Addendum EIC-1 — Middleman and Consensus Tag
Version 1.0 · 3 October 2026
Parent: Watchtower Working Standard v1.0 (1 September 2026)
Status: Optional. Not in force unless the sheet opens with the line “EIC-1 on.”
Layer: Station tag. Printed beside AIS and PAS. Never averaged into either.
Route first. Archive later. A loop is not a second dossier.
What this is
WWS scores one text and one wrap. AIS routes the reader. PAS warns on the package. They are never averaged. Parent bands stay where they are: 90–100 Archive Grade, 70–89 Usable First Look, 50–69 Cross-Examine, 0–49 Do Not Use as Record. Parent R8 stays the education-packet flag. This addendum does not move a band, reuse a flag, or add a domain.
EIC-1 names one station the parent does not name. The same unopened document is the spine of more than one outlet card, and those cards are then cited as agreement. That station is a tag. It is not a finding that the underlying claim is false. Rejecting the loop is not a finding that the event did not happen.
The Evidence Industrial Complex, for this addendum, is a pipeline with four stations:
Production. A shop selects what gets a PDF. Journals, advocacy NGOs, corporate-funded think tanks, litigant shops, and agency memos. Negative results can sit in the drawer. A conclusion can be discussed before the fieldwork is done.
Outlet. A newsroom translates the PDF into a feature and a card. It supplies the moral, the unnamed familiar, and the headline. It does not usually invent the event.
Consensus. A second masthead cites the first, or both cite the shop, and the pair is filed as “outlets of record agree.” Institutional consensus is not evidence. It is a second clock doing unpaid work inside the first article.
Consumption. The card is what most people see. Filing the card as the record is the failure mode.
EIC-1 watches the third station. The first is Domain 3 and R7. The second is Domain 2 and R5. The fourth is PAS and C1.
What changed, and what did not
Preserved from v1.0
Four-domain architecture.
Two-layer print: AIS beside PAS. Never averaged.
No partisan scoring. No credit for emotional sincerity.
No whole-piece kill switch.
Hard gates for First Look and Archive Grade, unchanged.
Parent R8, education packets only.
C1 and C1.1, unchanged. A card is still not the dossier.
Event and package stay separated.
Added
An optional line, “EIC-1 on.”
Flag M1, the consensus-loop flag.
A station tag: Loop, Book tour, or Not in season.
An ingestion block: a C1 card is not stored as a premise.
A routing order that opens the production document, not the next masthead.
Not added
No Domain 2 cap. A ceiling inside a domain is a second score.
No AIS ceiling of 31 legislated onto every card. Score the card on its own text.
No redefinition of R5, R6, or R7.
No use of Layer C, K1, or P1 by another name.
The one rule
Count documents. Do not count mastheads.
A row of headlines is not corroboration. A cluster of cards on one unopened PDF is one object. Promotion of that cluster into “the record” is a separate finding, and only M1 records it.
Run gate
Print the tag only when EIC-1 is on and the sheet already exists. Need both clauses for M1. If either fails, write “EIC-1 not in season” and return to WWS.
M1 fires only when both are true:
The load-bearing number or the moral comes from a document the reader cannot open from the page. No docket, statute, dataset, paper, or on-the-record primary. A title in a sentence is not a link.
That same document is the spine of at least two other outlet cards in the same window, and those cards are cited as agreement — the second masthead quotes the first, or both are filed as independent confirmation.
One book tour is not a loop. Two outlets writing up the same named book, paper, or author essay are distribution of one production document. Tag: Book tour. M1 stays off.
A masthead quoting a shop, then a second masthead quoting the first, is the loop. Tag: Loop. M1 goes on.
Shop class
Name the production document before naming the complex. Shop class is a routing fact, not a verdict.
Institute, advocacy NGO, corporate-funded think tank, litigant shop, agency memo. These are the production shops. Funding, donor review, and a pre-committed conclusion belong in the routing sentence. They are not proof the number is wrong.
Academic. A university press book or a peer-reviewed paper by named authors. It still has to be opened. It does not inherit the shop tag by being cited.
None. The unit is the primary: statute, docket, dataset, transcript.
A newsroom translating a named professor is a middleman under Domain 2 and R5. It is not a production shop. Do not stretch “policy institute” to cover a byline.
What EIC-1 does
Collapses an echo into one evaluation object. Secondary cards add zero weight. AIS does not rise because three mastheads cited the same unopened PDF.
Prints M1 beside the parent flags. It increments the flag count. It does not cap a domain.
Blocks ingestion of a C1 card. The card may be quoted for what it distributed. It is not stored as a factual premise. Route, or do not file.
Forces the routing sentence onto the production document, the statute, and the official score.
What EIC-1 does not do
It does not decide the claim is false.
It does not clear First Look or Archive Grade.
It does not average a tag into AIS or PAS.
It does not treat donor alignment as a kill switch.
It does not replace C1, Layer C, or K1. Fewer than two outlet cards on the same document: write “EIC-1 not in season” and stop.
It does not score a book tour as a loop.
Print block
Add this block under the parent scoresheet only when the sheet opens with “EIC-1 on.” Otherwise omit it.
EIC-1: on
M1: Y / N
Production document: title, author, year, linked Y / N
Shop class: institute / NGO / think tank / litigant / agency / academic / none
Echo: other cards in the window; do they cite the document, or cite each other?
Tag: Loop / Book tour / Not in season
AIS / PAS: the parent numbers, unchanged
Ingestion: C1 card stored as premise? Y / N — default N
Routing sentence: document, statute, score. Not the next masthead.
Routing protocol
When M1 is on, or when the tag is Book tour, ignore the middleman’s causal framing and the card headline. Then:
Extract named entities, dates, numeric baselines, and statutory references. Discard the moral.
Open the production document. If it is academic, open the paper or the book, not the review.
Open the primary the document itself relies on: statute, agency score, docket, dataset. A paraphrase of IRS records is not the records.
Isolate the contested claim. An aggregate can be a claim to check. A value-judgment built on that aggregate is not a second fact.
Log the numbers as claims to check until that primary is open. Do not promote them because a second masthead repeated them.
Calibration
These three units fix the gate. Do not re-litigate them by stretching the shop list.
Object. The New York Times social card, 2 October 2026, “The Rich People Democrats Aren’t Talking About.”
Unit: card. C1 already lawful (R6 = Y). EIC-1 adds the ingestion block and nothing else. Do not store the responsibility sentence as a premise. Parent card AIS stands. M1 is not reached: a card is not a second corroborating masthead of itself.
Object. Andrew Duehren, “The Rich People Democrats Aren’t Talking About,” The New York Times, 2 October 2026.
Unit: feature. Production document: Owen Zidar and Eric Zwick, The Everywhere Millionaire. Shop class: academic. Linked from the inspectable page: no. Echo: Atlantic, NPR, and the New Yorker wrote up the same book. They cite the book, not one another as the record. Tag: Book tour. M1: N. Domain scores unchanged. AIS remains the parent Cross-Examine. Do not cap Domain 2.
Object. Owen Zidar and Eric Zwick, “Americans Don’t Understand Who’s Rich,” The Atlantic, 14 September 2026.
Unit: author essay. The byline is the production. Shop class: academic. M1: N. EIC-1 does not improve this unit into Archive Grade. Adverse ground for the deduction, the statute, and the official score are still parent-gate questions.
Worked loop
Use this shape when the gate does fire. Do not staple it to the calibration objects.
A shop PDF states a finding. Outlet A translates it into a feature and a card, without a link to the underlying docket or dataset. Outlet B cites Outlet A. A third card cites “outlets of record.” Under parent WWS the second and third cards can look like sourcing. Under EIC-1 they are one object. M1: Y. Tag: Loop. AIS and PAS remain whatever the parent sheet scored on Outlet A’s text. The echo adds zero. Routing sentence: open the PDF, then the docket or dataset the PDF claims to rest on. The finding is a claim to check. The loop is not confirmation.
Core commitments carried forward
Unchanged from WWS v1.0: no partisan scoring; no credit for sincerity; no treating comment-thread or masthead consensus as evidence; event and package stay separated; a weak package is not proof the event is false; a strong package is not the record.
EIC-1 is a tag. It is not a tower.
No comments:
Post a Comment