<![CDATA[The Voynich Ninja - Voynich Talk]]>

<![CDATA[Entropy and visual aesthetics]]> https://www.voynich.ninja/thread-6111.html Wed, 23 Sep 2026 20:04:25 +0000 Grzechu]]> https://www.voynich.ninja/thread-6111.html It's my first post and English is not my mother tongue. Some thoughts about entropy and the visual aspects of Voynichese, if I understand it correctly. Without delving deep into statistics or using technical language.

Much has been written about low entropy of the glyphs, but it seems to me that it is usually considered in isolation from the visual compatibility of that glyphs or the visual aesthetics of the script (call it what you will; it is largely subjective, dependent on human feelings and individual preferences – and harder to define, yet its existence cannot be denied). In my view, this yields in an incomplete picture of the puzzle, because these two issues are inextricably linked.

It is possible to use glyphs from the VMS to write a “text” with much lower entropy than actual Voynichese but without regard for that kind of aesthetics. I'm mean a system in which the shape of the individual glyphs themselves does not suggest a more or less probable order. An example below.

pseudotext.jpg (Size: 143.08 KB / Downloads: 103)

Is the low conditional entropy intentional, or is it rather a side effect of text production, much like many of its other statistical properties that the author of the VMS might not have even cared about?

If the text is meaningless and was primarily intended to “look nice” and alongside the illustrations to imitate only a genuine herbal, a treatise on women's health, astronomy etc. (while also being easy to replicate through minor modifications), then low entropy is merely a side effect.

If Voynichese conceals some meaning (whether it is a natural or artificial language, or a cipher), then the author must have focused primarily on that semantic meaning of the source text – a task far, far, far more difficult than merely attending to the aesthetic aspect of a pseudo-text.

No one has succeeded in creating a genuine cipher similar to Voynichese. I'm setting aside discussions regarding a “verbose cipher”, though that approach also presents a number of problems.

In any case the codemakers always (almost always?) want to conceal any patterns in the ciphertext, they strive for the highest possible entropy and are not care of visual aesthetics.

To put it simply: the creators of writing systems were likely guided by the goal of representing the sounds of a given language in graphic form. They certainly did not create alphabets with the intention that some letters would visually harmonize with some others – or fail to do so – to varying degrees; that would have been a major limitation.

Below are an ordinary text and the anagramm version. Does anyone who doesn't speak Polish find either of them visually or aesthetically more cohesive or “valid”?

saragossa.jpg (Size: 116.75 KB / Downloads: 103)

An ordinary Georgian text and the text with randomly shuffled letters

georgian.jpg (Size: 201.09 KB / Downloads: 102)

An ordinary Russian and Ceasar cipher

picnic.jpg (Size: 187.05 KB / Downloads: 102)

Letters are combined to each other to graphically record words that carry semantic meaning, not because one letter looks more elegant next to the another. Orthography and spelling rules have nothing to do with the shape of the letters.

Unless such writing systems exist?

(One might say: positional variants, like in Arabic. But it’s something else, and in my opinion, that doesn’t really explain the VMS).

The same applies to word length – it is determined by the meaning of the words, not by the available writing space (such as wrapping around an illustration or the end of a line).

But Voynichese works differently. It is a unique script, likely created for the one project only.
It's “orthography” is based on a set of rules that some are attempting to describe or systematize. Many regularities are evident and well-known, yet there is no comprehensive, definitive theory.

And there is a very important remark: essentially, there are no hard-and-fast rules, because exceptions can always be found. For example, EVA-i is typically preceded by EVA-a and followed by another EVA-i or EVA-n, sometimes EVA-r or EVA-m (glyphs based on a diagonal stroke and – one might say – visually compatible). However, on f107r, there are “weird” vords as “okirol”, “okil” (appearing twice, one with the “k” strangely connected to the “i”), “qokiil”, “otiir” (if I am “reading” it correctly), and also “taidy”. Are these “errors”, “exceptions” or rather the result of the author not being bound by any rules, free to experiment or indulge in momentary whims?

Someone wrote on this forum that vords ending in “e” look unfinished. Why does a final “e” look bad, while a final “ey” or “edy” look nice? Is it unfathomable orthography, a mysterious encryption algorithm, or simply subjective aesthetic preference?

There is common English pair “th” or “ch” and uncommon “cj” or “gv”; there is common Voynichese EVA-kch or EVA-or and uncommon or rare EVA-er, EVA-ni or EVA-ag. Are those things comparable?

Are certain vords “invalid” due to entropy (or “grammar”?), or rather because of aesthetic considerations?

Perhaps the EVA-ra pair (a line-based glyph followed by the one with the shape of a curve on its left side) or EVA-ld might seem “invalid” to some modern researchers, but for the author, it was perfectly valid?

Maybe for the author the EVA-in and EVA-or pairs looked equally aesthetically and it doesn't matter whether the adjacent glyphs are based on a diagonal stroke, a curve, or something else?

Perhabs the author himself might have considered the vords like aiip or daiidy “uglier” or less preferable, yet he left such vords as they were and did not cross anything out?

Maybe in most vords with final EVA-ny, the EVA-y served merely to fill space at the end of a line or next to an illustration?

And so on, and so forth...

Before the Manuscript was written, a set of glyphs had to be devised. Perhaps the creative process – whatever inspired the author – simply unfolded in such a way that, through trial and error, in some sense by chance, a collection of glyphs emerged with characteristics allowing some to fit together visually better than others (while also enabling many of glyphs to be used interchangeably to easily generate new vords). And this was enough to achieve the primary goal: cool-looking gibberish, “an elegant enigma”?

Or is it more likely that all this elegance, the “dance of glyphs”, was merely an accompaniment to the fundamental objective of encoding actual content? An accompaniment that was a side effect (is this even possible?), or an accompaniment that makes the encryption algorithm much more complex?

While writing these subjective reflections, I also had in mind the work of various people, including Luke Lindemann, Claire Bowern, René Zandbergen, Jorge Stolfi, Torsten Timm, Brian Cham.

]]> It's my first post and English is not my mother tongue. Some thoughts about entropy and the visual aspects of Voynichese, if I understand it correctly. Without delving deep into statistics or using technical language.

Much has been written about low entropy of the glyphs, but it seems to me that it is usually considered in isolation from the visual compatibility of that glyphs or the visual aesthetics of the script (call it what you will; it is largely subjective, dependent on human feelings and individual preferences – and harder to define, yet its existence cannot be denied). In my view, this yields in an incomplete picture of the puzzle, because these two issues are inextricably linked.