← Ultimentality Wiki

The Outcomes of Attack

Definition

An attack record distinguishes what examination established from what changed in response. A sound objection can expose an error before the target is repaired. A test can also support a claim, reveal a functional dependency, remain inconclusive, or fail to address its target.

The framework retains four useful dispositions:

1. Deletion: remove a defeated claim and repair its dependencies.
2. Revision: change wording, scope, dependency, formulation, or governed behavior.
3. Status change or retyping: change the claim's warranted classification. This can include demotion or a stronger status supported by a new argument; neither is licensed by mere survival.
4. Invariant recognition: a specified removal test establishes a functional loss under the tested conditions. Record the dependency and its scope.

These are overlapping reporting categories, not a proven exhaustive and disjoint partition. Retyping is a kind of revision; invariant recognition concerns evidence about a relation and need not change that relation. Multiple dispositions can apply to one examination.

Supporting evidence without exemption

A failed objection by itself does not establish its target's truth. Distinguish an irrelevant objection or inconclusive probe from a discriminating test whose successful result supports the claim tested. Successful prediction, reconstruction, intervention, and demonstrated functional dependence can strengthen a warrant. None makes a claim immune to further evidence.

An invariant-recognition result supports the named dependency under the removal conditions. It does not prove that every alternative arrangement would fail or that the present formulation is uniquely possible.

Type and formal status

E: Derived, CV. The reporting scheme is revisable. Its usefulness rests on distinguishing evidence, disposition, and remaining uncertainty, not on protecting a count of four.

A: The scheme aims to describe actual examination and correction faithfully. It must accommodate both demonstrated failures and successful evidence without inventing either.

Recording a resolution

target: exact claim and scope
attack_type: kind of examination
evidence: observations or argument, with source
result: refuted / supported-within-scope / inconclusive / failed-objection
outcome: dispositions taken, or none
changed_content: actual changes, or none
changed_marks: warranted status changes, or none
changed_links: affected dependencies, or none
remaining_kill_condition: what would challenge the surviving claim

The fields are a proposed audit format. Missing fields make a record incomplete; they do not erase an actual resolution. A filled form does not prove that the described event occurred. Link the record to inspectable evidence and distinguish proposed work from completed changes.

What it regulates

The scheme prevents mere persistence from becoming proof, and prevents a justified retention from being mistaken for failure to learn. It also distinguishes a refutation from its implementation: an unrepaired error can be refuted and still remain in the governed system.

What regulates it

Causal falsifiability checks whether a warranted correction can affect the target. Same-level examination and attack-type matching connect evidence to the actual claim; they do not dismiss concrete evidence solely because it concerns an implementation of an abstract function. The two marks keep warranted status distinct from accuracy. The scar record preserves reasons and evidence for retained constraints.

Valid attack surface

Show that the scheme cannot represent a relevant result, confuses evidence with action, or permits an unsupported necessity claim to be recorded as a demonstrated dependency. Overlap is not itself a defect in explicitly overlapping categories; ambiguity that prevents useful reporting is.

What happens if isolated

Labels without evidence can misreport correction. Evidence without an effective update path can leave a known error in operation. A prohibition on all supporting evidence makes the account unable to recognize successful tests. Inspect which of these failures actually occurs.

What larger property emerges from the coupling

The result is an auditable correction record: defeats change what they should, successful tests supply their warranted support, unresolved questions remain identifiable, and surviving claims stay correctable.

What would actually kill the claim

The reporting scheme requires revision if it cannot distinguish failed objections, supported results, demonstrated dependencies, and implemented corrections in cases where those differences matter. A better scheme can replace its categories while preserving this function.

Prohibited misreadings

• Support as exemption: stronger evidence does not close a claim to correction.
• No change as no result: justified retention or a demonstrated dependency can be informative.
• A complete form as proof: verify the evidence it cites.
• A missing record as a missing event: distinguish uncertainty about history from disproof of it.

See also

Postfalsifiability · Causal Falsifiability · The Same-Level Attack Rule · Attack-Type Matching · The Self-Sealing Test · Crystallization · The Falsification Standard · The Kill-Table · The Scar Record