How Much Did Nostradamus Know? Information Theory and a Blind Test of Prophecy
Year: 2026
Type: article
Abstract: AuthorsKriger, Boris. Information Physics Institute, Gosport, Hampshire, UK; Institute of Integrative and Interdisciplinary Research, Toronto, Canada. ORCID 0009-0001-0034-2903 How much does a prophetic text say about the future, and how much is supplied by the reader who finds it fulfilled? This article answers the question for the 942 quatrains of Nostradamus's Prophéties (Lyon, 1555–1568) and gives a protocol that answers it for any oracular text. A prophecy is modeled as a fit relation between texts and events over a finite typology of event classes. A fulfilled quatrain that admits |F| of K classes teaches at most log2(K/|F|) bits about the course of events. For an undated prophecy, fulfilment becomes almost certain, its evidential weight decays to zero, and the interpreter must supply about log2 λ bits to name the fulfilling event, where λ is the expected number of fitting events in the reading window (anchor theorem). Vagueness has two independent sides: how many events fit one quatrain, H(E|Q), and how many quatrains fit one event, H(Q|E). Interpretive devices and corpus search inflate the fit without any foreknowledge. For every undated quatrain, exchanging past and future is a clean relabeling, so "prophecy" is a frame rather than a property of the text. Corpus measurements on all 942 quatrains: - the mean quatrain names 2.14 of 24 event classes and carries at most 3.15 bits; - well-known claimed fulfilments are each compatible with 30 to 111 quatrains; - seven quatrains state a year, and none comes true as written; - 84% of place mentions lie within 1000 km of Salon-de-Provence; - consecutive quatrains are lexically closer than chance (z = 8.0), so the order is not scrambled; - recomputed with a modern ephemeris, the astronomical configurations either recur every few decades or point to the author's immediate future (VIII.49: February 1559); - the Letter to Henri II contains two chronologies of the world that differ by about 580 years. Three double-blind experiments with independent interpreters and blind raters (300 rated items): - genuine quatrains, quatrains spliced from lines of different quatrains, and newly composed pseudo-quatrains fit history equally well (4.65, 4.61 and 4.60 on a 0–10 scale; p = 0.91); - genuine quatrains not recognized from the commentary tradition score below the forgeries; - real events attract better best matches than fictitious ones only when the interpreter searches the corpus; with random assignment the difference vanishes (0.60 vs 0.57); - advocacy cannot make a random pairing of quatrain and event convincing to a blind sceptic. The appearance of foreknowledge is produced by search over many weak fits, by the abundance of detail about real events, and by a tradition that remembers successes. The formal method of deconstruction for structured documents is applied to the corpus, its prefaces and its commentary tradition. The general protocol is extended to the Delphic oracle, horoscopes, the Yijing and scriptural prophecy. It measures predictive content only and leaves every other dimension of these texts untouched. The appendices contain the code, the experimental instructions and seeds, item-level tables, the complete record of every experimental item (texts, events, arguments, devices and blind scores), and the full 942-quatrain corpus used. Nostradamus; prophecy; information theory; falsifiability; deconstruction; double-blind experiment; oracular texts; Barnum effect; Delphic oracle; Yijing; horoscopes; computational humanities; ephemeris; Renaissance astrology; look-elsewhere effect; corpus linguistics Language: EnglishLicense: Creative Commons Attribution 4.0 InternationalVersion: 1.0Publication date: 2026-11-09 Related works Continues: 10.5281/zenodo.23166797 (A Formal Method for the Deconstruction of Structured Documents in Science, Law, Patents and Historical Sources) References: 10.5281/zenodo.23159974 (A Formal Method for the Deconstruction of Biographical Narrative) References: 10.13140/RG.2.2.26532.64648 (The Retrospective Drift Theorem) Communities / subjects: digital humanities, history of science and astrology, philosophy of science, statistics/methodology.
Author: Kriger, Boris
Language: English
Subfield: Religious studies
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7225640588
FWCI:
Cites:
Cited by:
Related to: