Counter-Model Builder

The framework

A motion is an application for a signature. What the movant files is, in substance, a draft of the court's order. Judges sign orders, not arguments.

The Counter-Model Builder states what a system supporting the opposition must do in response. Its thesis is that opposition is replacement, not response. The product is a counter-model: a complete, court-ready account of why the motion fails, grounded in the record, constrained by the governing standard, and auditable by the attorney whose name goes on it.

The requirements, in the order the system must meet them:

1. Fix the frame first. No argument exists before three things are recorded: the procedural posture, the standard of review, and the burden allocation. The controlling formulation of the standard is quoted, not paraphrased, and every later section is checked against it. A motion that paraphrases the standard is logged as a finding, because soft paraphrase is rarely an accident.
2. Decompose the motion onto a board. One axis lists the assertions the motion cannot win without: relief sought, standard invoked, burden assumed, each element attacked, each authority with the precise proposition it is cited for, and each characterization of the pleading or record. The other axis records each assertion's state: holds, contested, severed, or conceded. A cell without provenance (pin cite, pleading paragraph, citator entry, character-checked quotation) does not render as an argument. It renders as work remaining.
3. Treat every citation as a claim. A citation claims that authority transfers from a decided case to this one, and authority travels only by recognized moves (see The Twelve Bridges). Defects map to distinctions courts already credit, in rough descending order: wrong court, wrong posture, wrong element, wrong holding, different material facts, different rationale, different era. Where no move exists at all, the motion's "therefore" is argument, not law.
4. Distinguish only on material differences. Each proposed distinction carries a materiality statement under the rule's own terms, record anchors for every fact asserted, and any opinions that have already drawn the same distinction. Where the corpus holds none, the system says "no distinguishing precedent found," which is not the same sentence as "good law." Strength is reported in words (strong, moderate, weak), never as a percentage.
5. Read the silence. Omissions are findings: the element never addressed, the controlling case never cited, the ellipsis that removed a qualifier, the exhibit cited for a sentence it does not contain. The same machinery runs on the opposition's own side, so that directly adverse controlling authority is surfaced before drafting and forced into a disposition.
6. Force a disposition on every assertion. Fight in full, answer briefly, or concede. Silence is the one disposition refused, because an unanswered point is one the reply characterizes for free. Concessions are drafted with scope language stating what is given, what is expressly not given, and what inference the concession does not license.
7. Draft against the reply. The movant writes last, so each argument carries its strongest answer before filing, including opinions that rejected the same argument. Sections run in the order of the opposition's own theory of denial, threshold failures first. The introduction is written last.
8. Gate, but do not block. A citator gate (run on the opposition's own authorities), a quotation gate, a standard gate, a concession-scope gate, and repetition and budget gates each warn. The attorney adopts the fix, rejects it, or overrides with a typed reason, and the override is logged.
9. Make fabrication unrepresentable. No citation enters a draft except through retrieval that carries provenance. Generative components compose prose from retrieved, verified material; they do not mint authority.
10. Reserve six judgments for the attorney. The theory of denial; the posture of the response (full or partial, aggressive or surgical); each concession; whether each distinction holds; reliance on each authority the opposition cites; and the final attestation. Everything else is execution.

The system produces two artifacts: the filing the court reads, and the record the firm keeps of every proposal, adoption, rejection and override.

How to apply it

The following is an illustrative hypothetical built from the patterns the source essay describes; it is not drawn from a real docket.

A defendant moves under Rule 12(b)(6) to dismiss a four-count complaint that includes fraud. The memorandum runs twenty pages. Parsed onto the board, it reduces to roughly fifteen to twenty assertions doing all the work.

Frame. The system records the posture (pleading stage), quotes the plausibility standard from the governing authority, and notes that the motion's own statement of the standard is a paraphrase. That note goes on the board as a finding.

Audit. One assertion says the plaintiff "failed to prove" reliance and cites an affirmance of summary judgment. The audit asks whether that authority satisfies the move claimed for it. It does not: proof is not the pleading-stage test, and a holding on a full evidentiary record does not transfer to a motion that tests allegations. The cell moves from holds to severed, carrying the defect category "wrong posture."

Silence. The motion attacks four elements of fraud and skips a fifth. The board logs the skipped element as a finding and flags it as possible bait, because an opposition that pounces on it may meet a reply built for exactly that. A limitations argument ignores the discovery rule pleaded in the complaint, and a contract attached to the motion has no authenticating declaration. Both are logged as threshold failures: places where the movant never earned the merits.

Candor. The reverse sweep surfaces one directly adverse controlling decision on the third count. The attorney chooses to concede that count. The drafted concession states that the count is withdrawn, that the withdrawal is not an admission as to any other count, and where the pivot lands.

Drafting and gates. The draft opens with the threshold failures, then the posture defect, then the skipped element. Where courts have rejected an argument the opposition makes, those opinions travel with the draft as warnings. The quotation gate flags one block quote in the opposition's own draft whose ellipsis cuts a qualifier; the attorney adopts the fix. The standard gate flags a paragraph arguing evidence; the attorney overrides with a typed reason, and the reason is logged.

Signature. The six judgments arrive in a queue with reasoning, provenance and draft attached. When the queue is empty, the board shows both readings at once: how much of the motion still stands, and how complete the alternative order has become.

Known limitations and critiques

It is a requirements statement, not an evaluated system. The source essay specifies what an opposition system must do. It reports no measured performance, and nothing on this page should be read as evidence that any product meets the requirements.

Materiality is a legal call. The framework says so itself: whether a distinction holds is one of the six judgments reserved to the attorney. A system can propose distinctions and attach anchors, but similarity is not materiality, and a board full of plausible distinctions can still flatter a weak position.

Coverage bounds everything. "No distinguishing precedent found" is only as informative as the corpus searched, and so are the silence and adverse sweeps. The Pre-Filing Completeness Protocol is the Institute's attempt to make that limit inspectable rather than hidden.

Verbal strength ratings are coarse. Refusing numerical confidence avoids false precision, but "moderate" can hide real disagreement among reviewers. Critics may prefer calibrated estimates with published error rates; the framework's answer is that an unvalidated number invites the reader to stop thinking.

Gates that never block can be ignored. Logging overrides creates accountability after the fact, not prevention. The design deliberately trades hard stops for a record, consistent with the view that verification is a smoke alarm rather than a building code; a firm that never reads its override log gets neither.

Scope stops at the courthouse door. The system does not argue the hearing or draft a surreply, and local practice (page limits, reply rights) varies by forum.

Automation complacency. Execution handled by a machine can erode the attention the six judgments require. The queue design is meant to counter this, but that effect is untested.