# Responsible America — Editorial Policy

Status: internal operating procedure for anyone drafting, reviewing, or
approving reform pages. This document is self-contained: it defines
the workflow states, the required page structure, and the language
rules that apply to every reform record. It cross-references
`docs/governance.md`-style role concepts conceptually, but does not
depend on that document existing to be followed.

Last reviewed: 2026-07-30.

## 1. The editorial state machine

A reform record's `status` field (see `reform.schema.json`) must be one
of exactly: `draft`, `research_review`, `legal_review`, `fiscal_review`,
`editorial_review`, `approved`, `published`, `archived`. Movement
between states is one-directional except for corrections, which reopen
a `published` record back to the appropriate earlier state rather than
being edited in place.

### draft → research_review
**What must be true:** the record has values for the required schema
fields (`reform_id`, `reform_number`, `slug`, `title`, `short_title`,
`section`, `summary`, `problem_statement`, `current_system`,
`proposed_solution`, `implementation_method`, at least one supporting
and one opposing argument) and a first-pass answer to every question in
the Responsible America Test — even a rough answer, since research
review exists to strengthen weak ones. `fact_check_status` is
`pending_review`. `publication_status` is `draft`.
**Who moves it:** the drafting author (human or AI, subject to
`docs/ai-usage-policy.md`) requests the move; a research reviewer
(any team member designated to check sourcing) accepts it by changing
`status`.

### research_review → legal_review
**What must be true:** every material factual claim referenced in the
page has a corresponding claim record with `source_ids` populated from
sources actually retrieved and reviewed (`docs/fact-checking-policy.md`,
`docs/source-policy.md`), and no claim is left at `unsupported` or
`rejected` in the visible draft text — such claims are removed or
rewritten before this gate. `constitutional_amendment_required`,
`federal_legislation_possible`, `state_legislation_possible`,
`local_action_possible`, and `executive_action_possible` are set with a
real basis, not a guess.
**Who moves it:** a research reviewer signs off and hands to a legal
reviewer.

### legal_review → fiscal_review
**What must be true:** the record passes the checklist in
`docs/legal-review-policy.md` in full — legal analysis vs. legal
advice is distinguished, any statement about a named real person's
conduct is sourced to a checkable record, opinion is clearly framed as
opinion, and constitutional/statutory citations are accurate to the
actual text. A legal reviewer records their name and date.
**Who moves it:** a designated legal reviewer only. No other role may
advance a record out of `legal_review`.

### fiscal_review → editorial_review
**What must be true:** `estimated_federal_cost`, `estimated_state_cost`,
`estimated_private_cost`, and `estimated_savings` are each either
populated with a `status`, range/point figure, `period`, and `basis`,
or explicitly set to `not_applicable` — never left blank. Any savings
figure is labeled estimated vs. guaranteed per
`docs/research-methodology.md` Section 7, and any official government
score (CBO/GAO/JCT) is captioned separately from internal modeling. A
fiscal reviewer records their name and date.
**Who moves it:** a designated fiscal reviewer only.

### editorial_review → approved
**What must be true:** the full 23-point structure (Section 2, below)
is present and complete; the language rules (Section 3) are checked
line by line; `strongest_supporting_argument` and
`strongest_opposing_argument` are genuinely the strongest available,
not straw versions; and — for anything currently in Level 3 (Issues
Laboratory) seeking to move to Level 2 — the 28-question Responsible
America Test has real, specific, non-placeholder answers throughout,
not generic restatements of the problem.
**Who moves it:** a designated editorial reviewer, distinct from the
drafting author.

### approved → published
**What must be true:** `content_version` is set following semantic
versioning, and a `change_log` entry exists recording the approval
(date, version, summary, reviewer). `publication_status` moves to
`published` at the same time `status` does — the two fields are kept
in sync at this transition.
**Who moves it:** per `pd-086-publication-authority` (resolved Phase 9,
Owner Decision Session 1, 2026-08-01, Option B): the project owner
remains the ultimate publication authority at all times. The owner may
also designate one or more editors with bounded publication authority,
but only where the delegation is written, limited, revocable, and
restricted to reforms that have already cleared every recorded review
gate (legal, fiscal, specialist, and editorial). A delegated editor may
never alter an adopted policy position. As of this resolution, no
delegation has actually been made — `ra-record-review.mjs`'s
`ROLE_REQUIREMENTS` still requires `reviewer_role: owner` for
`publication_approval`; extending it to a specifically named, bounded,
revocable delegated role is a distinct future action, not something
this decision performs by itself.

### published → archived
**What must be true:** the reform has been superseded, withdrawn, or is
no longer an active platform position, and a `change_log` entry
explains why. Archived pages remain queryable; they are not deleted.
**Who moves it:** the project owner, or an editorial reviewer acting on
the owner's documented instruction.

### Reopening a published record for correction
An error found after publication does not get silently patched. The
record returns to the earliest state the correction actually requires
(a sourcing error returns it to `research_review`; a language-rule
violation returns it to `editorial_review`; a cost error returns it to
`fiscal_review`), a `change_log` entry documents the correction and its
reason, and the record proceeds back through every subsequent gate
before `published` again.

## 2. The 23-point reform-page structure

Every published reform page must contain all 23 of the following
sections, in this order. Each maps to specific `reform.schema.json`
field(s):

1. **The Problem** — `problem_statement`
2. **Why It Matters** — `problem_statement` plus scale/evidence framing (backed by claim records of type `statistic`/`historical_fact`)
3. **How the Current System Works** — `current_system`
4. **Who Benefits** — drawn from Responsible America Test answers on who benefits from the present system; not a dedicated schema field, sourced from `historical_background`/`current_system` plus claim records
5. **Who Bears the Cost** — same basis as #4, distributional framing; informed by `distributional_effects`
6. **The Proposed Reform** — `proposed_solution`
7. **What It Would Change** — `proposed_solution` narrowed to concrete effects
8. **What It Would Not Change** — explicit scope limits; drafted from the Responsible America Test question "What would it leave unchanged?"
9. **Constitutional and Legal Issues** — `constitutional_authority`, `constitutional_amendment_required`, `relevant_statutes`, `relevant_regulations`, `relevant_court_cases`
10. **Estimated Cost** — `estimated_federal_cost`, `estimated_state_cost`, `estimated_private_cost`
11. **Possible Savings** — `estimated_savings`, `funding_options`
12. **Arguments Supporting It** — `supporting_arguments`, `strongest_supporting_argument`
13. **Arguments Opposing It** — `opposing_arguments`, `strongest_opposing_argument`
14. **The Responsible America Compromise** — `proposed_compromise`
15. **Risks and Unintended Consequences** — `implementation_risks`, `unintended_consequences`, `abuse_scenarios`, `civil_liberties_risks`
16. **Safeguards** — `safeguards`
17. **How Success Would Be Measured** — `success_metrics`, `review_schedule`, `sunset_rule`
18. **Current Bills and Government Action** — `current_legislation`
19. **Where Elected Officials Stand** — `politician_positions`, `public_support_data`
20. **What Citizens Can Do** — `citizen_actions`, `contact_targets`
21. **Sources** — `sources`, `primary_sources`, `secondary_sources`
22. **Last Reviewed** — `last_fact_checked`, `next_review_date`
23. **Revision History** — `change_log`

A page is not published with any of these 23 sections silently blank.
If a section genuinely does not apply (for example, a reform with no
current bill pending has nothing for #18), the section states that
plainly ("No current federal legislation identified as of
[as_of_date]") rather than being omitted.

## 3. Language rules

Reform pages are written direct, factual, readable, respectful, firm,
specific, and transparent about uncertainty. Concretely:

- State the claim and its evidence; do not editorialize about how
  obvious or self-evident it is.
- Use absolute dates ("as of March 2026," not "recently" or
  "currently").
- Label every dollar figure with its period ("$14B annual," "$140B
  over 10 years," "$2B one-time") — a bare dollar figure with no period
  is not permitted.
- Where evidence is incomplete or mixed, say so in the text itself
  rather than presenting a single confident number.

**Phrases and patterns that are not used on this platform:**

- "Everyone knows," "obviously," or similar appeals that substitute for
  evidence.
- Absolute generalizations about a group — "all corporations," "all
  politicians," "the media always," "the government never." Claims are
  scoped to what the evidence actually supports (a named agency, a
  specific pattern, a bounded time period).
- "This proves" attached to evidence that only supports, suggests, or
  is consistent with a conclusion — reserved for claims with direct,
  dispositive evidence.
- Treating political disagreement as disloyalty or treason.
- Accusations of corruption, bribery, or criminal conduct without a
  specific, checkable source for the assertion (see
  `docs/legal-review-policy.md`).
- Dehumanizing language about any group or individual.
- Sensational or definitive claims about UAPs beyond what the cited
  government records actually state (see section S of the taxonomy,
  `unsupported-claims`).
- Casual comparisons to dictatorship, fascism, or authoritarian regimes
  used as rhetorical shorthand rather than a documented, sourced
  parallel.

## 4. Level 3 → Level 2 promotion

A reform in the Issues Laboratory (Level 3) is promoted to the
numbered Renewal Plan (Level 2) only by passing every gate in Section 1
through `approved`, with genuine, specific, non-placeholder answers to
all 28 Responsible America Test questions. "It sounds attractive," a
compelling anecdote, or momentum from public interest are not
sufficient grounds for promotion on their own — the legal, fiscal, and
editorial reviews must each independently sign off.