Public checks

Data integrity

Whether the archives behind The Books, The Floor, The Gavel, The Desk, and The Race still hold together. Technical companion to Methodology. Archives live in the data cache; last automated refresh is on status.

01 · Status

Verdict, open findings, and what Pass / Advisory / Fail mean.

Live verdict

Current build

Advisory

Critical checks passed. 11 advisory findings are listed below. Those are documented source quirks, not silent failures.

State data years: 2023, 2022, 2021, 2020, 2019, 2018, 2017, 2016
Primary state source: New York State Office of the State Comptroller
Check tally: 7 pass · 5 advisory · 0 fail
OSC watch checked: Aug 4, 2026, 5:09 PM UTC

Open advisory findings

Advisory

In plain terms: these are the reasons the verdict is Advisory instead of Pass. Each finding says what still holds and what to treat as approximate.

CATEGORY_PARTS_EXCEED_SPEND · ME
- Category parts exceed federal spending for ME (2016) by $5,427,822,708
Reliability
The state’s total federal spending, revenue paid, net balance, rank, and return per dollar still come from OSC workbook totals. Those stay on the critical path. What breaks down is the category split: program lines already add up to more than total spending, so the leftover “other” line is set to zero.
Affects
Spending-by-category charts for that state and year. Not the ranking or map.
REVENUE_TYPE_SUM_MISMATCH · 4 years
- FY2018: byType sum $3,324,242,420,390 vs totalReceipts $4,107,741,496,584
- FY2021: byType sum $4,040,349,490,934 vs totalReceipts $6,818,157,647,017
- FY2022: byType sum $4,889,824,615,926 vs totalReceipts $6,271,507,596,876
- FY2023: byType sum $4,432,379,961,193 vs totalReceipts $6,134,432,040,451
Reliability
The national headline receipt total (totalReceipts) is still the figure for “how much came in.” The tax-type breakdown (byType) comes from a related but not identical Monthly Treasury Statement cut, so the typed pieces will not add up to 100% of the headline.
Affects
Revenue composition charts for the flagged fiscal year. Headline totals and state IRS collections are separate series.
ISSUES_NO_VOTES · 2 findings
- fairtax-act-119-hr-25: no archived House or Senate votes
- share-savings-seniors-118-hr-5376: no archived House or Senate votes
Reliability
Bill identity, pipeline stages, and Congress.gov summaries can still be sound. This flag means no House or Senate roll calls were archived for that watchlist bill.
Affects
Floor Votes and Members indexes for that bill.
GAVEL_NO_ARGUMENTS
- 534 of 627 docket cases have no CourtListener oral-argument audio. Audio is only expected after argument when CourtListener returns a recording; pending, unargued, and unenriched rows are normal, and the Court did not systematically record arguments for most of its history.
Reliability
Case identity, citations, and opinion clusters can still be sound. CourtListener has no oral-argument audio for those dockets yet. That is expected for pending or unargued matters, unenriched watchlist rows, and the long era before the Court routinely recorded arguments.
Affects
Gavel Arguments rows for cases without a recording.
GAVEL_NO_FULL_TEXT
- 102 of 627 docket cases have no locally archived opinion text yet (full-layer PDF / CourtListener extract still pending for those rows).
Reliability
Metadata and outbound links can still be sound. Opinion text lands when the full-layer fetch extracts a CourtListener opinion or SCOTUS PDF; many recent docket stubs are still waiting on that pass.
Affects
In-app opinion reading on Gavel case detail pages.
GAVEL_NO_PDF
- 565 of 627 docket cases have no locally archived opinion PDF yet.
Reliability
Metadata can still be sound. A local SCOTUS PDF is archived when the download succeeds; cases without one still link out to the Court.
Affects
Local PDF links on Gavel Docket / Opinions.
OSC_SOURCE_WATCH_STALE
- All 8 archived OSC workbooks match osc.ny.gov. Last checked Aug 4, 2026, 5:09 PM UTC. Watch result is 40 days old (older than 30). Re-run npm run data:check-osc.
Reliability
The last successful remote hash check is older than the staleness window. Archives may still match OSC. We have not re-verified recently.
Affects
How fresh the OSC source-watch signal on this page is. Re-run npm run data:check-osc.

What the labels mean

Every automated check on this page, and the live verdict above, uses one of three labels.

Pass: This check cleared. For that rule, the archived numbers match what the site expects.
Advisory: Something in the source data does not line up cleanly. We keep the finding visible. Critical ranking rules can still pass.
Fail: A critical rule broke. Treat site numbers as under review until this clears.

Advisory is not a soft fail. It means we found a quirk in the source material, wrote it down, and still consider the critical ranking rules intact. Fail means stop and review the build.

02 · Evidence

Shipped archives, OSC hash watch, and every automated check.

Archives in this build

The Ledger ships local JSON archives for each path. Integrity checks below run against these shipped files, not against a live scrape on every page view. You can download the same files from the data cache.

The Books · States
8 OSC years (2016–2023)
New York State Office of the State Comptroller
The Books · Flow
8 revenue · 64 CBO · 7 award · 8 outlay years · 73 agency profiles
Treasury / CBO / USAspending archives
The Floorfetched 2026-08-07T16:47:47Z
22 bills · 93 roll calls
Library of Congress Congress.gov API
The Gavelfetched 2026-08-07
627 cases · 525 with text · 62 with PDF · 9 justices
CourtListener bulk data (Free Law Project) + SCOTUS PDFs
The Deskfetched 2026-09-13T17:21:09Z
47 presidencies · 20 curated actions
The Ledger Desk watchlist + U.S. presidency catalog
The Racefetched 2026-09-13T15:34:09Z
1996/1998/2000/2002/2004/2006/2008/2010/2012/2014/2016/2018/2020/2022/2024/2026 · 8,106 seats · 44,538 candidates · 4,601 donor pages · 13,513 with votes
FEC bulk (weball + pas2 + cm) + FEC presidential summary tables + FEC/MEDSL election results + state primary adapters

Live election results

The Race map has Live and Archive modes. Live overlays Senate and House tallies from civicAPI, an unofficial aggregator of state reporting; race calls are civicAPI's. Archive uses FEC Official presidential results for 1992–2024 electoral votes (Certificates of Vote / NARA) and FEC / Clerk Senate and House winner archives by cycle. Archive finance and certified or best-available vote attachments on The Race remain FEC / state SOS / MEDSL and are not rewritten on every live poll. Registry: 1,224 race ids (built 2026-09-13) · 48 states · House districts 429/435 · no feed yet: ND, VT.

OSC source watch

State rankings come from Excel workbooks published by New York’s Office of the State Comptroller. We keep copies of those files in the project. A script compares a SHA-256 hash of each local file to the live download on osc.ny.gov.

If the hashes match, the site still reflects the current published workbook for that year. If they differ, that is a review signal. We do not replace archives automatically.

Remote vs archive

Advisory

All 8 archived OSC workbooks match osc.ny.gov. Last checked Aug 4, 2026, 5:09 PM UTC.

FFY 2016Pass
FFY 2017Pass
FFY 2018Pass
FFY 2019Pass
FFY 2020Pass
FFY 2021Pass
FFY 2022Pass
FFY 2023Pass

Year	Local hash	Remote hash	Status
FFY 2016	bee458cc0943…	bee458cc0943…	Pass
FFY 2017	0633f9558c2d…	0633f9558c2d…	Pass
FFY 2018	3bc6cb508abd…	3bc6cb508abd…	Pass
FFY 2019	2143fa57add6…	2143fa57add6…	Pass
FFY 2020	7080059645b7…	7080059645b7…	Pass
FFY 2021	2116c5f27e81…	2116c5f27e81…	Pass
FFY 2022	9c1dd3c80cba…	9c1dd3c80cba…	Pass
FFY 2023	d816c6ca2d76…	d816c6ca2d76…	Pass

This watch compares byte hashes of published Excel files to the in-repo archives. A match means the site's state rankings still reflect the current OSC file for that year. Drift does not auto-update The Ledger; rebuild after human review. Also watching FFY 2024, 2025 for first publication.The watch also checks years that are not in the States list yet: FFY 2024 (not published); FFY 2025 (not published).

Automated checks

These checks run against the JSON archives shipped with the site: OSC state years, Books flow datasets, the Floor bill watchlist, the Gavel docket, The Desk presidencies catalog, and The Race elections archive. A critical failure in the state load path is meant to stop a broken state dataset from rendering as a normal page. Floor, Gavel, Desk, and Race checks surface watchlist sync and archive completeness on this page.

7passed

5advisory

0failed

Books · states

01Pass
state-count
Fifty-state coverage
FFY 2023: 50 states · FFY 2022: 50 states · FFY 2021: 50 states · FFY 2020: 50 states · FFY 2019: 50 states · FFY 2018: 50 states · FFY 2017: 50 states · FFY 2016: 50 states
Rulestates.length === 50
02Pass
derived-metrics
Derived metrics match formulas
Net balance, per capita, return per dollar, and status are recomputed from revenue, spending, and population on every load.
Rulenet = spending − revenue; perCapita = net / population; return = spending / revenue; status from net sign
03Pass
ranks
Rank order
Rank 1 is the lowest (most negative / strongest net contributor) net balance; ranks are unique 1–50.
Rulesort by netBalance ascending → rank = index + 1
04Advisory
categories
Spending category residuals
CATEGORY_PARTS_EXCEED_SPEND: Category parts exceed federal spending for ME (2016) by $5,427,822,708
RulesocialSecurity + medicare + medicaid + defense + other ≈ federalSpending
11Pass
golden-pins
Golden OSC pins
All 5 pinned OSC figures match the shipped archives.
Rulerevenue, spending, population === pinned workbook cells
12Advisory
osc-source-watch
OSC remote source watch
All 8 archived OSC workbooks match osc.ny.gov. Last checked Aug 4, 2026, 5:09 PM UTC. Watch result is 40 days old (older than 30). Re-run npm run data:check-osc.
RuleSHA-256(local archive) === SHA-256(osc.ny.gov Excel)

Books · flow

05Pass
congress-deficit
CBO deficit identity
Holds for all 64 historical CBO years (± $2B).
Rulerevenues − outlays ≈ deficitOrSurplus (± $2B)
06Advisory
flow-schema
Books flow datasets
REVENUE_TYPE_SUM_MISMATCH: FY2018: byType sum $3,324,242,420,390 vs totalReceipts $4,107,741,496,584 · REVENUE_TYPE_SUM_MISMATCH: FY2021: byType sum $4,040,349,490,934 vs totalReceipts $6,818,157,647,017 · REVENUE_TYPE_SUM_MISMATCH: FY2022: byType sum $4,889,824,615,926 vs totalReceipts $6,271,507,596,876 · REVENUE_TYPE_SUM_MISMATCH: FY2023: byType sum $4,432,379,961,193 vs totalReceipts $6,134,432,040,451
RuleZod parse on import + critical integrity rules

The Floor

07Advisory
floor-archive
Floor watchlist archive
ISSUES_NO_VOTES: fairtax-act-119-hr-25: no archived House or Senate votes · ISSUES_NO_VOTES: share-savings-seniors-118-hr-5376: no archived House or Senate votes
Rulewatchlist id/slug ⊆ build · stages contain currentStage · latestAction present

The Gavel

08Advisory
gavel-archive
Gavel docket archive
GAVEL_NO_ARGUMENTS: 534 of 627 docket cases have no CourtListener oral-argument audio. Audio is only expected after argument when CourtListener returns a recording; pending, unargued, and unenriched rows are normal, and the Court did not systematically record arguments for most of its history. · GAVEL_NO_FULL_TEXT: 102 of 627 docket cases have no locally archived opinion text yet (full-layer PDF / CourtListener extract still pending for those rows). · GAVEL_NO_PDF: 565 of 627 docket cases have no locally archived opinion PDF yet.
Rulewatchlist id/slug/docket ⊆ build · decided ⇒ dateFiled + opinions · SCOTUS URL present

The Desk

09Pass
desk-archive
Desk presidencies archive
47 presidencies · 20 curated actions · built 2026-09-13T17:21:09Z.
RuleZod shape · unique presidency/action slugs · watchlist action ids ⊆ build

The Race

10Pass
race-archive
Race elections archive
1996, 1998, 2000, 2002, 2004, 2006, 2008, 2010, 2012, 2014, 2016, 2018, 2020, 2022, 2024, 2026 cycles · 8,106 seats · 44,538 candidates · 5,142 with top donors · 4,601 donor pages · 13,513 with vote totals · 24 presidential rows on FEC PresCand tables · built 2026-09-13T15:34:09Z.
RuleZod shape · unique candidate/donor slugs · every candidate seatId ∈ seats · presidential finance prefers PresCand Table 1/2 when available

03 · Recompute

Formulas, golden pins, and a rank sample from both ends.

Formulas

On state pages, derived fields are not taken on faith from the JSON alone. When the archives load, the site recomputes:

Net balance: federal spending received − federal revenue paid
Per capita: net balance ÷ population
Return per $1: spending ÷ revenue
Status: contributor if net < −$1 · recipient if net > $1 · else balanced
Rank: sort by net ascending · lowest net = rank 1

For FFY 2023, the sum of state net balances is $1.3T. In a federal deficit year, a large positive national sum is expected. That sum is not treated as an error.

Golden pins

These are specific OSC workbook figures we store as known-good values. The list below compares them to the archives shipped with this build. If a rebuild swaps a column or year, the pins fail here and in the test suite.

California FFY 2023
Pass
Revenue
$589.5B
Spending
$601.4B
Net
$11.9B
Population
38,965,193
Minnesota FFY 2023
Pass
Revenue
$76.7B
Spending
$79.7B
Net
$2.9B
Population
5,737,915
Mississippi FFY 2023
Pass
Revenue
$22.2B
Spending
$58.5B
Net
$36.2B
Population
2,939,690
California FFY 2022
Pass
Revenue
$679.3B
Spending
$664.3B
Net
−$15.0B
Population
39,029,342
Texas FFY 2022
Pass
Revenue
$381.5B
Spending
$478.4B
Net
$96.9B
Population
30,029,572

Year	State	Revenue	Spending	Population	Implied net	Pin
2023	California (CA)	$589.5B	$601.4B	38,965,193	$11.9B$1.02 / $1	Pass
2023	Minnesota (MN)	$76.7B	$79.7B	5,737,915	$2.9B$1.04 / $1	Pass
2023	Mississippi (MS)	$22.2B	$58.5B	2,939,690	$36.2B$2.63 / $1	Pass
2022	California (CA)	$679.3B	$664.3B	39,029,342	−$15.0B$0.98 / $1	Pass
2022	Texas (TX)	$381.5B	$478.4B	30,029,572	$96.9B$1.25 / $1	Pass

Rank sample · FFY 2023

A sample from both ends of the FFY 2023 ranking: the strongest contributors and the strongest recipients. The recompute column checks that status and net balance still match the formulas above.

Strongest contributors

#1 New Jersey
−$17.7B · contributor
Pass
#2 Massachusetts
−$3.6B · contributor
Pass
#3 Washington
−$808.9M · contributor
Pass

Strongest recipients

#48 Texas
$76.9B · recipient
Pass
#49 Florida
$79.6B · recipient
Pass
#50 Virginia
$139.7B · recipient
Pass

Rank	State	Net	Return	Status	Recompute
Strongest contributors
#1	New Jersey	−$17.7B	$0.88	contributor	Pass
#2	Massachusetts	−$3.6B	$0.97	contributor	Pass
#3	Washington	−$808.9M	$0.99	contributor	Pass
Strongest recipients
#48	Texas	$76.9B	$1.20	recipient	Pass
#49	Florida	$79.6B	$1.26	recipient	Pass
#50	Virginia	$139.7B	$2.18	recipient	Pass

04 · Limits

What integrity does not force equal across federal datasets.

What we do not force equal

Integrity here means each series is consistent with its own archived source rules. It does not mean every federal dataset equals every other.

OSC state balances are not required to equal IRS gross collections or USAspending award totals for the same state.
Agency obligations from USAspending are not required to equal Treasury MTS outlays for the same agency name.
National MTS receipt totals are not required to match the sum of OSC state tax payments dollar for dollar.
Floor Featured bills are an editorial Start-here overlay on a Congress.gov inventory for declared congresses. Inventory rows are list-level; full roll-call archives attach to Featured bills. Senate XML may be missing when fetch could not reach it.
Gavel cases are an argued-merits docket with Featured editorial notes. Opinion text comes from CourtListener quarterly bulk dumps filtered to the watchlist, not a live scrape. Search metadata is not required to include full concurrence / dissent join lists, and oral-argument audio may be missing for some dockets. PDFs remain a smaller SCOTUS slip-opinion archive, not the full docket.
Desk actions are a curated watchlist, not a full Federal Register crawl. Presidency bios may omit fields the unitedstates executive file does not carry.
Race finance for House and Senate follows FEC weball. Presidential races prefer FEC Presidential Table 1/2 summary workbooks when available; weball alone undercounts some campaigns. Donor pages cover committees in top-donor lists, not every Schedule A individual. Live map tallies stay on civicAPI and are covered under Archives above.
Known residual: 2016:ME. For that state-year, OSC program parts already exceed total spending, so the residual other line is floored at zero. That is flagged advisory, not ignored.

Those boundaries are editorial rules. They are documented in Methodology. The verdict on this page reflects the archives in the current build, plus the last committed OSC hash check. It is not a live scrape of every .gov or CourtListener endpoint. If a critical check fails, a golden pin misses, or the source watch shows drift you have not reviewed, treat the numbers as under review.

Live verdict

Open advisory findings

What the labels mean

Archives in this build

Live election results

OSC source watch

Automated checks

Fifty-state coverage

Derived metrics match formulas

Rank order

Spending category residuals

Golden OSC pins

OSC remote source watch

CBO deficit identity

Books flow datasets

Floor watchlist archive

Gavel docket archive

Desk presidencies archive

Race elections archive

Formulas

Golden pins

Rank sample · FFY 2023

What we do not force equal

Live verdict

Open advisory findings

What the labels mean

Archives in this build

Live election results

OSC source watch

Automated checks

Fifty-state coverage

Derived metrics match formulas

Rank order

Spending category residuals

Golden OSC pins

OSC remote source watch

CBO deficit identity

Books flow datasets

Floor watchlist archive

Gavel docket archive

Desk presidencies archive

Race elections archive

Formulas

Golden pins

Rank sample · FFY 2023

What we do not force equal