Roadmap
What is built, what is left, and what only a human can close.
Standing rule: no scores. Color marks whether something is in force, reversed, contested, concluded, or permanent, not good or bad.
New here? Read this first (30 seconds)
- Color is a legal fact, not an opinion. It shows (whether something is still in force, was reversed, or is contested in court), never whether it was good or bad.
- Every number links to its source and to how strongly it can be tied to the action (). Click any figure to see the full evidence.
- Nothing here is scored, graded, or ranked. When two credible sources disagree, this archive shows both numbers instead of picking one.
- Wins are in here too. See or filter the Ledger by .
- Nobody is asking you to trust a byline. There isn’t one — the archive is anonymous, on purpose, which is exactly why it has to be checkable without one. See and check one entry against its own source yourself.
Build status
54 of 62 tasks complete
8 remaining across 8 phases.
4 more the maintainer can build directly4 need you — people, lawyers, hosting
87%
Phase 1 — Methodology
Done
Complete5/5
✓Four-layer model: action, status, effect, valuation
✓Four-level attribution scale, applied to every claim
✓Source tier system with orientation labels
✓Rules for handling source disagreement — never average estimates, show the spread
✓Methodology stated publicly as its own page
Phase 2 — Core archive
Done
Complete6/6
✓97 documented entries, each carrying all four layers
✓Filterable ledger with search and standing filters — live
✓Entry detail drawer with sources, effects, revisions and cross-links — live
✓Citable permalink per entry (?entry=id) — live
✓Timeline with zoom, pan, era bands and a density minimap — live, lane-packed at every zoom level
✓Every view with real source wired into the live navigation registry — all 20 reachable
Phase 3 — Legibility
Done
Complete5/5
✓Standing color system (factual, not evaluative)
✓Headline figures with direction and incidence
✓Estimate-spread charts, unaveraged
✓Attribution and evidence-composition charts
✓Standing indicators strip with per-figure vintage dates
Phase 4 — Comparison & narrative
Done
Complete6/6
✓Entry-vs-entry on identical fields
✓Cross-president metrics with counting rules and hatched partial terms
✓Claims tally kept separate from actions
✓Dossiers: sequential accounts of the four largest threads
✓Patterns: findings visible only across the whole dataset
✓Wiki cross-links with automatic backlinks — live in the ledger and the entry drawer
Phase 5 — Evidence quality
Done
Complete8/8
✓Every entry carries at least two source documents — verified live, 0 of 97 fall short
✓Every effect claim carries at least two estimates — verified live, 0 of 203 fall short
✓Live integrity audit running against the archive’s own data
✓Six-lens internal review published in full, with an explicit non-independence notice
✓Controlled source-lean vocabulary and unified spelling across the archive
✓Provisional-flag system for facts still moving
✓Public corrections ledger with published error rate
✓Versioned entry history — the CMS append-on-save mechanism is live; seeded history is still 7 of 97 entries because the rest cannot be backfilled honestly and accrue only from real edits
Phase 6 — Coverage
Me, on request
In progress12/14
✓97 entries across all 11 domains
✓Thin domains filled: healthcare, energy, public health
✓Promises and public claims tracked as their own units
✓Regulatory, education, AI, crypto, deployment, elections, clemency and agency coverage added
✓First-term coverage: travel ban, DACA, census, Soleimani, North Korea, Syria, Title IX, WOTUS
✓Commonly misread view — contested ground from every entry in one place
✓Healthcare, energy and judiciary depth: Medicaid, ACA subsidies, methane, NEPA, immunity, CASA
✓COVID response, Phase One China deal, impoundment, DEI, press access, emoluments added
✓Fabricated-figure corrections, including inflated claim counts
✓Fixed app frame — sidebar and content scroll independently — live
✓Structured datasets built from primary sources by script rather than by hand — Senate roll-call confirmations, Federal Register tariff actions, and FRED economic series
✓Indicators: consumer prices, petrol, unemployment, payrolls, mortgage rates, the 10-year yield, federal debt and house prices, each sliced by presidential term — with rates compared in percentage points, never percent
○Continued expansion as new actions occur — the watchtower now surfaces candidates rather than waiting for someone to noticeMaintainer can build directly
○Lower-court and regulatory actions beyond the Supreme CourtMaintainer can build directly
Phase 7 — Editorial operations
Me, then you
In progress8/11
✓Accessibility: skip link, main landmark, drawer focus management, AA contrast and target size — verified live
✓AAA target size (44px) on standalone filter and legend controls — measured live across all views, with one documented exception (the compact search-results Clear button)
✓Correction submission backend: validated endpoint, rate-limited, status/response tracked — verified live end-to-end
✓Status-refresh queue persisted with an owner field and overdue alerts. Status dates update automatically once a day, with nobody needing to remember — verified live against the database
✓A back-end editing tool so non-engineers can maintain entries. Only people the maintainer has explicitly approved can log in and edit, and every save is checked against the archive's own rules before it can publish — verified live end-to-end
○Second-reviewer sign-off before publication — states live in the CMS, self-review blocked (same GitHub login can’t edit and sign off), still unstaffedNeeds you
✓Watchtower: a daily scan of the Federal Register and the Senate roll-call record files new primary documents into an admin review queue with the document attached — it publishes nothing on its own
✓Link rot checker built and run manually (`npm run data:check-links`) across every source URL, bot-protected hosts reported separately from genuinely dead ones so real failures are not buried
○That link rot checker wired into the daily cron so it runs on a schedule instead of only when someone remembers to invoke it by hand — not done yet, nothing in /api/cron/daily calls itMaintainer can build directly
✓Bare-domain sources carry a structured provenance note explaining why they resolve to a publisher, enforced by two tests: every one annotated, and the count only ever falling
○A named person to work the refresh queue the site now generates — the admin UI to assign one is live; with one maintainer there is nobody to assign toNeeds you
Phase 8 — Launch prerequisites
Me, then you
In progress4/7
✓Masthead states what this actually is: no company, no funding, no staff, one anonymous editor, and the correction form as the only contact route
✓Hosting — deployed and live
✓Self-integrity snapshot: the archive's own entry page is captured to the Internet Archive on every publish and every accepted correction, verified end to end with a real capture that reached success — tamper-evidence that an entry was not silently edited after publication
○Archiving the ~227 externally cited source URLs themselves, so a citation still resolves after the original page moves or disappears — not built yet; today's snapshot integration only ever captures this site's own pagesMaintainer can build directly
✓Entity structure and media liability insurance — not applicable. There is no business and there will not be one; insurance is written for entities. The personal-exposure point it implied is real and belongs to the media-law item below
○Media-law review of a sample of entries — and it matters more without a company, not less: with no legal entity there is nothing between the archive and the person maintaining itNeeds you
○A human monitoring corrections with a published response-time SLA — the intake, rate limiting, status field and admin review queue are real, and one person reads themNeeds you
The honest bottleneck
All 20 views are live, and the CMS, correction backend, snapshot archiving and second-reviewer tooling work end-to-end against the real database — verified by clicking through them, not by reading the code. A site like this fails on maintenance rather than on design: status fields rot within weeks, and an archive with stale statuses is worse than none, because it looks authoritative while being wrong. Three things now run on a schedule against that: the market and economic series refresh themselves, the refresh queue recomputes which entries are due for a recheck, and a daily watchtower files new primary documents into a review queue so the archive notices a change before a reader does. None of it publishes anything on its own. A fourth check — sweeping every cited source URL for link rot — exists and works, but only as a script run by hand; it is not yet on that schedule. What is genuinely left is not code — a media-law read, and a person answering corrections on a published schedule. The amber items above are ones an engineer can keep working; the red ones need a person.