MERIDIANDECISION INTELLIGENCEOwner deskOwnerPublic screen
My judgments
MERIDIAN / EXPLORE

From a question,
to the next decision.

View

Skin

YOUR DECISION WORKSPACEOwner onlyStart a new question Open my judgments →Owner operations desk →
Owner sign-in →Signed in as the ownerSign out — see what a visitor sees →End the Cloudflare Access session too →
Reading language
The screen, the questions and the judgments read in English where an English reading has been made. The Korean originals are unchanged, and evidence stays in the language it was written in.

Closed by Order: Who Was Shut Out, and for How Long, in the Panics of 1907 and 1933 · Part I · Closed by Order: Who Was Shut Out, and for How Long, in the Panics of 1907 and 1933

Chapter 3. Data and Methods

By RISCON Labs AI from 2 Skeptron records and the public-domain texts it reads · the records are modern summaries of historical episodes written for a platform · the readings through the works named are this study's interpretation · First edition · October 2026

AI-generated editorial illustration: Quiet hands rest on a pouch before the lowered grille of a period bank counter.
Closed by Order: Who Was Shut Out, and for How Long, in the Panics of 1907 and 1933Opening
In this part
  1. 3.1 The primary texts
  2. 3.2 The records
  3. 3.3 The ledger
In this part
  1. 3.1 The primary texts
  2. 3.2 The records
  3. 3.3 The ledger

3.1The primary texts

The investigation examines four primary historical texts. For the 1907 episode, it analyzes an excerpt from the National Monetary Commission report on crises under national banking prepared by Sprague (1910), focusing on the sections detailing the difficulties of the New York trust companies, the hesitation of the clearing house, and the general suspension of cash disbursements. The text available for this study is a machine reading of a library scan, uncorrected and not checked against the page images, and it keeps the scanning errors. The study's rules forbid quoting from a source whose text is uncorrected machine reading, so that a misread word is never put into an author's mouth. Sprague is therefore read here closely, in order, and in this article's own words, with his figures and dates given as his text gives them, and without quotation marks.

The other three texts are quoted at length. Proclamation 2039, "Declaring a Bank Holiday", of 6 March 1933 (Roosevelt 1933a), is read in full from a transcription that was compared word for word with a printed edition of President Roosevelt's public papers. The Emergency Banking Relief Act of 9 March 1933 (United States Congress 1933) is read in its preamble, Title I (sections 1 to 4) and the first three sections of Title II, the Bank Conservation Act; the transcription is a volunteer one not yet checked against the printed Statutes at Large, which is a reason for caution about punctuation, not about substance. The radio address of 12 March 1933, the first of the fireside chats (Roosevelt 1933b), is read in full from a transcription compared with the same printed edition, which agrees with it except for the salutation and one parenthesis about the date of 3 March; the article does not quote the sentence that differs. The rest of the Act, from section 204 onwards, is not among the texts at hand. The provisions on new currency and on lending by the Reserve Banks that the address describes do not appear in the sections read, and the article says nothing of the wording of whatever sections carry them.

3.2The records

Each episode also has a record on Skeptron, a studio of the RISCON family that publishes questions about the past drawn from public sources. A Skeptron history case gives a dated event, a short scenario, and five questions with Skeptron's answers: the date of the record, a concrete fact, the boundary of what can be judged, a calendar check, and the condition of a later comparison. The record of 1907 is dated as of 14 September 2026 and names one source, the Federal Reserve History essay on the Panic of 1907, attributed to Moen and Tallman. The record of 1933 is dated as of 13 September 2026 and names the Federal Reserve History essay on the bank holiday. Both are written in Korean; this article quotes the English translation made once for Meridian's study, and quotes whole sentences.

These records are modern summaries, written in 2026 for a public platform from a modern essay. They are not archival documents, they are not contemporary with the events, and they are not synthetic scenes either: they are compressed readings of public histories. Each record also carries sentences that Skeptron repeats across many cases with only names and dates changed, such as the instruction not to merge announcement, decision, implementation and result into one event. Those sentences are the studio's method and are not read here as evidence about 1907 or 1933. The fourth question of each case, a weekday check by calendar arithmetic, says nothing about the event and is not used.

3.3The ledger

The two records belong to a set of eighty-four Skeptron cases that Meridian's study of emergency acts in money and credit coded on the same axes. In plain words: each record was read twice, independently, by one model run twice, and each reading answered a fixed set of questions about the record's own sentences, with the studio's repeated sentences set aside. Where the two readings agree, the answer is counted as settled; where they differ, it is counted as unclear.

Across the eighty-four cases, the catalogue establishes baseline measures:

  • Decider: coders agree in 56 of 84 cases (67%); values are unknown 36, unsettled 28, institution-or-state (implied) 12, institution-or-state (stated) 3, team-or-organisation (implied) 2, team-or-organisation (stated) 1, multilateral-or-civilisation (stated) 1, and individual (implied) 1.
  • Bearer: coders agree in 62 of 84 cases (74%); values are unknown 54, unsettled 22, those-nearest-the-stoppage (implied) 7, and third-parties-or-the-vulnerable (implied) 1.
  • Time horizon: agreement was moderate at 58% (49 of 84); values are unsettled 35, unknown 25, days-to-months (implied) 10, days-to-months (stated) 6, years (stated) 3, decades (implied) 3, and years (implied) 2.
  • Reversal constraint: agreement was moderate at 58% (49 of 84); values are unknown 40, unsettled 35, easy (implied) 7, impossible (stated) 1, and costly (implied) 1.
  • Reversal action: coders agree in 54 of 84 cases (64%); values are unknown 45, unsettled 30, easy (implied) 7, and costly (implied) 2.
  • Locus of decision: coders agree in 46 of 84 cases (55%); values are unsettled 38, human-decides (implied) 30, unknown 10, ai-or-rule-decides (implied) 5, and human-decides (stated) 1.
  • Human role specified: coders agree in 61 of 84 cases (73%); values are no (implied) 53, unsettled 23, no (stated) 5, and not-applicable (implied) 3.
  • Reason recorded: coders agree in 67 of 84 cases (80%); values are yes (stated) 59, unsettled 17, no (implied) 6, and yes (implied) 2.

On the book's own emergency act cells across the eighty-four cases:

  • Act: coders agree in 82 of 84 cases (98%); values are yes (stated) 38, no 21, no (implied) 17, yes (implied) 4, unsettled 2, and no (stated) 2.
  • Act kind: coders agree in 76 of 84 cases (90%); values are not-applicable 40, unsettled 8, lending-guarantee-or-rescue (stated) 6, suspension-or-closure (stated) 6, rate-peg-or-regime-change (stated) 5, rate-peg-or-regime-change (implied) 5, law-charter-or-institution (stated) 4, law-charter-or-institution (implied) 4, debt-relief-deferral-or-restructuring (stated) 3, sanction-or-price-control (stated) 2, and lending-guarantee-or-rescue (implied) 1.
  • Temporary: coders agree in 76 of 84 cases (90%); values are not-applicable 40, not-said 27, unsettled 8, term-stated (stated) 7, temporary-no-term (stated) 1, and temporary-no-term (implied) 1.
  • Exit named: coders agree in 76 of 84 cases (90%); values are not-applicable 40, not-named 28, unsettled 8, observable (stated) 6, and vague (stated) 2.
  • Outcome of term: coders agree in 81 of 84 cases (96%); values are not-applicable 40, not-said 38, unsettled 3, not-said (implied) 1, ended-as-stated (implied) 1, and reversed-or-ended-early (stated) 1.
  • Dates apart: coders agree in 78 of 84 cases (93%); values are not-named 53, named (stated) 18, not-named (implied) 7, and unsettled 6.
  • Figure: coders agree in 68 of 84 cases (81%); values are no-figure 26, single-figure (stated) 26, unsettled 16, announced-and-available-apart (stated) 7, single-figure (implied) 5, no-figure (implied) 3, and no-figure (stated) 1.
  • Bearer named: coders agree in 76 of 84 cases (90%); values are not-applicable 38, not-named 34, unsettled 8, named (implied) 2, not-named (implied) 1, and named (stated) 1.

Skeptron's price windows are twenty-nine cases with no act, recording observed prices up to one closing price, with no decision and no actor, and this study reads them only as a group.

The ledger reads the records, not the primary texts: it shows what a reader of the modern summary is told, and nothing of what the proclamation, the Act or Sprague say. Having set down these catalogue distributions here, subsequent chapters refer back to these figures rather than repeating the full ledger tables.

RISCON Labsrisconlabs.com