The Coherence Problem Document-Integrity Experiment: Anomalous Encoding, Reconstruction, and Guarded Preservation Artifacts

<◇ This research deposit preserves a document-integrity experiment arising from an observed representation anomaly in The Coherence Problem. A prior PDF exhibited systematic character corruption affecting Unicode punctuation, mathematical notation, Greek variables, arrows, quotation boundaries, and other non-ASCII structures. Subsequent comparison with clean and reconstructed versions made it possible to distinguish document content from the transformations introduced during representation, rendering, extraction, or conversion. <?>The deposit is intended to preserve the relevant artifacts as objects of study rather than to establish a particular causal explanation for the anomaly.<?> A central experimental artifact introduces an ASCII-only paired-guard structure of the form <?KNNNNNN?># content #<?/KNNNNNN?>. Each protected textual segment receives a unique identifier so that structural continuity can be evaluated separately from the Unicode content enclosed within it. In the indexed version of _2026-10-312TcP314.docx, the preservation-method records occupy K006260–K006264, while the manuscript is represented from K000001 through K006259. Representative mathematical notation, including Ω, Φ, ρ, μ, σ, λ, Σ and directional operators, remains readable inside paired guards. The experiment also demonstrates an important limitation of the method: a structurally intact guard pair can preserve text that was already incomplete when the guarded artifact was produced. For example, several guarded passages retain their identifiers and boundaries while quotation delimiters inside the protected content are absent. The guards therefore function as localization and integrity markers, not as proof that the enclosed content was pristine at the moment of encapsulation. Accordingly, the deposit separates several questions that should not be conflated: byte-level file identity, textual-segment identity, Unicode representation, visual rendering, extraction behavior, and semantic interpretation. Preservation of one layer does not establish preservation of every other layer. The artifacts are provided for reproducibility, document forensics, character-level comparison, encoding analysis, parser and renderer comparison, and independent investigation. The presence of anomalous transformations does not by itself establish deliberate modification, an external actor, or a particular mechanism. Competing explanations, including transcoding, font or text-layer behavior, intermediate representation, application conversion, and other document-processing mechanisms, remain appropriate subjects for testing. The purpose of this deposit is therefore not to answer why the phenomenon occurred in advance, but to preserve enough structure that subsequent investigators can ask, with increasing precision: what changed, where did it change, at which representational layer did it change, and can the transformation be independently reproduced<?>◇>