Stone by Stone: How Claims Are Checked

These are the rules every article on Persuasion Dissected follows. They are written out so a reader can hold any page to them and report the place where a page falls short.

Which source decides

The answer depends on the kind of claim. Age and prestige do not settle it.

• What was said or done on the record. Wording, procedure and original figures go by the primary source: the paper, the trial record, the text of the speech, the newspaper of the day.
• Whether an effect is real, and how large. The whole body of evidence decides. Meta-analyses and large replications are read together with the original study, and preregistered work counts for more. The original authors’ explanation is reported as their explanation at the time. Where results conflict, the article says the results are inconsistent and does not pick one.
• Historical fact. Dates, counts, sums and sequences start from the primary record, read with suspicion, since memoirs, indictments, pleas and propaganda all take a side. They are checked against an independent source or a scholarly history. A detail resting on one interested party is marked by X’s own account.
• Motive and cause. These are always labeled as interpretation, with the named scholars who hold them. A person’s stated motive is not treated as fact. An article on why a movement rose or failed sets out at least two named scholars or schools.
• Everything else. Encyclopedias are leads and do not get cited. Journalism and popular history are used only where nothing better could be read, and are marked as such. A practitioner’s own stories are labeled practitioner account and do not stand beside experiments as equal evidence.

When a source could not be read

A source is used only when its published terms allow this site’s use. If its terms rule that out, it is left out, even as a way of checking a fact. Where a publisher’s terms could not be found, the only material taken from it is material that is free to use on other grounds, such as a public domain text or a government document.

Where a record could not be consulted, the article says so and cites what could be read in its place. A paper whose full text could not be read appears as a bibliography entry only, or any figure from it is marked as reported in with the source that reports it. Source lists can state how much of a source was read, such as full text or excerpt, and the article’s claims have to match that level. A number or quotation taken from an old scan is compared with the page image before it is used.

How to read an article

The ask, verbatim. The line above the headline quotes one sentence with its coordinate, and the article takes apart what that sentence asks an audience to accept. If the persuader’s own words could not be read, the line quotes the nearest public domain primary sentence and says who said it, such as a commission or a court. Text still under copyright does not go there.

The case card. Case dissections carry a card with fixed fields: Case, Date range, Place, Persuader(s), Audience, The ask, Channel, Outcome and Primary record. The card and the matching CSV file are the same data. Study retraces and timeline comparisons have no card.

The crossing map. Each stone is one premise the audience accepted, with its source, and a text list beside the figure gives the same information. The map marks the last stone, or marks none when the case has none. It is an image for organizing the analysis. It does not explain why anyone believed anything, so an article will not say that an audience believed because it stepped across on its own.

The study stub. A research claim is followed by a short bracketed note in one of three forms. Any field that could not be confirmed reads unknown, and an experiment whose group sizes could not be confirmed does not have its results quoted.

• Experiment: [study stub: experiment; Exp. k, n=… vs n=…, N=…; preregistered: yes / no / not applicable / unknown]
• Meta-analysis or systematic review: [study stub: meta-analysis; k=… studies; total N=… / not reported; preregistered: …]
• Archival work, method critique, narrative review or historical analysis: [study stub: archival / review; N not applicable; preregistered: not applicable]

The counter-move. This passage puts the reader in the original audience’s seat and names the stone where a person could have stopped, along with information available at that time. It is a way of reading the case and not advice for a live dispute.

Numbers

Sums, gaps, shares and day counts are worked out again from the original figures. That tests the arithmetic and the presentation; whether the inputs are right still depends on the sources behind them.

Figures that belong to different parties stay with those parties. The 1848 Chartist petition is the standing example: 5,706,000 and 1,975,496 are two sides’ claims, printed together with the gap of 3,730,504. Plato’s thirty votes and the 281 in the Hicks translation of Diogenes Laertius are listed separately, and any inference is marked as one. A count worked out here is identified as this site’s, even where a source also prints the number. The 381 days of the Montgomery boycott are recomputed from 5 December 1955 to 20 December 1956. A National Archives teaching page also prints 381 days without naming the two dates, and the article says both.

Data tables hold single facts, each row with its own source and able to stand alone. A qualifier such as an approximate value gets its own column. These tables are assembled from original figures and do not copy any database’s table or text.

Headlines, short answers, descriptions, captions and table cells may not sound more certain than the body. A limit such as by his own report or about travels with the figure wherever the figure goes. Claims about what an unnamed crowd believed need a source, a sample and a period, or they are not made.

Quotations and translations

Classical texts are cited by their standard section numbers, and the translator is named where known. A translation is quoted at length only when its translator’s protection has ended in both the United States and the European Union. Where no such translation can be confirmed, the passage is translated for this site and marked as this site’s translation, with the original alongside.

Works still under protection get a short quotation with a page reference, or a bibliography entry alone. They do not appear in headlines, short answers, descriptions or cards.

Images

A work counts as public domain only when its status holds in the United States, in the European Union and in the country of first publication, and the credit states the basis. A public domain label that an institution places on a work is a lead and not proof. A government’s own work is judged on the legal basis stated for it. The accepted open licenses are CC0, the Public Domain Mark, CC BY and CC BY-SA; noncommercial and no-derivatives licenses are excluded.

Scans of documents keep their paper color and their edges in the article, and any crop or other change is stated in the credit. Diagrams are designed from sources and data and do not trace a protected work.

Anything labeled Illustration shows a kind of scene. It never shows a real person’s face, stands in for a real document or event, or carries a scale, a number or a step in an argument. Reuse terms for all of this are on Take These Maps.

Age limits

The threshold year is the current year minus 50, which makes it 1976 in 2026.

• Trials: the verdict and all appeals ended by the threshold year, and the defendant, the lead lawyers and the judge have died.
• Hoaxes: the exposure and any court proceedings ended by the threshold year, and the principals, accomplices, exposers and named victims have died.
• Negotiations: the talks ended by the threshold year, and the main negotiators have died.
• Movements: the movement has ended and its main phases ended by the threshold year. If an organization survives, only that earlier period is covered.

Anyone who cannot be confirmed dead counts as living. A living person is not an article’s subject and appears only through a public role on the record of that time, with no guess at motive. A living researcher is discussed only through published papers and methods.

What the method does not cover

This site does not teach persuasion technique, and it does not give legal, financial, medical or salary advice. Where a kind of fraud still exists today, the article says that anyone facing a suspected fraud should contact the local authorities, and it gives no steps for confronting anyone or gathering evidence. Further limits are on the about page.

Errors in any of this are recorded on the corrections page.