Union Delta Review¶
Review ritual — instantiates Inclusive Membership Union Design
A version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.
A union is not built once; it is rebuilt as its sources change, and each rebuild can differ from the last. Union Delta Review is the recurring ritual of comparing the newest union against its predecessor and explaining every difference. Its central discipline is distinguishing a real change — a member genuinely added to or removed from a source — from an artifact, most often a source that went dark and dragged its members out of the union without anyone intending it. Its defining move is change attribution: for each added or dropped member it uses the provenance record to point at the source and version responsible, so a delta is never just a number but a caused, explained event. It works entirely in the time dimension — old version versus new — and takes no interest in the static shape of a single snapshot.
Example¶
A bank maintains a screening union of sanctioned and high-risk parties assembled from several official watchlists — a national sanctions list, an international list, and an internal risk register — rebuilt on each release. This week's rebuild shows the union shrank by 1,200 names. Read naively, that looks like 1,200 people were cleared. The delta review refuses that reading and attributes the change: the provenance timestamps show that 1,150 of the dropped names all trace to the international list, whose feed returned an unusually small file at 03:00 — a partial-load outage, not 1,150 real removals. The remaining 50 drops trace to a genuine national-list update and are legitimate. The review applies the absence policy: the outaged source is marked unavailable, its previous members are carried forward, and only the 50 real removals stand.
The outcome is that a dangerous artifact — a screening list that silently forgot 1,150 sanctioned parties — is caught and reversed, because the review demanded a cause for every drop instead of trusting the count.
How it works¶
- Diff against the prior version. Compute the set of added and dropped members between the previous published union and the new one; the delta, not the snapshot, is the object of review.
- Attribute every change to a source. Use provenance timestamps and versions to trace each add and drop to the specific source and update that caused it.
- Classify real vs. artifact. Separate genuine membership changes from source-availability artifacts — outages, partial loads, cadence gaps — using the absence policy.
- Adjudicate before publishing. Carry forward members lost to an outage, let real changes through, and record the decision, so each published union is a reviewed, explained increment over the last.
Tuning parameters¶
- Review cadence — every rebuild, or only when the delta exceeds a threshold. Reviewing every version catches small silent drifts; threshold-triggered review saves effort but can miss a slow bleed.
- Anomaly threshold — how large a per-source swing must be to demand investigation before publishing. A low threshold catches subtle outages but cries wolf; a high one only flags disasters.
- Outage response — what to do with members lost to an unavailable source: carry forward, freeze the whole union, or publish with the gap flagged. Carry-forward protects continuity but can mask a source that is genuinely shrinking.
- Attribution depth — whether every changed member is traced to a source, or only aggregate per-source deltas are reviewed. Per-member attribution is thorough but heavy; aggregate review is fast but can hide offsetting errors.
When it helps, and when it misleads¶
Its strength is that it makes the union trustworthy over time: changes are explained, not just observed, and the most dangerous failure — a source outage silently deleting members — is caught before it propagates. The statistical hazard it guards against has a name: data missing not at random, where an absence is systematically tied to a source failing rather than to a real change, so treating the gap as real bias the whole union.[n1]
Its failure mode is the mirror image of what it prevents: over-aggressive carry-forward. If every drop is reflexively treated as an outage and members are carried forward, the union stops tracking genuine removals and slowly fills with stale members that no source still vouches for. The classic misuse is reading a raw count delta as truth — "the list shrank, so membership fell" — without attribution, which is exactly the sanctioned-party bug above. A subtler trap is reviewing only aggregate deltas, letting a real add and an artifact drop cancel out to a deceptively small net change. The guarding discipline is to attribute every change to a source and version, distinguish outage from removal by policy, and keep carry-forward honest by expiring members that stay unsupported too long.
How it implements the components¶
null_and_empty_source_policy— applied in the time dimension: it decides, per delta, whether a source's missing members reflect a real removal or an outage, and what carry-forward that warrants.provenance_retention_record— consumed as the attribution key: version and timestamp stamps let each add and drop be traced to the source and release that caused it.
This review works version-over-version; it does not profile the static shape of a single snapshot — per-source coverage and overlap at one point in time (overlap_profile, downstream_union_scope_label) is Overlap and Coverage Dashboard, its nearest twin: the dashboard answers "what does the union look like now?", this review answers "what changed, and why?".
Related¶
- Instantiates: Inclusive Membership Union Design — governs how the union is versioned and how changes are explained over time.
- Consumes: Provenance Tagging Protocol supplies the version and timestamp stamps every change is attributed with.
- Sibling mechanisms: Canonical Identity Resolution Pass · Deduplicating Union Pass · Inclusive-OR Membership Test · Overlap and Coverage Dashboard · Provenance Tagging Protocol · Source Collection Ingestion Workflow · Type Compatibility Checklist · Union Specification Sheet
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Union Delta Review operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it a version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.
Independent corroboration: The frozen evidence defines Union Delta Review as 'A version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop', so its operative form is Assessment, Review & Assurance.
Nearest alternative: Analysis, Modeling & Optimization — Union Delta Review includes features of an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution, but its defining operation is a bounded evaluation of existing evidence or work that produces a finding or disposition.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Computer Science & Software Engineering
Origin pattern: Convergent development
Present-day reach: Universal
Rationale: W3C PROV-CONSTRAINTS documents that a provenance model represents entities, activities, generations, invalidations, and ordering constraints needed to compare changing unions. This is direct, mechanism-specific evidence for computer science as the best-evidenced historical home of the operation—A version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.—rather than evidence merely that the operation is useful there. The retained alternates record genuine adjacent lineages; later portability is represented separately by domain_reach=universal.
Related originating lineages:
- Engineering & Design — Engineering design, reliability, and systems-safety practice supplies a parallel or contributing lineage for the mechanism's defining operation: a version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.
- Library & Information Science — Library and information-science stewardship supplies a parallel or contributing lineage for the mechanism's defining operation: a version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.
- Mathematics — Mathematical modeling, proof, and abstract-structure practice supplies a parallel or contributing lineage for the mechanism's defining operation: a version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.
- Organizational & Management Science — Organizational Management supplies a historically relevant adjacent lineage or formative practice for the operation—A version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop.—but the adjudicated evidence more directly locates the defining lineage in computer science.
- Systems Thinking & Cybernetics — Systems science's feedback, boundaries, control, and regulation tradition contributes a separate formative lineage to the mechanism's union delta review logic.
Review resolution: The blind reviewers disagree on primary lineage (organizational_management versus computer_science). The defining operation is: A version-over-version review of the union that separates genuine membership change from source outages, using provenance timestamps to explain every add and drop. The researched W3C PROV-CONSTRAINTS establishes that a provenance model represents entities, activities, generations, invalidations, and ordering constraints needed to compare changing unions. That source therefore supports computer science as the historical origin. organizational management remains in the uncapped alternates where it contributes a formative practice, but application or governance is not itself proof of origin. origin_mode=convergent records lineage construction; domain_reach=universal separately records later applicability.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; medium confidence.
Sources consulted:
Notes¶
[n1] Data is missing not at random (MNAR) when whether a value is absent depends on the reason for its absence — here, members vanish because a source failed, not because they left the population. Treating such gaps as real changes biases every count, which is why the review attributes each drop to a cause before trusting it. ↩