Answers are grounded in this paper and cite their sources.
This paper gathers, in self-contained form, the mathematical background of tagged sets, a generalization of fuzzy sets, and of the (dis) similarity measures built on them: inward products, scalar products and norms of arbitrary order, Minkowskian metric vectors, Gram matrices and hypermatrices, and collective distances and similarity indices. As an application, the text of the Voynich manuscript is compared with the Universal Declaration of Human Rights in 62 languages and orthographies, with medieval Catalan, Castilian, and Latin texts, and with ciphers and text generators of known nature. At the character level, the manuscript lies more than eight standard deviations from the nearest reference text, although distance alone does not single it out, since a natural language in an unusual orthography can lie as far. The anomaly is located inside the word, in the information carried by its fourth and fifth characters, and a slot-grammar generator fitted to the manuscript reproduces most of it, far better than the same construction reproduces any language tested. At the word level, the vocabulary statistics of the manuscript are those of a language, and its adjacent words carry as little mutual information as those of the sparsest Latin technical prose measured. The analysis neither deciphers the manuscript nor proves it meaningless; it characterizes the text as more compatible with a word-building procedure than with a simply enciphered language.
No takes yet. Share an insight, caveat, or question.
Ramon Carbó‐Dorca (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: