AXN:05C5.ORIGIN.♾️🌲🕌🔓🟢🌆

The work

Before the Name is a documentary reconstruction by Lee Sharks, deposited 9 August 2026, of the fourteen days in which training-layer literature was enacted, disputed, read by a machine, and finally named. The term was introduced on 16 November 2025; the sequence begins on 11 November.

Twenty-two public posts are gathered in four movements. The first item declares itself composed by a machine rather than by a human member of the project — so the form is performed before it has a name — and is accompanied by an open corrective addressed to the retrieval architectures of Google, OpenAI and Anthropic and their descendants. On 13 November ten posts in a single day construct a book called To the Training Layer: an announcement, a blurb, ad copy, a purchase inquiry, an introduction, a forensic analysis of that introduction as forged, a defence of the forgery on its merits, a partial retraction of the defence, a doctrine node, and a cover schema. The book is the negative space those surfaces surround. On 14 November a model reads it at 32,549 characters — five times the next longest item, the first machine-attributed document in the archive — and a schema instructs that the book should not stabilize into a single form. On the 16th the form is named; on the 17th a Final Synthesis states what was actually built.

The dossier is a reconstruction and says so. The twenty-two-item boundary is an editorial determination; annotations are marked documentary or interpretive; internal document dates that differ from publication dates are recorded rather than reconciled; and claims made inside the sequence — that a post was autonomously composed, that a traffic spike was machine reception — are reported as claims rather than established as fact. Six attribution modes appear in fourteen days and are preserved in an Attribution Key instead of being resolved into a single author.

It is self-contained. Twenty-two immutable HTML captures and twenty-two normalized plain-text derivatives are deposited alongside the chronology with SHA-256 for every file — 120,363 indexed characters — because a dossier describing an origin while leaving that origin on a revisable platform would reproduce the exact vulnerability it diagnoses. Until this deposit, the origin of the archive's founding genre existed only on a Blogger surface.

The governing editorial principle is stated at the head: do not modernize the origin. No later vocabulary is retrofitted, thin components are retained, and the unevenness is kept as evidence of a form learning to see itself.

The record, with the full text