Research thesis: The Voynich problem may not be only that we cannot read the manuscript. We may no longer know what relationship its surviving representations were intended to have with the world, the practices of its makers, or the minds of its original receivers. If that bridge is missing, resemblance can become one of our most dangerous guides.
This article is the explanatory foundation for Tangential Voynich. The experimental branch deliberately places the manuscript inside systems it almost certainly does not historically belong to. This article asks the prior question: why would we ever need to do something so strange?
The Comfortable Mistake
The Voynich Manuscript invites attachment. A page appears botanical, and the mind reaches for plants. A circle contains stars, and the mind reaches for astronomy or astrology. Human figures stand in pools or tubes, and the mind reaches for bathing, medicine or anatomy. Rows of short strings resemble words, and the mind reaches for language.
None of those moves is absurd. Comparison is how knowledge works. The danger begins when the comparison becomes so familiar that we stop remembering it was a comparison.
The stronger the apparent resemblance to a familiar category, the greater the danger that the observer stops testing the category itself.
This is not a law saying the closest-looking interpretation must be false. A plant-like image may ultimately have something to do with a plant. A circular diagram may ultimately belong to an astronomical or calendrical system. The warning is subtler: familiarity lowers our cognitive resistance. It turns a hypothesis into background.
Once that happens, later observations are no longer neutral. A botanical frame makes ambiguous parts botanical. A pharmaceutical frame makes containers appear pharmaceutical. A linguistic frame makes blank-delimited strings become words. The frame begins manufacturing its own confirmations.
Reality Does Not Arrive Already Tokenised
eduKate’s Cognitive Art work begins from a simple problem: reality is continuous, while human reasoning operates by cutting, selecting, naming, comparing and compressing. The Token work makes this explicit. A token boundary is often not a boundary reality supplied for us. It is a boundary a receiver decided was useful.
That gives us a chain:
REALITY → OBSERVATION → EXTRACTION → TOKENISATION → REPRESENTATION → INTERPRETATION → ACTION
Every arrow can lose information. Every arrow can also introduce structure.
Observation selects where attention lands. Extraction selects which aspects matter. Tokenisation decides where one thing ends and another begins. Representation chooses a form that can be stored, transmitted or manipulated. Interpretation reconstructs a working model from that representation.
By the time a system reaches representation, it may look almost nothing like the reality from which it came.
A Map Does Not Look Like a City
An MRT map does not resemble Singapore in scale, geometry or physical texture. It deliberately destroys most geographic truth so that route truth becomes easier to see. The representation is useful because it is selectively false.
A musical score does not resemble sound. Algebra does not resemble motion. A chemical equation does not resemble a reacting flask. A circuit diagram does not resemble electrons moving through a conductor. A legal contract does not resemble the social behaviour it regulates. Computer source code does not resemble the pixels that eventually appear on a screen.
In all of these cases, a trained receiver knows how to expand the representation back into a useful world.
The representation alone is incomplete.
The Receiver Carries Part of the Information
Consider a subway map. The marks on the paper do not contain every fact needed to use the railway. The receiver already understands ideas such as station, interchange, direction, line, service, closure, entrance and destination. The representation and the receiver form a coupled system.
USABLE MEANING ≈ REPRESENTATION + RECEIVER MODEL + CONTEXT + PRACTICE
Remove the receiver model and the same marks become strange.
This is where Voynich becomes more difficult than a normal cipher problem. We do not merely lack a plaintext. We may lack the class of receiver that knew how the object was meant to be expanded.
If the original user was a physician, apothecary, astrologer, teacher, workshop specialist, court practitioner, memory technician, calculator, compiler of excerpts, multilingual reader or something outside our present categories, some of the missing information may have lived in the receiver rather than on the page.
The manuscript might therefore be complete enough for its historical user while remaining radically underdetermined for us.
Voynich as an Inverse Representation Problem
Let the original reality or practice be R. Let the historical representational process be f. What survives is approximately:
P = f(R)
We possess P and try to recover R.
The temptation is to imagine that there is a clean inverse:
R = f⁻¹(P)
But representational systems are often many-to-one. Several realities can produce similar representations. A representation can intentionally discard dimensions. It can depend on tacit knowledge. It can compress a procedure into a cue. It can preserve relationships rather than appearances.
Therefore the inverse may not be unique.
f(R₁) ≈ P f(R₂) ≈ P f(R₃) ≈ P
That changes the research objective. We should not ask only, “What interpretation fits?” We must ask, “What observations discriminate among multiple realities capable of producing a similar representation?”
Why the Images May Be More Dangerous Than the Script
The Voynich script announces its distance from us. We look at it and know that we do not know.
The images behave differently.
A root-like form feels like a root. A star feels celestial. A circular arrangement feels cosmological. A human body feels anatomical. Because recognition is rapid, uncertainty can disappear before reasoning even begins.
The least understandable parts of Voynich may contaminate us least. The most recognisable parts may require the strongest controls.
This does not make the illustrations useless. It makes them high-risk evidence.
The safer first pass is geometry before identity: branch events, lobes, symmetry, attachments, nodes, channels, rings, boundaries, repetition, negative space and topology. That logic is already active in the Voynich and Padua visual-grammar lane, where components are tested before objects are named.
The Script Has the Same Problem
The writing looks like writing. That observation is fair. But our modern transcription immediately adds another layer of representation.
The EVA transcription and segmentation work already captured the core issue: every decision about what constitutes a glyph, a space, a connected form or a word-like unit creates a representation, and that representation becomes the world the statistics can see.
MANUSCRIPT → MODERN OBSERVATION → MODERN GLYPH CLASS → MODERN TOKEN BOUNDARY → TRANSCRIPTION → STATISTICAL WORLD
If our segmentation is wrong, the mathematics can be perfectly calculated and still answer the wrong question.
This is why the recent Voynich unit work matters so much conceptually: the assumptions glyph = letter, blank-delimited unit = word and space = lexical boundary cannot simply be smuggled into the measurement. The unit has to be earned.
Money: The Cleanest Demonstration
Money makes the problem obvious because its visible form and systemic function are radically separated.
A banknote is coloured paper or polymer. A bank balance is a number in a ledger. Yet what money does may involve transferable claims, purchasing capability, settlement, accounting, trust, institutional guarantee, social coordination and deferred exchange.
Imagine a future archaeologist who recovers a banknote but not the institutions that made money work. The archaeologist can accurately study the portrait, ink, serial number, decorative geometry and signatures while misunderstanding the dominant function of the object.
The analysis would not be foolish. It would simply be trapped at the representation layer.
That is the Voynich warning in miniature.
Familiarity Is Not Evidence
Humans are good at normalising models. Once a representation proves useful often enough, it becomes cognitively invisible. We stop remembering how much interpretation is built into it.
We say “word” and forget that the boundary is a hypothesis. We say “plant” and forget that the identification is a hypothesis. We say “astronomical section” and forget that the section name was inherited from resemblance. We say “recipe paragraph” and begin counting recipe-like structures.
This is related to eduKate’s Normalcy Blindness work: a familiar phenomenon can become so cognitively backgrounded that the observer no longer interrogates it. Familiarity changes the search space.
The Representation Trap Has Three Levels
Level One: Appearance
What does it resemble?
Level Two: Representation
What distinctions does the system preserve, suppress or encode?
Level Three: Function
What did the representation allow a historical receiver to know, decide, remember, calculate, retrieve or do?
These levels can align. They do not have to.
APPEARANCE ≠ REPRESENTATION ≠ FUNCTION
A transport map is the canonical modern example: appearance is lines and coloured nodes; representation is a simplified network; function is route planning.
If Voynich contains similar separation, visual resemblance could be orthogonal to operational purpose.
Why Tangential Voynich Breaks the Attachment
If our nearby comparisons are saturated with inherited expectations, one way to expose those expectations is to move deliberately far away.
A railway control system is historically impossible as the origin of a fifteenth-century manuscript. That makes it cognitively safe. We can borrow routing, terminals, bottlenecks, state transitions and graph measures without accidentally deciding that the manuscript is a railway manual.
A compiler can force questions about tokenisation, operators, legal transitions and scope. Music can force questions about motif transformation and cadence. A supply chain can force questions about directionality. An ecosystem can force questions about co-occurrence, niches and regime shifts.
Each system is deliberately disposable.
The borrowed model enters, asks questions, makes predictions and then leaves.
The Purification Step
The core Tangential test is simple:
METAPHOR → PREDICTION → MEASUREMENT → REMOVE METAPHOR → KEEP OR DISCARD RESIDUE
If a composer says a line ending behaves like a cadence, remove the word cadence. Perhaps what remains is: a restricted family of terminal forms becomes more frequent near line endings.
That is testable.
If a railway engineer calls one form an interchange, remove railway vocabulary. Perhaps what remains is: the form connects two distributional neighbourhoods that otherwise have weak overlap.
That is testable.
If nothing remains after the metaphor is removed, the analogy has produced no transferable evidence.
The Counterintuitive Rule
The closer something appears to a familiar interpretation, the more carefully we should inspect whether the familiarity belongs to the manuscript or to the receiver.
This is not anti-history. It is a way to protect history from us.
Historical comparisons remain indispensable. Padua, Venice, Prague, Vienna, herbals, medical miscellanies, astronomical manuscripts and professional books constrain what was possible. But capability, resemblance, language, production geography and function must stay in separate evidence lanes until a justified bridge connects them.
Tangential Voynich creates distance so that when we return to historically plausible worlds, we can see which features were present before the story arrived.
A New Research Question
The old question is:
What does Voynich look like?
The new question is:
Which observations remain after we repeatedly change the representation through which we look at Voynich?
Those survivors matter because they are harder to explain as receiver artefacts.
They are candidates for invariants.
The Research Contract
- Resemblance may generate a question; it may not close the question.
- A token name is provisional until its function is validated.
- Visual identity is separated from visual geometry.
- Statistical structure is separated from semantic interpretation.
- Historical capability is separated from provenance.
- The borrowed Tangential system must make a prediction.
- The borrowed system must then be removed.
- Only observations that survive neutral restatement enter the evidence ledger.
- Held-out data and hostile controls outrank beautiful analogy.
- Failure is preserved when it narrows the search space.
World Return
The Voynich Manuscript may be unread because a language is lost. It may be unread because a cipher is unknown. It may also be unread because the conceptual bridge between representation and use has disappeared.
If that last possibility is real, looking harder at resemblance can make us more confident without making us closer.
The task is therefore not to stop comparing.
It is to become conscious of what comparison does to the observer.
Tangential Voynich exists to break the comfortable bridge.
We detach plant from plant-like. Word from word-like. Astronomy from circle. Bath from body-and-liquid. Meaning from statistical structure. Provenance from resemblance. Function from appearance.
Then we rotate the manuscript through distant systems and ask each one to expose a different structural property.
Finally, we remove those systems too.
The goal is not to discover which foreign world Voynich belongs to. The goal is to discover which parts of the world we keep seeing were placed there by us.
That is the Representation Trap.
And that is why Tangential Voynich must exist.
Continue: Voynich Research Library · Voynich | The Mathematics and Science · Tangential Voynich | What If We Deliberately Put the Manuscript in the Wrong System? · Information & Representation · Cognitive Art
TANGENTIAL VOYNICH · EXPLANATORY SERIES INDEX
Why We Break the Representation
Read these as one argument. First break the assumed relationship between appearance, token, representation and function. Then break and calibrate the observer. Finally rotate the comparison world itself, because even the baseline that tells us what counts as “unusual” is another representation choice.
Then Enter the False Worlds
System worlds: Systemic Parallax · Railway · Operating System · Compiler · Database · Filesystem · Telecommunications · Warehouse.
Object worlds: Courtly Luxury Object · Cabinet of Curiosities · Ceremonial Gift · Presentation Manuscript · Reliquary or Vessel · Automaton or Marvel · Nested Object · Conversation Piece or Heirloom · Made Simply to Sit There and Fascinate.
Break the safe representation. Rotate the observer. Rotate the null. Then let the wrong worlds attack what remains.