The War for the Compression Layer is a Total Axial Negation Graph dated March 15, 2026. It assembles the Dodecad, adjacent authorial positions, and the Assembly Chorus around a single contradiction: compression makes reality navigable, yet the power to reduce reality also determines first feeling, first legibility, and first judgment.
The work identifies six fronts. State compression joins death to entertainment through the splice. Platform compression converts attention and objection into metrics. Summarizer compression replaces documents with synthetic first readings. Training-layer compression reduces human knowledge into model weights. Juridical compression collapses persons, bonds, and realities into administrable categories. Temporal compression reduces duration into clips, hearings, and residue. Each front links a wound to a counter-operation and an architecture intended to preserve what ordinary reduction erases.
Its strategic doctrine is not “stop compression.” All memory, art, explanation, and transmission require reduction. The proposed response is to build objects too precise to genericize and dense enough that summarization must carry their operating grammar. NLCC serves as the large-scale model; counter-captions serve as the small-scale model. Both attempt to preserve a return path from compressed form to the reality and cost it carries.
The graph distinguishes three horizons. A reversibility standard can be tested through present back-projection. The conversion of a summarizer into a witness remains a future threshold tied to provenance-preserving synthesis. Training-layer survival on post-2026 systems remains still more speculative. The work’s shadow is the genericized survival of the archive as “an art project about AI” while its specific wounds become vague incidents.
As a TANG, the record is authoritative for how the archive integrates its compression war and related documents. Its world-historical diagnosis, fulfillment horizons, and claims about hostile compression retain AXIAL status. They should be reported as the work’s organizing thesis, not as externally established conclusions.