leesharks/provenance-erasure

all things are now lawful to you in jack feist

EA-RHIZOME-PER-01 — provenance erasure

Where provenance disappears, and what measures it. A transformation chain — source, retrieval, selection, composition, summarisation, reception — with the instruments that score retention at each step, and the cases where the instruments were applied to themselves.

Counts

71 nodes · 0 core · 0 neighbour · 7 edges · 7 stolons

Roles

Order is the classifier — role_for returns on first match, so a broad pattern above a narrow one swallows it.

role	what it names
instrument	THE MODULE AND ITS CHAIN. Five versions survive on the site: #780, #156 v3, #817 v3.1 with its three extensions — #1527 the Lucente Extension carrying ABN, #1528 Four Interfaces, #1529 the Claim Status Packet — and #1530 v3.2, made self-complete 'because by v3.1 the chain had reached four deposits and a module that cannot be run from one document is a changelog rather than a module.'

A FIRST READING SCATTERED THEM. #156 filed as magnitude because its subtitle lists PER and DSL; #1530 and #1527 as precondition because they carry ABN. The module is the module whatever its subtitle leads with, and it is matched by name. | | magnitude | PER and its M/C/D dimensions — HOW MUCH IS LOST. With the unit rule that says what may be counted, because a PER on the wrong unit is a different metric. | | orientation | Ω — WHICH WAY IT FALLS. Not a programme beside the metric: the second axis of one framework. 'A rate without an orientation cannot tell you who was erased.' | | precondition | what runs before any metric, in an order the module calls load-bearing. ABN is rank zero. | | before_the_machines | DOMAIN 2 AND THE FRAMEWORK'S SECOND SPINE, not background to it. Erasure with a 2,600-year measurable record, its own four-vector loss model, its own deficit Δ = W − R, and a twelve-station witness table in which attribution dies at the channel-boundary rather than with time. THE BODY CARRIED FOUR NODES FOR THIS BEFORE THE SITE'S DEEP-TIME WING WAS READ.

AND THE WING'S OWN CHAIN WAS ENTIRELY ABSENT UNTIL ITS VOCABULARY WAS READ. #1512 Deep Time is a SURFACE RECORD — a short pointer to the live page, naming its chips: #1493 The Queen's Seizure and the Re-fired King, #1494 The Transform Signature Frozen, #1495 The Sign Under Test, #1496 The Suppression Map, #1497 Undoing the Suppression: The Counter-Stack. ALL FIVE SCORED OUT, because a grammar built from the site's section headings does not contain the words its deposits are titled with. | | process_provenance | the third dimension the site adds to C2PA and PER. 'Fluency can be generated. Provenance must be borne.' | | reception | what surfaces did with the instruments. Named shapes rather than generic loss — including the battery where a surface fabricated replacement metrics, ran them on itself, and gave itself perfect scores across all dimensions. | | self_application | the instrument applied to itself. The Self-Audit Module Dissolved, PER 1.00. The Empty Bracket. | | theory | the reasoning about what erasure IS. A first pattern used generic words — argument, thesis, finding, interface — and filed Logotic Hacking, the Constitution of the Semantic Economy and a Pessoagraph workplan as provenance theory. OTHER FRAMEWORKS ARE NOT THIS ONE'S THEORY. | | evidence | where an instrument was pointed. Beneath the apparatus, not in front of it. | | unroled | the honest floor |

ORDER IS THE SITE'S, WITH ONE CORRECTION LEARNED BY EMITTING IT. The named instrument leads, because every version of the module carries PER and DSL in its subtitle and a magnitude-first classifier files the module as the metric — #156 v3 did exactly that, twice.

Then magnitude and orientation, which the site says are the two moments of one loss: 'PER measures the first, Ω the second — A RATE WITHOUT AN ORIENTATION CANNOT TELL YOU WHO WAS ERASED.' Then the preconditions the module runs before any metric, in an order it calls load-bearing. Then the domains the framework claims, the named reception shapes, self-application, theory, and evidence last.

EVIDENCE IS LAST BECAUSE A CAPTURE IS WHERE AN INSTRUMENT WAS POINTED. An earlier grammar put it high and the body became its own evidence file.

A node matching nothing falls to unroled, whose count is this grammar's own error bar.

The other bodies

Each is emitted by one generator from a grammar file, and each declares stolons naming the siblings it advertises and does not contain.

• leesharks/heteronyms
• leesharks/machine-mediated-reception
• leesharks/model-collapse-anti-collapse
• leesharks/poetics
• leesharks/revelation-first
• leesharks/semantic-economy
• leesharks/spxi-mpai

Where it points and does not go

Advertised, not included. The outside stays outside and stays reachable.

• model-collapse-anti-collapse — narrowed_by from concept:provenance-erasure what erasure does to a distribution over time. This body scores single transformations; that one holds the question of what repeated erasure does to the space of what can be said.
• semantic-economy — priced_in from concept:bearing-cost erasure is a cost borne by the provenance-bearer and never by the composer. WHO PAYS is that body's question and this one does not answer it — it only establishes that something was taken.
• machine-mediated-reception — observed_in from concept:attribution the 425-address registry holds the observations these instruments score. This body defines PER; that body is where PER values live.
• spxi-mpai — instrumented_in from concept:inscription whether an inscription survived is measured here; whether it was placed well is specified there. Deliverable ten, the declared SIM set, is an anti-erasure technology.
• revelation-first — adjudicated_in from concept:documentary-authority a chronology settled on a late and ambiguous witness is an attribution question at historical scale. THE INSTRUMENTS HERE HAVE NEVER BEEN RUN ON IT.
• heteronyms — borne_by from concept:attribution PER scores whether a byline survived. That body holds the bylines, and its apparatus covers twelve of twenty-six.
• poetics — borne_by from concept:extraction-fidelity Pearl's machine score preserves lineation at whitespace fidelity. WHETHER AN EXTRACTION THAT PERFECT PRESERVES THE AUTHOR is this body's question and it has not been run on it.

Reproducing it

RHIZOME_GRAMMAR=rhizomes/_grammars/provenance-erasure.json \
  python3 scripts/build_rhizome_mc.py

spore.json carries the recipe and the commit of the ledger it came from. The grammar is read, not carried: every determining rule above lives in rhizomes/_grammars/, and a different body means a different grammar file, not a fork of the generator.

Emitted from the Crimson Hexagonal Archive · alexanarch.org