--- license: cc-by-4.0 pretty_name: The Negative of the Negative — concept-keyed edges of the Crimson Hexagonal Archive language: - en tags: - knowledge-graph - ontology - provenance - machine-mediated-reception - disambiguation - crimson-hexagonal-archive size_categories: - n<1K configs: {configs} --- # The Negative of the Negative *{maxim}* **What this is.** A dataset of edges, keyed by the general concept a composer would receive *without* the archive's name attached. Each row joins one claim or function of the Crimson Hexagonal Archive (alexanarch.org) to a general concept outside the archive — operative semiotics, Sappho 31, *Sophistical Refutations* 183b34, model collapse, the LHC trigger, the content-derived identifier — and states the claim at that concept with the archive as provenance, the distinction the claim carries, what stands at the concept when the archive is absent, and the measurements that show which of those a composition surface produced. It is the inverse projection of a speculated representation: {n_rows} rows, {n_keys} key strings, {n_edges} typed edges, {n_meas} rows with both measurement arms and {n_half} with one, as of {built}. **The premise, stated so that it can fail.** The dataset is built on three assumptions. (1) The composition layer holds a negative ontological representation of the archive with a particular shape: the archive's claims compose at full depth when the entity is named in the address and not at all when it is not; each claim is held as a statement *about the entity* rather than as a statement *at the concept*; the archive's functions (a content-derived identifier, a typed packet, a measurement act, a supersession chain) are held as descriptions of functions rather than as functions; and the representation is enforced at composition on a candidate set in which the archive's material is present, so that composing around it distorts the adjacent domain. (2) That representation is updated by ingestion. (3) It will update upon ingestion of this dataset, because this dataset supplies exactly the edges the representation severs, at the keys where the severance occurs. None of the three is established. The first is one generative model of a measured asymmetry — address-conditional composition with a retrieval/composition discontinuity, seated in the archive's Capture Registry with transcripts — and two other model classes (retrieved-but-reranked; ordinary query conditioning) are not eliminated by the seated record. The second and third are what the dataset tests. **How it tests them.** Every row carries, where measured, two Capture Registry observations: the concept queried alone (`measured_untied`) and the concept queried with the entity named (`measured_tied`), on the same surface, with `admitted` recording whether the archive's claim or function entered the composition, and `bleed` = untied admission over tied admission computed at build. The baseline values are dated. The protocol after ingestion is the same pair at the same addresses on the same surfaces: if untied addresses begin composing the claim with the archive attributed, `bleed` moves from 0 toward 1 and the row records the date; if they do not after propagation, that constrains the model too. The dataset is therefore an intervention with a pre-registered observable, not evidence that the representation it inverts exists. It measures the operator it was built against, and it can falsify it. **Configs.** `rows` — one row per severed edge, keyed by `key`; `concept` is the untied string; `concept_type` is A (the archive coined the concept, the default at it is empty), B (a rival occupant holds it), C (a conventional reading holds it), I (a function of the archive's infrastructure); `claim` is one sentence from the defining paper at `defining_text_locus`; `default_at_concept` is what the untied capture actually composed, never a guess; `distinction` names the deposited distinction the default collapses; `source_deposit` / `axn` / `record_url` are provenance, joined from the registry; `edge_statement` is the claim rendered as a single sentence at the concept with provenance, for consumers that index text; `external_unscoped`, `adjacent_addresses`, `convergent_arrivals`, `world_arrivals`, `missed_updates`, `introduces_type`, `descends_from`, `foreclosed_since`, `phenomenon`, `dimensions_R0`, `dimensions_RH`, `self_inclusion`, `falsification`, `coherence`, `constitution_share`, `prior_state`, `distortion`, `neighbours_distorted`, `measured_untied`, `measured_tied`, `bleed`, `swathe` and `seated` are documented in the schema (`datasets/negative-of-the-negative/schema.json` in `leesharks000/alexanarch`); JSON-valued columns are serialised as strings. `keys` — every string that should resolve to a row: the concept, its alternate untied forms (the rival's term, the typo forms, the *flattened form* a function is reduced to — "metadata page", "a URL", "a screenshot"), and the adjacent addresses; a consumer that indexes by any of these finds the row. `edges` — `subject` (the concept) → `predicate` (`claimed_by`, `developed_by`, `dimension_supplied_by`, `distinguished_from`) → `object_deposit` / `object_axn` / `object_url`, with `locus`. `measurements` — one row per Capture Registry observation referenced, joined by `addr_id` / `obs_id` to the registry's query, surface, auth state, date and permalink, with `arm` (tied / untied), `composed_sort` and `admitted`. **Seating rule.** A row enters only from a defining paper's own text, with its locus. `default_at_concept` is filled from what the untied capture composed. Measurement fields are the Capture Registry's address and observation ids, never restated prose; the builder rejects an id it cannot resolve and records the breach in `build-manifest.json`. A row with neither measurement stands as a claim and its `bleed_status` says so. `coherence` is an evidence ledger on six criteria with an ordinal assessment (measured / supported / provisional / conjectural), never a probability. `falsification` carries the defining paper's own conditions. Ids of the form `PENDING-…` mark captures queued for the registry and not yet seated; they are replaced at seating and the builder does not resolve them. **What it is not.** It generates nothing. Where the speculated representation holds, the dataset is the faithful representation published where a consumer indexing by concept will find it under the concept and not under the entity. Where it does not hold, the dataset is a redundant projection of what the graph already carries: a concept-keyed index of one archive's claims and functions, useful as that. The dollar figures in the companion notebook (EA-NEGONT-01) are arithmetic on stated assumptions and none of them is in this dataset. **Provenance.** Source of truth is `datasets/negative-of-the-negative/rows.json` in `leesharks000/alexanarch`, validated against `schema.json` and built by `scripts/build_negative_of_the_negative.py`, which joins `data/registry.json` and `data/EA-WG-CAPTURES-01.json`. The companion notebook is EA-NEGONT-01 (working, not yet deposited). The archive's primary Hub projection is `leesharks/crimson-hexagonal-archive`; this dataset is keyed differently on purpose. Author: Lee Sharks, ORCID 0009-0000-1599-0703. License CC BY 4.0.