Receipts — a claim-level evidence standard
Version 1.0 · 2026-06-23 · Brian Nuckols, Memory Hole / Hesitation Media
Status: citable specification. Outlet-neutral: any journalist or researcher may adopt it. Cite as Receipts: a claim-level evidence standard, v1.0.
What it is
Receipts is a way of publishing a claim so the reader can check it instead of trusting it. Every claim that carries the argument is graded in one plain word, shown with the proof it rests on, and marked with what would prove it wrong. The standard is descriptive, not inferential: it shows you what the conclusion is standing on and leaves the judgment to you.
One line: every load-carrying claim, graded and shown with the receipt behind it, so you never have to take the reporter’s word for it.
The five grades
| Grade | What it means to you |
|---|---|
| Established | Independently confirmed. You could verify it yourself today from the cited record. |
| Supported | Solid evidence with one soft spot: a single strong source, or one short step of reasoning. |
| Reported | On the record from a credible source, but not independently confirmed. |
| Thin | Rests on one weak or uncorroborated source, or a long interpretive leap. |
| Disputed | Credible evidence cuts both ways. |
How a grade is set: three questions
- Where is it from, and can they be trusted on this? A court filing and an anonymous tip are not the same footing, and a source can be reliable on one matter and not another.
- Who else confirms it, independently? Two articles citing the same press release are one source, not two.
- How far is the leap from the evidence to the claim, and could it be proven wrong? “The filing says X” is a short, testable leap. “Therefore the system is captured” is a long one. A claim that nothing could disprove earns no grade above Thin.
A claim is Established when all three are strong, and slides toward Thin as any one weakens. The two judgment questions (can they be trusted here; how big is the leap) are answered in writing before publication. That written answer is the claim’s receipt and its “breaks if” line.
How each claim is shown
Claim. the sentence being rated
Grade. one of the five words
Receipt. what it rests on, with the source archived where it was read and fingerprinted (hashed) so the copy you see is the copy that was used
Breaks if. the one thing that would overturn it
The evidence profile
The grades do not stay scattered through the piece. They are summarized.
Per investigation: mark the claims the conclusion depends on, then show the spread, e.g. “12 central claims: 7 Established, 2 Supported, 2 Reported, 1 Thin.” Render it as a single bar so a reader sees in one glance how much of the argument stands on confirmed ground.
The honesty flag: name the weakest claim the conclusion depends on, out loud, e.g. “The conclusion turns on claim #9 (Thin). If it falls, the piece falls.” No summary may hide which weak claim carries the roof.
Across a body of work: the running distribution of grades over time, the rate at which a thesis was downgraded or killed after review, and the standing correction rate. Descriptive statistics over the whole output.
To adopt it
Use the five grades and the per-claim format above, publish the evidence profile and the honesty flag on each piece, and archive every receipt where it was read with a published hash. You may say “graded under Receipts v1.0” without affiliation to Memory Hole. The grades are the standard; the name is free to travel.
Migration from the prior system
Receipts replaces the MHEES six-axis code and the NATO Admiralty two-axis grade outright. The retired
notation (O·S·C·K·D·T with daggers; Admiralty letter-number grades; the EJ-1 E1–E5 public tier) maps
forward as follows, and existing graded ledgers are re-mapped when next edited rather than all at once:
- Admiralty / MHEES “strong on all axes” → Established
- one soft axis (single strong source, or a one-step inference) → Supported
- credible but uncorroborated single stream → Reported
- weak or uncorroborated source, or a long interpretive leap, or an unjustified judgment axis → Thin
- contested corroboration, or a standing counter-reading → Disputed
The EJ-1 standard adopts the five grades as its public tier in place of E1–E5. The Trustless white paper drops the Admiralty reference and cites Receipts.