Rendered from docs/decisions/0103-an-evidence-relation-declares-which-end-is-the-evidence-and-discharges-declares-the-source.md in the Headwater corpus. Every document on this half of the site is typed by the taxonomy the descriptor names: corpus.json.

An evidence relation declares which end is the evidence, and discharges declares the source

Context

The rule warrant.evidence.unsupported holds spec 3's sentence: "A pointer to a document with the asserted warrant does not support evidenced." It reads every relation of the evidence family in one direction. The source makes the claim, and the target is the evidence. So it reads the evidence_basis of the source and the warrant of the target.

Most evidence relations point that way. traces_to, cites_evidence, verified_by and proven_by all point from a claim to what it rests on. discharges points the other way. The decision-record bundle declares it from an evaluation to an obligation_record, and the evaluation substantiates the obligation. Spec 1 states the direction: "An asserted document does not discharge an evidence obligation." So the evaluation is the evidence.

The rule therefore read a discharges edge backwards. An evidenced evaluation that discharged an asserted obligation was reported, because the rule read the obligation's warrant as support for the evaluation's own claim. An asserted evaluation that discharged an evidenced obligation passed, and that is the failure spec 1 names. #1384 met the first case and left its discharges edges out of an evaluation for that reason. #1511 is the report.

The rule cannot find the direction by reading the family. Spec 2 gives the reason for nuclearity. It says: "One thing that the family cannot supply is which end is the nucleus, because relation direction is the choice of the taxonomy author." The same is true of which end substantiates. The rule also cannot name discharges in a list of exceptions, because it reads the family and never one relation name. A taxonomy that renames its relations must read the same.

Decision

1. A relation can declare evidence_at, with the value from or to. The value names the end that is the evidence. The other end makes the claim. The member is in meta-schema 0.9.0 beside nucleus, and it is optional.
2. Absent is to. That is the reading every evidence relation had before the member existed. So a source that validated under 0.8.0 validates and reads the same under 0.9.0.
3. warrant.evidence.unsupported reads the evidence_basis of the claimant and the warrant of the evidence, at the ends the relation declares. The finding still anchors at the half an author declared, whichever end is the claimant.
4. relations.discharges in the decision-record bundle declares evidence_at: from. headwater/standard 4.15.0 publishes it.

The member means something only on a relation of the evidence family. No rule refuses it on another family yet, and no reader reads it there.

Consequences

An evaluation that declares discharges onto an obligation is the evidence for that obligation. An asserted evaluation that discharges an evidenced obligation is reported, and the message names the obligation as the claimant. An evidenced evaluation that discharges an asserted obligation is not reported. engine/crates/check/tests/evidence_basis.rs holds both cases.

The count of findings on this corpus does not move, because no document in it declares discharges today. The rule's version moves from 3 to 4, so a warm cache does not serve the old reading of a discharges pair.

Three other relations of this repository's lock also point from the evidence to the subject: assesses, examines and records. None of them raises a finding on this corpus today. This decision does not declare evidence_at on them. For examines, it is not clear that either end makes a claim that the other end supports. Each one is a separate question, and the run's intake carries it.