Decoding the Voynich Manuscript: A 15th‑Century Generative Instruction Set

For six centuries, the Voynich Manuscript has been a tantalizing puzzle for cryptographers, linguists, and AI researchers alike. Every attempt to read its ink‑laden pages has ended in frustration, as the text refuses to yield to conventional decipherment techniques. The new research, presented in a paper hosted on the authors’ GitHub pages, takes a radical departure from the past: it treats the manuscript not as a language to be translated, but as a Generative Instruction Set (GIS) – a procedural code that can be executed.

“When I look at code, I don't ask ‘what does this say?’ I ask ‘what does this do?’” – a computer scientist involved in the project

From Glyphs to Commands

The key insight came when the team noticed that specific glyph clusters consistently appeared alongside particular topological features in the illustrations. For example, the prefix qo‑ always precedes text that describes the root system of a plant. In programming terms, qo‑ behaves like a scope declaration that switches context to the root hierarchy.

// Modern pseudo‑code
Scope root = new Scope();

// Voynich script
qo- [followed by morphological commands]

Similarly, the glyph p‑chor was mapped to a container core operation. When the manuscript instructs “p‑chor”, the artist drew a vase that houses the plant’s roots – a literal constructor rather than a descriptive phrase.

A Hierarchical Botanical DSL

The researchers mapped the entire botanical section of the manuscript, revealing an object‑oriented structure:

Element GIS Token Interpretation
Class definition cthy Declares a new botanical class
Core vs. appendage chor / dar Defines relationships between structures
Morphological shape tor, sh, ol Describes tuber, spiky, or round forms

This hierarchy functions like a Platonic database of idealized natural forms, generated by a 15th‑century codebase.

Interactive Exploration

The authors released a JavaScript parser that lets anyone type Voynich tokens and see the corresponding structures rendered in real time. The parser interprets the glyphs as drawing commands, effectively turning the manuscript into a live generative system.

“This isn't a translation. It's an execution.” – the research team

Users can experiment directly via the interactive terminal on the project’s website, which also hosts a full PDF of the paper.

Implications for Historical Computing

Recasting the Voynich Manuscript as a GIS challenges long‑standing assumptions about medieval knowledge production. It suggests that the manuscript’s creator possessed a sophisticated understanding of procedural logic, long before the formalization of programming languages. For AI researchers, the GIS offers a new dataset for symbolic reasoning and program synthesis, bridging historical artifacts with modern computational paradigms.

The work also raises intriguing questions about the provenance of the manuscript: if it is indeed a generative system, what tools and knowledge did its author have access to? And how might this perspective influence the study of other undeciphered texts?

The image above shows Folio 19r, the plant whose roots are encased in a vase – a textbook example of the GIS in action.

A New Lens on an Old Mystery

By shifting focus from translation to execution, the research opens a fresh window onto the Voynich Manuscript. It transforms a centuries‑old enigma into a living, programmable artifact, inviting developers, historians, and AI practitioners to explore the intersection of early computational thought and modern technology.

For those eager to dive deeper, the full paper and interactive tools are available at the project’s GitHub page: https://yaucheukfai.github.io/voynich-gis/