Reappearance Association Protocol¶
Association protocol — instantiates Entity Persistence Across Observation Gaps
Decides whether a fresh sighting is the same entity that went dark — scoring it against an explicit identity criterion and abstaining into a monitored ambiguous hold rather than forcing an unsafe rebind.
When something reappears, the whole continuity effort comes down to one decision: is this the same entity that went dark, a different one that merely resembles it, or genuinely undecidable? Reappearance Association Protocol is the decision procedure for that moment. It scores a returning presentation against an explicit identity criterion — the stated evidence that would make two temporally separated sightings the same entity rather than similar — and it treats "I can't tell yet" as a first-class outcome, holding the candidate as an ambiguous, monitored hypothesis instead of forcing a match. Its defining commitment is the abstention path: because a false rebind and a false split rarely cost the same, the protocol refuses to convert weak evidence into a confident answer, and it keeps unresolved cases visible until discriminating evidence arrives rather than guessing and moving on.
Example¶
A marine biologist runs a manta-ray photo-ID program. Each ray carries a near-unique pattern of belly spots, and the study's continuity depends on recognizing an individual seen this season as one catalogued three years ago. A new photograph comes in. The Reappearance Association Protocol states the identity criterion up front — spot-pattern correspondence above a matcher score, consistent scarring, plausible size growth, a compatible location and season — and scores the candidate against every catalogued individual, not just the closest name.
The result is one of a few dispositions, never a silent overwrite. A strong, conflict-free match rebinds the sighting to the existing individual and inherits its whole history. A weak or contested score does not mint a confident new ray and does not force it onto the nearest match; it goes into a held, ambiguous state — a "possible re-sight of #A-114, needs a clearer belly shot" — that the protocol keeps on a watchlist. A genuinely novel, well-photographed ray with no plausible match is catalogued as a new individual, with predecessor/ambiguity links preserved if it was ever confused with another. The asymmetry is deliberate: wrongly merging two rays corrupts a life-history record in ways a cautious extra individual never does.[n1]
How it works¶
- State the identity criterion explicitly. Define, before matching, what evidence establishes sameness and what defeats it — features, trajectory, context, and disqualifying conflicts.
- Score against all candidates. Compare the returning presentation to every retained identity, producing a match score and surfacing conflicts (e.g., an incompatible simultaneous sighting elsewhere).
- Gate the disposition. Rebind on a strong, conflict-free pass; mint a new identity with predecessor links on a clear miss; and route everything in between to a hold.
- Monitor the residual ambiguity. Keep held hypotheses on a watchlist, request or await discriminating evidence, and resolve them later rather than closing them prematurely.
Tuning parameters¶
- Association threshold — how strong a match must be to rebind. Conservative thresholds cut false merges but raise false splits and holds; permissive ones do the reverse. This is the core false-merge / false-split dial.
- Merge/split cost asymmetry — how differently the two errors are priced, which shifts the threshold and the width of the abstention band. Steeper asymmetry widens the hold zone on the costlier side.
- Reversibility of a rebind — whether a match is provisional and revisable or an irreversible merge. Provisional links tolerate lower thresholds because mistakes can be undone; irreversible merges demand near-certainty.
- Hold-resolution policy — how long an ambiguous hypothesis stays open and what evidence closes it. Long holds preserve caution but accumulate unresolved cases; short ones force decisions that may be premature.
When it helps, and when it misleads¶
Its strength is exactly the high-stakes return where getting identity wrong is expensive and features are shared across entities. By making the identity criterion explicit and giving "undecided" a real home, it directly guards against both false rebind and false split, and it prevents the unsafe forced match — the archetype's insistence that abstention beat a confident error when a wrong merge would transfer history, permissions, or suspicion to the wrong entity.
Its failure mode is threshold miscalibration under look-alikes: set the bar too low and common features drive confident false merges; too high and legitimate returns fragment into duplicates. A subtler misuse is letting the abstention path rot — holds that are never revisited become their own backlog of permanent maybes, quietly as harmful as a wrong answer. And a matcher trained on one population can carry uneven error rates across groups, so a threshold that is safe on average can be unsafe for some. The guard is to keep both false-merge and false-split rates measured, keep provisional links reversible, and treat the hold list as work to be resolved, not a place to hide hard cases.
How it implements the components¶
reappearance_association_gate— it is the gate: it routes each returning presentation to rebind, hold, split, merge-after-review, or new identity, with the asymmetric harms exposed.identity_persistence_criterion— it states and applies the explicit test for when two separated presentations are the same entity, and what evidence defeats a match.residual_ambiguity_monitor— it keeps unresolved candidates on a watchlist and drives them toward discriminating evidence rather than closing them prematurely.
It does not reconcile competing reports from multiple observers, nor rank sources or preserve their provenance — reconciliation_record, source_authority, and provenance_record are Multi-Observer Sighting Reconciliation — and it does not propagate the entity's state through the gap before it returns; the observation_gap_state_model is Predictive State Filter. This protocol decides sameness on return; it neither fuses sources nor forecasts.
Related¶
- Instantiates: Entity Persistence Across Observation Gaps — supplies the evidence-gated, abstention-capable decision for reconnecting a return to a retained identity.
- Consumes: Dormant Entity Registry — matches the returning presentation against the retained dormant records it holds.
- Sibling mechanisms: Predictive State Filter · Dormant Entity Registry · Absence-Evidence Calibration Test · Multi-Observer Sighting Reconciliation · Persistent Identifier Resolver
Editorial Notes¶
Form Classification¶
Form family: Decision, Gate & Allocation
Rationale: Reappearance Association Protocol operates as a case-specific gate, selection, routing, prioritization, or resource disposition because it decides whether a fresh sighting is the same entity that went dark — scoring it against an explicit identity criterion and abstaining into a monitored ambiguous hold rather than forcing an unsafe rebind.
Independent corroboration: The frozen evidence defines Reappearance Association Protocol as 'Decides whether a fresh sighting is the same entity that went dark — scoring it against an explicit identity criterion and abstaining into a monitored ambiguous hold rather than forcing an unsafe rebind', so its operative form is Decision, Gate & Allocation.
Nearest alternative: Analysis, Modeling & Optimization — Reappearance Association Protocol includes features of an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution, but its defining operation is a case-specific gate, selection, routing, prioritization, or resource disposition.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Engineering & Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Deciding whether a new observation belongs to a pre-existing track after a gap is the engineering data-association problem; robotics and intelligence provide major operational specializations.
Related originating lineages:
- Robotics & Automation — Robotic perception materially developed track re-identification and abstention under ambiguity.
- Security Studies & Intelligence Analysis — Surveillance practice contributes identity criteria and monitored ambiguous holds.
Review resolution: The blind reviewers disagreed on primary lineage. Light authoritative research resolves the defining form in favor of engineering_design: Deciding whether a new observation belongs to a pre-existing track after a gap is the engineering data-association problem; robotics and intelligence provide major operational specializations. The rejected primary is retained only when it materially shaped the mechanism, and present-day breadth is recorded separately as domain_reach=specialized.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
[n1] Capture–recapture (mark–recapture, and its modern photo-identification form) is the ecological method of re-identifying individuals across sightings to estimate population and life history; its central data-quality risk is exactly the misidentification this protocol guards against — a wrong match silently corrupts an individual's record. Referenced as an established method, not a cited study. ↩