[[home|← Ultimentality Wiki]]

# Attack-Type Matching

## Definition

Attack-Type Matching is the discipline that a counterexample refutes only when its *kind* matches the kind of claim it targets. Every content-bearing claim in the framework carries a type — frame condition, stipulation, carving, binding, coupled controller, role topology, authority, world-mapping, continuity, or answerability — and each type exposes exactly one characteristic surface where a real attack can bite. An attack aimed at the wrong surface is not a weak refutation; it is a category error, and it lands nowhere. Where [[same-level-attack-rule|the same-level attack rule]] governs the *level* of abstraction at which a counterexample must engage, this page governs the *kind*. Both conditions must hold at once: an admissible attack matches its target in type **and** in level. Matching is a filter on admission, not a verdict on outcome; whether an admitted attack then lands is decided downstream by [[outcomes-of-attack]] and [[causal-falsifiability]], never by the table below.

## Type and formal status

**E.** This page is **Derived**, and its mark is Mixed, held as two separable marks per the [[two-mark-system]]. *As organization* — a lookup table routing attacks to claims — it is **Exposition**: it indexes distinctions made on other pages and asserts no new load-bearing relation. *As classification* — the claim that these are the operative claim-types and that each is answerable to *this* attack and not another — it is **Derived, CV**: a carving contestable by counter-instance (a claim whose only honest attack is absent from the table) or by a better carving (a shorter or more discriminating partition of claim-types). The count of ten is itself CV; no row is exempt, and promoting this table from a serviceable carving to a law would be the [[textual-nephilim]].

**A.** The table aspires only to *route* attacks accurately. It makes no world-mapping claim and predicts no result, so it carries no independent [[alethic-axis|A]] burden; that burden falls on the pages it routes to. This is a treatise-side extension, held contestable — nothing here is forced, canonical, or foundational.

## The matching table

Each row names a claim-type and the single attack surface that a valid refutation of that type must reach. The rightmost column names only where a *successful* attack of that kind can come to rest: each entry is a selection from the four [[outcomes-of-attack|outcomes of attack]] — deletion, revision, demotion-or-retyping, and invariant recognition. A failed attack lands nowhere in that column; it is logged as a failed attack, never as confirmation.

| Claim type | Valid attack | Where a landed attack can rest |
|---|---|---|
| Frame condition | frame declension or internal incoherence | revision or invariant recognition |
| Stipulation | decline or show inapt use | revision or demotion/retyping |
| Carving/count | counter-instance or better carving | revision, demotion, or deletion |
| Binding | wrong actuator, error, or causal effect | revision or deletion |
| Coupled controller | removal test or failed mutual regulation | invariant recognition or deletion |
| Role topology | missing function or invalid dependency | revision or deletion |
| Authority claim | contest the authority | demotion/retyping or revision |
| World-mapping claim | predictive or explanatory failure | revision or deletion |
| Continuity claim | lineage or identity break | deletion or demotion |
| Answerability claim | causal non-effect or capture | deletion or revision |

**Reading the rows.** A *frame condition* (an [[frame-internal-tier|FT]] claim) yields only to frame declension or a demonstrated internal incoherence, never to a fact gathered inside the frame; see [[refutation-and-frame-declension]]. A *stipulation* has no truth-value to falsify — it can be declined or shown to misfire in use. A *carving or count* — the four primitives, two Forces, three corruptions, five roles — is [[carving-tier|CV]] and answers to a counter-instance or a better carving; disliking a number is not an attack, because contestable never means probably wrong ([[contestability-gradient]]). A *binding*, [[predicate-binding|B(p) = (p, eₚ)]], is attacked by showing the actuator is wrong, the regulated error is misidentified, or the pairing produces no causal effect ([[causal-error-mark]], [[emotion-as-regulated-binding]]) — recalling that a binding is regulation toward a setpoint, not a conserved invariant. A *coupled controller*, composed by [[force|⊕]] and never by addition, is attacked by a removal test or by exhibiting failed mutual regulation ([[cross-regulated-necessity]], [[reciprocal-attack-surfaces]]). A *role topology* — [[splcw|SPLCW]] and [[the-ring|the Ring]] — is attacked by a missing function or an invalid dependency, remembering that the Ring is maximal and never declared closed, and that seizing the [[witness-outside-the-ring|witness seat]] is the [[nephilim]], not a repair. An *authority claim* ([[authority-canonical-tier|AC]]) is attacked by contesting the authority itself. A *world-mapping claim* answers to the [[alethic-axis|A]] axis through predictive or explanatory failure — and the two axes never predict each other. A *continuity claim* is attacked by a lineage or identity break ([[continuity]], [[symbolic-immortality]], [[continuation-and-colonization]], [[responsible-successor]]). An *answerability claim* is attacked by causal non-effect — [[effective-and-ornamental-answerability|ornamental]] rather than effective correction — or by [[capture-of-corrective-layer|capture]] of the corrective layer.

The outcome column is indicative; the authoritative outcome set, and the rule that a failed attack is logged as a failed attack, live in [[outcomes-of-attack]].

## What it regulates

It regulates the **type-blind attack**: the move that pits a process against an operator, a substrate against a function, a token against a type, persistence against direction, or a surface expression against a regulated error — and then reports the mismatch as a refutation. Those invalid moves are catalogued negatively by [[same-level-attack-rule]] and [[category-error-atlas]]; this page supplies the positive complement, naming for each claim-type the one surface a genuine attack must reach. It equally regulates the inverse failure — attacking a claim as though it were a *stronger* type than it is, treating a stipulation as a world-mapping claim or a carving as a frame condition — which manufactures spurious refutations by importing a burden the claim never carried.

## What regulates it

The classification is regulated first by the pages it routes to: a row is only as good as the target's own declared valid attack, and any target that revises its attack surface forces the corresponding row to revise. It is regulated downward by [[outcomes-of-attack]] — a "matched" attack that can produce no admissible outcome was mis-typed — and laterally by [[same-level-attack-rule]], since type-match without level-match is still inadmissible. Against becoming a shield it is regulated by [[strengthening-without-absolutization]] and [[self-sealing-test]]: a claim's *type is itself contestable*, so retyping a claim in mid-dispute to dodge a landed attack is a tracked, attackable act, never a free exit.

## What happens if isolated

Held alone — a matching table with no [[outcomes-of-attack|outcome]] discipline and no [[causal-falsifiability|causal]] test — the page degrades into a gatekeeper's charter: every inconvenient objection is waved off as "the wrong type of attack," and the type-labels harden into a rhetorical shield. Isolated the other way — outcomes and causal tests with no type-matching — every objection counts equally, a process complaint is permitted to "refute" an operator, and the type boundaries the framework depends on dissolve. The page is load-bearing only in coupling: type-matching decides *which* attacks are admissible; [[causal-falsifiability]] and [[outcomes-of-attack]] decide whether an admitted attack *lands* and require that its failure be logged honestly.

## What larger property emerges from the coupling

Coupled to its regulators, the table yields **type-disciplined falsifiability**: a standard under which the framework can be attacked at every point yet cannot be dissolved by category error, and cannot itself use category error as an excuse. This is what lets [[postfalsifiability]] mean something — invariants that survived *matched* attacks, not invariants that were merely never attacked in the right currency. The matching table is the admissions rule of that standard; [[falsification-standard]] states the standard, and the [[attack-surface-matrix]] is its ledger.

## What would actually kill the claim

The classification dies if a same-level, well-typed counterexample is produced and then dismissed *only* by arbitrary retyping — that is, if the type-labels can always be redrawn to keep any chosen claim safe. It also dies if the ten types cannot be separated (an attack valid against one is valid against all), or if a claim-type is exhibited whose only honest attack lies outside the table and cannot be added without collapsing another type. It does **not** die because someone declines the entire vocabulary of typed attack; that is [[refutation-and-frame-declension|frame declension]] — legitimate but not costless, surrendering any principled way to say why one objection here is better than another.

## Prohibited misreadings

- **Type-matching is not a veto.** Calling an attack "the wrong type" without showing *why its surface cannot reach this claim* is itself a move that must be justified and can be wrong. The rule carries an appeal path; see [[same-level-attack-rule]].
- **A matched attack is not a defeated attack.** Matching makes an attack *admissible* only; whether it lands is decided by [[outcomes-of-attack]] and [[causal-falsifiability]], not by this table.
- **A failed matched attack is not confirmation.** It is logged as a failed attack. Counting it as evidence that the claim is true is the [[self-sealing-test|self-sealing]] failure.
- **Claim-type is not frozen.** Types are [[carving-tier|carvings]] — contestable and re-drawable — but redrawing a type to escape a landed attack is a tracked, attackable act, not a private exit. See [[strengthening-without-absolutization]].
- **The count of ten is not forced.** It is a CV partition, not a foundational number; a better carving supersedes it, no row is exempt, and this table is Derived like every other content-bearing claim — never the [[textual-nephilim|promotion]] of a lookup table into a law.

## See also

[[same-level-attack-rule|The Same-Level Attack Rule]] · [[outcomes-of-attack|The Outcomes of Attack]] · [[causal-falsifiability|Causal Falsifiability]] · [[refutation-and-frame-declension|Refutation and Frame Declension]] · [[self-sealing-test|The Self-Sealing Test]] · [[falsification-standard|The Falsification Standard]] · [[attack-surface-matrix|The Attack-Surface Matrix]] · [[category-error-atlas|The Category-Error Atlas]]