Adverse-Precedent Search Protocol¶
Protocol — instantiates Precedent-Guided Decision Governance
Requires search for cases that limit, distinguish, conflict with, or undermine the proposed precedent treatment before a decision is finalized.
A decision maker who searches only for cases that agree with the answer they already prefer can quietly define the case universe to fit that answer. The Adverse-Precedent Search Protocol is the standing discipline that forbids this: before a precedent treatment is finalized, the record must show that someone actively hunted for the cases that cut the other way — the ones that limit, distinguish, conflict with, or flatly contradict the proposed line. Its defining move is that coverage of the unwelcome precedent, not the supporting one, is what the protocol certifies. A search that returns only confirming authority has failed the protocol even if every case it found is real, on-point, and correctly cited. The output is not a decision and not a comparison; it is a certified sweep — a documented claim that the adverse universe was looked for and either found or genuinely absent.
Example¶
A patent examiner is about to allow a claim on a new battery-electrode coating, leaning on a family of earlier allowed applications with similar chemistry. Under the office's adverse-search protocol, allowance cannot issue on the strength of those supportive files alone. The examiner must run and log a second, differently-shaped search aimed at what would defeat the claim: prior-art references disclosing the same coating, later office actions that narrowed the earlier allowances, and any reexamination that clouded them. The protocol fixes the query classes (classification codes, keyword-plus-synonym expansions, non-patent literature) and requires the examiner to note each database searched, each filter applied, and — critically — each promising hit that was reviewed and set aside, with a one-line reason.
The sweep turns up a conference disclosure predating the whole supportive family. It does not decide anything; the examiner still has to judge whether that reference is truly on point. But it changes what the file can honestly claim: the allowance, if it issues, now rests on precedent that survived an adversarial look rather than on a search rigged for agreement.
How it works¶
The protocol is a required second pass with its own success criterion. What distinguishes it from ordinary retrieval:
- Adverse query classes are mandatory, not optional. Alongside "cases like mine," the searcher must run "cases that limit mine," "cases that reach the opposite result," and "later treatment of my supporting cases." Absence must be affirmatively recorded, not inferred from silence.
- Coverage is logged, not just results. Databases searched, filters and exclusions applied, date ranges, and the ranking tool used are all part of the record, so a reviewer can see what the search could not have found.
- Promising-but-rejected hits are documented. Each adverse case that was surfaced and then set aside gets a reason, converting a silent omission into a reviewable judgment.
- The automated ranker is treated as a suspect, not an oracle. Where a similarity engine orders the hits, the protocol requires a check that its ranking is not silently burying adverse results below the fold.
Tuning parameters¶
- Search breadth vs. searcher time — how many adverse query classes and databases are mandatory. Wider sweeps catch more limiting authority but cost hours; scale breadth to the decision's stakes and reversibility.
- Exclusion transparency — how aggressively filters may prune results, and how much of that pruning must be justified in the log. Tight filters speed review but can hide the very case that matters.
- Rejected-hit threshold — how strong an adverse match must be before a written reason is required. A low threshold documents more judgment but bloats the record.
- Ranker-audit depth — how far past the top results a searcher must read before trusting the tool's order. Deeper reads guard against buried adverse cases but slow every search.
- Re-run trigger — whether the sweep must be repeated when new authority lands before the decision issues.
When it helps, and when it misleads¶
Its strength is that it attacks selective citation at the source: it makes the absence of a contrary-case search a visible, auditable failure rather than an invisible convenience. In fields with a professional duty of candor[1], it operationalizes that duty into a checkable routine instead of a hope.
Its failure mode is theater. A protocol can be satisfied in form — boxes ticked, databases named — while the searcher runs deliberately weak adverse queries designed to find nothing, so the log certifies a sweep that was engineered to fail. It also cannot judge what it surfaces: it guarantees the contrary case was looked for, not that its significance was correctly weighed. And an over-heavy protocol invites fatigue, where searchers pattern-match their way through the ritual. The guarding discipline is to audit the adverse queries themselves — sampling logs to check that the contrary-case searches were genuinely capable of returning something — and to keep the automated ranker on a short leash rather than trusting its top page.
How it implements the components¶
This protocol fills the retrieval-integrity slice of the archetype and nothing downstream of it:
precedent_retrieval_and_adverse_search— it is the adverse-search half of this component: mandatory contrary-case queries, logged coverage, and documented rejected hits.machine_assisted_similarity_guardrail— by requiring a check that the ranking tool is not burying adverse results, it keeps the automated similarity engine subordinate to an accountable human sweep.
It does not decide, communicate, or transition anything. prospective_change_rule and reliance_fair_notice_and_transition_assessment belong to its protocol twin Precedent Change Notice and Transition Plan, which acts after a change is decided; this protocol acts before any decision, and it does not implement the similarity comparison itself (material_fact_similarity_model, held by Material-Fact Comparison Matrix) or the treatment choice (precedent_treatment_decision, held by Follow–Distinguish–Overrule Memo).
Related¶
- Instantiates: Precedent-Guided Decision Governance — supplies the balanced case universe every later step reasons over.
- Sibling mechanisms: Follow–Distinguish–Overrule Memo · Material-Fact Comparison Matrix · Precedent Authority and Validity Table · Precedent Case Brief · Precedent Change Notice and Transition Plan · Precedent Citation Graph · Precedent Conflict Panel · Precedent Consistency Audit
Editorial Notes¶
Form Classification¶
Form family: Protocol, Workflow & Routine
Rationale: The mechanism requires a structured second search pass with mandatory adverse query classes, coverage logging, and documented treatment of rejected hits before finalization, so its operative form is a repeatable search protocol.
Nearest alternative: Rule, Policy & Commitment — The obligation to perform adverse search is a standing rule, but the mechanism specifies and depends on the ordered search-and-documentation enactment.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Law & Governance
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: Mandatory search for limiting, distinguishing, and conflicting authorities is a canonical legal-research and precedent-governance discipline.
Related originating lineages:
- Library & Information Science — Systematic retrieval, documented coverage, and search completeness contribute the information practice.
- Philosophy — Active search for defeaters and counterevidence supplies the broader epistemic norm.
Review resolution: Mandatory search for limiting, conflicting, or later adverse authority is a specialized legal-research protocol. Library science supplies retrieval and philosophy contributes confirmation-bias discipline, but precedent governance remains the single primary lineage.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] American Bar Association. “Model Rules of Professional Conduct: Rule 3.3, Candor Toward the Tribunal” (n.d.). Requires advocates to disclose known, directly adverse controlling authority not already disclosed by opposing counsel. registry ↩