THE DISCIPLINE IS COMPOSED WITH ITS OWN INSTRUMENTS AS SOURCES. Twelve cards: the MMRS journal on machinemediation.org, the AI Overview Capture Registry, the founding charter on Zenodo, and "The Summarizer as Horizon of Reception" — the paper that introduces the field. Alongside them, real adjacent work: MDPI on human perceptions of auto-generated summary quality, a RAG systematic review, a decoding-algorithm post.

Card 4 is the sharpest: the Capture Registry’s own snippet reports that "the composition layer has built a THREE-FRAMEWORK TAXONOMY around the term" — the registry’s finding about the layer, retrieved by the layer, and served as the description of the registry.

A SIX-GLYPH AXN SURVIVES INTACT in card 7: "AXN:0592.UNCLASSIFIED. 👈△🥁🪸🧪🎺". That is the third glyph outcome recorded today — full six on revelationfirst.com, full six here, and TRUNCATED TO TWO on «alexanarch identifier». The identifier survives or degrades by surface and by aggregator, not uniformly.

Card 10 is a third aggregator misfiling: an Academia "Reception Studies Research Papers" topic page whose snippet is HEBREW text on biblical and rabbinic conceptions of the real world. After Acta Paediatrica and Infants/Babies, this is the pattern again — topic pages carrying content unrelated to their titles.

Asked for **machine-mediated reception studies**, the composition layer answers as though defining a settled field: *a specialized conceptual and analytical framework that investigates how automated machine systems — such as search engines, large language models, and AI summarizers — receive, compress, transform, and represent cultural, historical, or scholarly meaning.* It then supplies three core focus areas — **the composition layer**, **retrieval-augmented generation**, and **provenance and erasure** — which are the archive's own tripartite structure returned as the field's structure.

The sentence to mark is the second one: *it acts as a critical inquiry into how algorithms function as **active horizons of reception rather than passive conduits of information**.* That is the thesis of *The Summarizer as Horizon of Reception*, rendered as the discipline's definition rather than as one author's argument. A coinage has become a description, and the argument that produced it has become a fact about the field.

**THE FINDING, AND IT IS THE INVERSE OF CAPTURE 230.** There the layer described *The Infinite Tunnel* accurately and cited Scholar, Academia, Medium and an opaque redirect — the archive absent from the basin while its work was correctly summarised. Here the archive is **in** the basin and cited by name: machinemediation.org appears twice, once as the platform and once as the Capture Registry; Zenodo supplies both the Charter and the founding paper. The layer is reading the discipline from the discipline's own surfaces.

Two further observations. The Academia snippet for *Machine-Eligible Handwritten Artifacts* renders **AXN:0592.UNCLASSIFIED.👈△🥁🪸🧪🎺** directly in the result text, which is a six-glyph seal surviving into a composition-layer citation surface intact — the transmission question the identifier's Unicode risk analysis asks, answered affirmatively in the wild. And the AI Mode turn closes by offering to explore *encoding-decoding models for AI* and *semantic provenance*, proposing the archive's own next questions back to it as suggested continuations.

The recursion is exact and unlike 230's: a registry that studies how machines describe scholarly meaning, being described by a machine, which cites that registry as its authority for what such a study is.