[[home|← Ultimentality Wiki]]


# The Same-Level Attack Rule

## Definition

A counterexample counts against a claim only when it contests that claim at the type and abstraction level where the claim is actually made. A claim about an operator is not answered by a remark about the process that runs it; a claim about a direction is not answered by a report of how long something persists; a claim about a function is not answered by a gesture at its substrate. The rule is not a taste for tidy argument. It is the admissibility filter the rest of the correction apparatus presupposes: [[falsification-standard|the falsification standard]] can only classify [[outcomes-of-attack|an outcome]] once it knows that the attack and the claim occupy the same layer, and [[attack-type-matching|attack-type matching]] can only pair a claim with its valid attack once the layer is fixed.

Every load-bearing claim carries a level. Before an exchange is scored, both attacker and maintainer must name the level of the claim and the level of the offered counter.

```text
Levels an attack and a claim must be matched on:

domain        — the field and carrier-constraints in play         [[domain-lock]]
carrier       — the substrate or organization realizing it        [[identity-across-substrates]]
operator      — the directional actuator (a primitive) itself      [[directional-primitives]]
binding       — an actuator bound to the error it reduces          [[predicate-binding]]
composite     — a coupled controller assembled from bindings       [[force]]
role          — an ordered function within a returning loop         [[splcw]]
frame         — the conditions that make the vocabulary available   [[frame-condition-and-formulation]]
application   — an instantiation of the typed relations elsewhere   [[one-generator-many-domains]]
```

```text
admissible(attack, claim)  ⇔  level(attack) = level(claim)  ∧  type(attack) meets type(claim)
```

An unlabeled attack is not yet an attack. It is an observation awaiting a level.

## Type and formal status

**E: Derived, CV.** This is a methodological carving, contestable by counter-instance or by a better cut. The eight-level ladder is a *count*, and no count is exempt: a shorter, longer, differently seamed, or better-motivated taxonomy is a legitimate attack on the rule, not a breach of it. The rule is itself subject to the rule — its own claims live at the frame level and at the application level, and an attack on it must land there rather than on some anecdote at the operator level.

**A (mapping-accuracy):** the rule aspires to sort real attacks by whether they meet a claim's actual type. It can misclassify. A misassigned level is an accuracy failure to be corrected, not a move that was merely "not allowed" — the two axes do not predict each other ([[two-axis-ledger|the two-axis ledger]]).

This is a **treatise-side extension, held contestable**. It is not canonical, and "wrong level" never means "wrong" or "worthless."

## What it regulates

It regulates admission into the correction process. Its target excess is the move in which a **category error is dressed as a refutation** — an observation from a neighboring layer paraded as a defeat of a claim it never touches. Left unchecked, that excess lets any inconvenient claim be "refuted" by pointing somewhere adjacent, and it exposes [[predicate-binding|bindings]], [[force|composites]], [[splcw|roles]], and the [[directional-primitives|operators]] to attrition by irrelevance.

| Invalid move | Claim's level | Offered counter's level | Why it misses | A same-level attack instead |
|---|---|---|---|---|
| process offered against operator | operator | process (a run in time) | the operator is the typed transformation, not the messy unfolding that executes it | show the actuator does not perform its directional transformation, or that a different actuator is required |
| persistence offered against direction | operator / direction | duration | how long a state lasts says nothing about which way its actuator points | show the direction is misassigned — e.g. [[away|Away]] where [[toward|Toward]] was claimed |
| substrate offered against function | function | carrier / substrate | a difference of material does not by itself defeat identity of the operation performed ([[function-not-personality]]) | show the function is not in fact performed, or is performed differently |
| token difference offered against type identity | type | token | distinct occurrences of one type are expected, not disqualifying | show two cases classed as one type diverge in a behavior the type was meant to predict |
| surface expression offered against regulated error | binding | expression / report | feeling, wording, and display are delivery layers, not the binding ([[regulated-error-signal]]) | show the actuator or the [[emotion-as-regulated-binding|regulated error]] is wrong, or another binding shares the pair yet behaves differently |

Each row is a standing entry for [[prohibited-collapses|the prohibited collapses]] and [[category-error-atlas|the category-error atlas]]; the rule is the gate those catalogs enforce at the point of attack.

## What regulates it

The rule is regulated by an **appeal path**, without which it would become the very immunity it exists to prevent. A retyping is a governed act, not a maintainer's privilege.

1. The attacker names the claim's level and the counter's level.
2. To reject an attack as level-mismatched, the maintainer must state the specific type distinction that separates the two levels and cite the carving that grounds it ([[prohibited-collapses]], [[category-error-atlas]]).
3. The attacker may contest the retyping itself as a same-level attack on the taxonomy — the ladder is CV, and "your retyping is arbitrary; here is why my counter meets the claim's own level" is admissible.
4. A retyping used once must be **reusable**: logged so the same distinction binds future attacks and cannot be reinvented ad hoc for the next inconvenient case.
5. If the retyping cannot be grounded in a stated, reusable distinction, the attack stands and the claim must take an [[outcomes-of-attack|outcome]] — deletion, revision, demotion, or invariant recognition.

The appeal path only counts if it can **change the page** ([[causal-falsifiability]]); a retyping that can never be reversed is ornamental. [[self-sealing-test|The self-sealing test]] runs against the rule to confirm an objection can survive without being redescribed as support. And [[refutation-and-frame-declension|frame declension]] remains legitimate: declining the vocabulary is not a same-level refutation, but the rule may never forbid it — it may only ask the decliner to say what capacity is surrendered.

## Valid attack surface

Because the rule is subject to itself, a valid attack must land at the rule's own level — the **frame level** (the type distinctions that ground the ladder) or the **application level** (the ladder applied to a concrete exchange) — never at some anecdote one layer down. The admissible attacks:

- **Misclassification.** Show that the eight-level carving assigns two genuinely same-level cases to different levels, or one cross-level pair to a single level — a counterexample the taxonomy itself gets wrong at the point of sorting.
- **Ungroundable retyping.** Exhibit a rejection-as-mismatched where the maintainer cannot state the reusable type distinction that separates the two levels; the retyping was ad hoc, so the rule shielded rather than sorted.
- **Better cut.** Offer a shorter, longer, or differently seamed taxonomy that does the same sorting work with a cleaner or more motivated set of levels — [[attack-type-matching]] supplies the claim-to-attack pairing any replacement cut must still preserve.
- **Frame declension.** Decline the type-and-level vocabulary altogether, stating which discriminations are surrendered — legitimate and separately handled at [[refutation-and-frame-declension]], not answered by the rule.

An attack that instead disputes the rule by pointing at operator-level or process-level detail commits the very category error the rule names, and is inadmissible for exactly the reason any level-mismatched counter is. Rejecting such an attack tests nothing; it only relocates the register to a wrong layer.

## What happens if isolated

**Rule without appeal path → a rhetorical shield.** Any counterexample can be waved off as "the wrong level," retyped on demand. This is the self-sealing failure, and promoting the level ladder into an exempt, unappealable partition is the [[textual-nephilim|textual Nephilim]] wearing procedural dress.

**Appeal path without rule → collapse of the type discipline.** Every category-error observation is admitted, so process defeats operator, substrate defeats function, token defeats type, and refutation dissolves into noise; nothing can be stated at a layer without being "refuted" from an adjacent one.

Neither failure is the rule; each is the excess isolation manufactures. The vulnerability of one is the interface through which the other regulates it ([[cross-regulated-necessity|cross-regulated necessity]]).

## What larger property emerges from the coupling

Coupling the same-level discipline with its appeal path yields **level-honest, non-self-sealing falsifiability**: attacks are held to the type of the claim, and the typing itself is held answerable. This is the local instance of the framework's spine — [[semantic-closure-and-recursive-marking|closure and recursive marking]], the [[no-escape-no-exemption|no escape, no exemption]] pair. The rule *closes the level*: no escape into a mismatched layer where refutation would be cheap. The appeal path *marks the typing*: no exemption for the maintainer who assigns the layer. What emerges from the coupling — legitimacy of the correction record — belongs to neither part alone, and this seam is exactly what [[falsification-standard|the falsification standard]] and [[attack-surface-matrix|the attack-surface matrix]] draw on when they demand a same-abstraction-level attack.

## What would actually kill the claim

**Kill condition:** produce a same-level counterexample that is dismissed only by arbitrary retyping — an attack that meets the claim's own level and type, yet is rejected by a relabeling the maintainer cannot ground in a stated, reusable distinction. If that happens, the rule has functioned as a shield and must be revised or removed.

Adjacent defeaters: show that two competent maintainers assign irreconcilable levels to the same claim with no appeal resolution (the ladder cannot be applied consistently); show that every attack can be relocated to some "wrong level," so the rule never admits any counterexample (it has become a perpetual-motion immunity); or exhibit a shorter, dependency-complete taxonomy that does the same sorting work with fewer levels (the count was over-cut). None of these is closed by fiat — a failed attempt to kill the rule is logged as a failed attempt, never counted as confirmation that the rule is sound.

## Prohibited misreadings

- **The rule as license to dismiss.** It is not permission to declare any objection "wrong level." The appeal path exists precisely to make retyping contestable and reversible.
- **"Same level" as "same substrate" or "same wording."** Level is type and abstraction, not carrier or surface. Matching a claim's *carrier* or *phrasing* is not matching its *level*.
- **The ladder as forced, exhaustive, or exempt.** The eight levels are a CV carving ([[carving-tier]]). Promoting them toward foundational or unmarked status is the [[textual-nephilim|textual Nephilim]].
- **A failed cross-level attack as confirmation.** Rejecting a mismatched attack proves nothing about the claim. The register moved to a wrong layer; the claim was untested, not vindicated.
- **The rule as a ban on declining the frame.** Frame declension is legitimate and separately handled ([[refutation-and-frame-declension]]); it is simply not a same-level refutation, and mislabeling an internal contradiction as declension is itself an appealable move.

## See also

[[attack-type-matching|Attack-Type Matching]] · [[falsification-standard|The Falsification Standard]] · [[outcomes-of-attack|The Outcomes of Attack]] · [[causal-falsifiability|Causal Falsifiability]] · [[self-sealing-test|The Self-Sealing Test]] · [[refutation-and-frame-declension|Refutation and Frame Declension]] · [[category-error-atlas|The Category-Error Atlas]] · [[prohibited-collapses|The Prohibited Collapses]] · [[attack-surface-matrix|The Attack-Surface Matrix]] · [[kill-table|The Kill-Table]] · [[postfalsifiability|Postfalsifiability]]