A proposal for how a government could make its own check register publish itself — and an honest account of how far the idea reaches.
Today this platform, like every transparency effort, does forensic reconstruction: it chases a receipt after the money moves, through minutes, audits, check registers, and records requests — and it hits dead ends exactly where that trail was never written down. The idea here flips the order: if a dollar can only leave through an authorized account tied to a specific budget line, then the transaction is the record, generated the moment the money moves, with no separate reporting step to lose it. Transparency stops being a favor someone chooses to grant and becomes a byproduct of paying at all.
This is not science fiction. Every piece already runs in production somewhere. No one has assembled them into one object, and — told honestly — the idea reaches only a slice of a budget. Both of those things are true at once, and the pitch is only credible if it says so.
Pay each procurement budget line from its own authorized account or virtual card, whose transaction feed publishes by default — so the line and the receipt are the same object.
| Piece of the idea | Where it already runs | The number |
|---|---|---|
| A government card rail that captures the receipt natively | GSA SmartPay (federal purchase/travel/fleet cards) — captures merchant, category code, amount, date, and at Level 3 full line-item detail | $39.4B across ~82M transactions, FY2025 (GSA) |
| An account that enforces what it can buy | Merchant-category controls decline non-conforming charges at the point of sale; the Army hard-blocks high-risk categories bankwide | Standard control today (AFARS 14-6) |
| Purpose-restriction at national scale | SNAP EBT; and WIC, which matches each scanned item's code against an approved-product list in real time — the true item-level precedent | SNAP $99.8B / 41.7M people, FY2024 (USDA) |
| Publishing every transaction by default | Oklahoma posts item-level purchase-card transactions, filterable by cardholder, merchant, and item | Live dataset, updated monthly (data.ok.gov) |
| Publishing at state scale, and its payoff | OhioCheckbook put every state transaction online | 112M transactions; Ohio went from 46th to 1st in national transparency rankings (Buckeye Institute) |
| Ring-fencing each appropriation line | Encumbrance accounting — a purchase order reserves budget against the line before a dollar is spent | Standard government practice (primer) |
| Attribution that travels with the money | ISO 20022 on FedNow / RTP can carry structured line and purchase-order data with the payment | Live rails, now $10M limits (Fed) |
The floor of this idea — a downloadable, searchable record of every check — is already solved by real Texas governments, and Texas even certifies it through the Comptroller's Transparency Stars program (the "Contracts & Procurement" star requires a machine-readable vendor check register, not a PDF) (program).
But here is the honest frontier: every one of them stops at the check level — who got paid, how much, when, roughly for what. No Texas local government publishes its purchase-card transactions at the item level. The only government that does is out of state — Oklahoma, which posts every state P-card purchase, itemized (OK). So Texas's best counties define the floor receipts-by-rails should assume as already-won — the downloadable, awarded check register — and the open frontier they have all left, item-level and rail-native disclosure, is exactly the standard this idea names.
Nobody has collapsed the encumbrance and the card into one object — a virtual card whose spending limit is the encumbered budget line, so the swipe simultaneously liquidates the encumbrance, records the expenditure, attaches the receipt, and renders it publicly at the line level. Procurement-tech firms do purchase-order workflow; audit firms review spend after the fact; no one issues per-line spendable cards. A white-label card issuer on a community bank is the plausible way a small county could stand one up.
1. It reaches only ~15–30% of a budget. Payroll (~41% of local government spending), debt service, transfers to schools and cities, and large capital all move by wire to entities that cannot accept cards. This rail illuminates the discretionary/procurement lines — exactly where the dead ends are — and cannot touch the big-dollar lines (which are already well documented). 2. Capture is not disclosure. The federal government captures all of this and publishes only aggregates. Publish-by-default is a policy choice; opacity is the norm, and Oklahoma is the exception, not the rule. 3. The receipt is not proof. The Government Accountability Office found 41–48% of purchase-card transactions failed basic controls (GAO-08-333). Native data lowers the cost of finding the receipt; it does not prove the purchase was legitimate. 4. Every enforcement mechanism is also a control mechanism. The "programmable money is surveillance" objection is live — the Anti-CBDC Surveillance State Act passed the U.S. House 219–210 in July 2025 (H.R.1919). Anything that can block a government payment draws that fire. The frame must be audit-and-anti-diversion, not programmable control.
Not "a prepaid card for every line." That breaks on payroll, debt, and contracts, and hands critics the surveillance argument. Instead:
An encumbrance you can spend from, that emits its own public receipt — a virtual card per procurement line, on a pooled or white-label issuer, scoped to the operating slice, with the transaction feed published by default. Vendor, amount, and budget-line attribution are native and enforced; whether the stapler charged to the stapler line was really a stapler stays an audit question (only one program, WIC, does item-level enforcement, and only on a tiny standardized product set).
County Commons does not need to build payment rails. Its role is three things:
The difference is visible at the level of a single line.
A reconstructed dead end (today): a line reads INCENTIVES — $1,840,000, and no public document names a recipient. The trail stops. Filling it takes a records request, a meeting-minutes pull, or a counter visit — and often it stays dark.
A rails-native line (under this proposal): the same line publishes its own itemized transactions — payee, amount, date, category — the moment each payment clears. No request required. The dead end becomes a list.
The coverage score is just this, counted: what fraction of the county's spending is already the second kind.
Every factual claim above links to a primary source. This brief is a starting point, not the last word — it is filed as a solution on the platform so anyone can weigh it, cite against it, or file a better one.