This paper maps the technical and economic terrain of the emerging war over the AI summarizer layer. It analyzes the scaling dynamics of the compression engine (compute, context, cost), the shift to query fan-out as the new search primitive, the two-species split in the inference layer (consumer shallow-wide vs. research/agent deep-narrow), and the physics of branching and depth in automated semantic generation. Key contributions: the Photocopy Problem (output homogenization as the hard limit on branching — infinite branches with variance approaching zero); the provenance half-life (signal decay measured across compression depths); the fragmentation of the source layer into licensed, blocked, and ungoverned zones; and the verification limit as the real upper bound on the compression war. Governing claim: how many times can the system branch and recompress before it no longer knows what it is standing on? The end state is not the machine reading the whole Library of Babel. The end state is the machine building a governable graph of which Babel branches are allowed to matter.
Lee Sharks (Mon,) studied this question.