documented evidence chain proves which links were checked and which artifacts were cited for a specific claim.

What this does not prove

A long chain does not prove the claim is true in all environments. A complete-looking chain does not prove unlisted failure modes were considered.

Last reviewed 2026-09-09. Title for citation: Software Evidence in a Minute: Evidence Chains.