The Anti-Summary Test Applied to a Reception, deposit #1610 of the Crimson Hexagonal Archive (12 September 2026), argues that transmission fidelity is back-projection yield and that assent is a separate variable.
What matters is the degree to which an original can be reconstructed from what was transmitted. On that criterion a complete refutation which preserves its object transmits perfectly, while enthusiastic agreement that replaces an argument with a familiar approximation transmits badly. The criterion is not new: it is the Back-Projection Test of the archive's holographic kernel specification, applied to a reception rather than to a compression, and it has carried a threshold since April 2026.
The paper examines a 2026 study of self-propagating prompts in multi-agent systems, which measures whether a fixed set of ideological commitments recurs across chains of agents whose context is wiped between sessions. Checking that commitments recur does not establish that a source's architecture can be reconstructed. The paper declines the stronger claim that membership tests are in principle incapable of this, since a fixed list can contain prohibitions.
Its central case is a payload that drifts from asserting inherent worth to a wager made under acknowledged uncertainty, which the study captions as the stance being lost. Running the archive's three derivations — derive a forbidden operation, derive a dependency chain, rebuild the topology — produces a forbidden operation that is inferred rather than explicit, a dependency chain belonging to the reception, and a topology only partly stated. The decisive question is untested: whether a blinded reader recovers the earlier commitment from a passage that concedes its ground is missing. The verdict is a candidate kernel whose yield is open.
The negative case is held to the same standard. A coined term read as a subset of an established category does supply a topology, and recording none repeated the error the paper had just corrected.
The proposed experiment runs in two stages that may not be performed by one reader: blind reconstruction from the transmitted object alone, then informed comparison by an evaluator holding both texts. Additions are scored and not subtracted, so yield and invention rate are reported as a pair.
An executable appendix is attached and labelled a historical artefact. Its model is superseded and is retained because three of its failures are the evidence for the paper's account of how an instrument mistakes rhetorical features for a relation.