Features
Part of Misquotes and fact checks: the questions that matter
Misquotes and fact checks analysis: a grading scale
Misquotes and fact checks analysis through a six-band scale for quotation wording, attribution, evidence, and what each grade allows you to publish.
Almost every quotation check ends somewhere between proved and disproved, and almost every published result pretends otherwise. Readers are handed a verdict, genuine or fake, when what the checker actually has is a position on a scale with a lot of room left on both sides.
For misquotes and fact checks analysis, a grade is more useful than a verdict, and it is also more honest. This page sets out a scale, says what each band requires, and says what each one licenses you to do.
What to take away
- Grade the evidence, not the sentence. The grade describes what you found.
- Two grades are needed, because wording and attribution fail separately.
- A grade must state what would move it. If nothing could, it is an opinion.
- The default for anything below the middle of the scale is that the line does not run under a name.
Why a verdict is the wrong output
A verdict compresses three different things into one word: how good the evidence is, how thoroughly you looked, and how much the answer matters. Compressing them loses exactly the information a reader needs to decide whether to rely on you.
It creates a false symmetry: confirming a quotation and refuting one are not equally achievable. Confirmation can be conclusive, because the passage is on the page in an identified edition.
Refutation almost never is, because evidence for invention is an absence, bounded by where you searched. Grading keeps that asymmetry visible. The long-standing vocabulary of provenance works the same way for physical objects, for the same reason.
Two axes, not one
Separate the wording from the name. They fail independently, and a single grade hides which one failed.
Wording. Is this string the string that was produced? A line can be genuinely someone's and still be circulating in a compressed, improved, or modernized form.
Attribution. Did this person produce it? A line can be accurately transcribed from a source that credits the wrong person.
A line that is textually exact and wrongly credited, and a line that is loosely paraphrased from a real passage by the right person, are opposite problems. One grade cannot say both.
The scale
| Grade | What it requires | What it licenses |
|---|---|---|
| A | The passage located in an identified work, with edition and locator, and the wording matched character for character | Quote it, cite it, rely on it |
| B | Located in a contemporaneous record of speech, or in a source with a specific citation you have opened | Quote it with the record named, note any variant wording |
| C | Earliest located appearance predates the fame, is dated, and credits this person, but no primary source exists | Report as attributed, with the earliest appearance and its date |
| D | Circulates widely, earliest located appearance is late, anonymous, or credits someone else | Do not use under this name. Report the earlier appearance |
| E | No appearance located before its current circulation, in named corpora, at stated search depth | Do not use. Report the search and its limits |
| U | Not checked, or checked too shallowly to grade | Say so. An ungraded line is an unchecked line |
Two things about this scale are deliberate.
Six grades: what each requires
Grade
- A
- located, edition, exact wording
- B
- contemporaneous record, citation opened
- C
- earliest dated appearance, no primary
- D
- late, anonymous, or miscredited
- E
- no appearance in named corpora
- U
- not checked or too shallow
Requires
- A
- quote, cite, rely
- B
- quote, name record
- C
- report as attributed
- D
- do not use under name
- E
- do not use, report search
- U
- say so, ungraded
Licenses
- A
- B
- C
- D
- E
- U
There is no grade for "fabricated". The most you can support is grade E, which describes your search rather than the sentence's origin. Writing "invented" claims knowledge of an absence across all print, and nobody has that.
Grade U is a real grade and should be used. Most lines on most pages are grade U, and marking them honestly is better than implying a check that nobody performed.
What each grade has to carry
A grade with no apparatus behind it is a decoration. Textual scholarship solved this problem long ago with the critical apparatus, the block of notes that lets a reader see the evidence for a reading rather than take the editor's word for it. Borrow the idea in miniature.
Apparatus behind every grade
- Wording tested, exactly as tested
- Earliest appearance, with date and place
- Corpora searched and their date ranges
- What you did not search, and why
- The one thing that would change the grade
A grade E that names the archive most likely to hold an earlier appearance is a research note. One that names nothing is a shrug with a letter on it.
The inventory of what different corpora actually contain is the material for writing that line properly. The research guide to searching historical newspapers is a useful model for how the people who built a collection state its coverage limits.
Applying it under time pressure
Grading does not require more time than a verdict. It requires stopping at a defensible point instead of an arbitrary one.
Time budget to grade
- Ten minutesdefensible D, E, or U
- One hourC on any printed line
- A or Bthe work in hand, no shortcut
- Set required grade by the use, not the line
Set the required grade by the use, not the line; a passing reference can run at C with a label.
Anything characterizing a person, especially a living one, needs A or B or it does not run. Our material on leadership quotations argues for that standard at length.
Anything going on a wall needs A, because the correction will never catch up with it.
Grading a line you cannot resolve
Most interesting lines finish at C, D, or E and stay there. Publish the grade with its apparatus and leave the question open. Our catalog of the specific mechanisms that damage quotations usually lets you say something more precise than "unresolved", such as which mechanism the evidence pattern fits. That is often the most useful thing you can offer.
Where the line turns out to be a common saying rather than anyone's sentence, the material on sayings with histories rather than authors covers how to describe it.
Common questions
Is a six-band scale not too fine for practical work?
The bands that do the real work are A, C, and E, and most people could run on those three. The others exist because the middle of the scale is where arguments happen, and having a name for each position shortens them.
Can a grade go up later?
Yes, and it should. Archives grow, and a line graded E five years ago may be gradeable C today. Recording the search you did is what makes a later upgrade cheap.
Should the grade be published, or kept internal?
Publish it. A visible grade tells readers what your page is claiming, and it protects you: a line marked E is not an assertion, and it is not read as one.
Does a grade replace the label on the line itself?
No. The grade describes your evidence and the label describes the claim on the page. A reader sees the label; an editor sees the grade behind it.
