Engineering your own voynich

[RenegadeHealer]

Trust me Marco, your model and explanation were anything but sloppy! And of course it remains to be seen if the requirements I made up are even a step in the right direction for reverse-engineering the VMs text. It may indeed be a ciphertext whose plaintext is not recoverable. I wonder if there are any ways this could be reliably distinguished from stochastically-generated strings of glyphs that never had any communicative or information-storing intent in the first place; I lean towards no.

I somehow managed to completely miss René's paper from this last spring, even though it's right up my alley. Thank you for that satisfying read. What René describes is indeed exactly what I'm thinking about, but more developed. That man is an absolute legend. To be honest I'm really not sure which one researcher had the biggest influence on my volvelle idea, but from now on I'll be citing Zandbergen, 2021 for it.

I especially like his easygoing and proactive approach to the hard problems of the Voynich text (Currier languages, LAAFU, frequent exact reduplication, frequent quasi-reduplication, infrequent multi-word strings, Grove words, small Levenshtein edit distance, one-legged gallows, m, labelese, and most of all binomial vord length distribution). For all of these except one-legged gallows, he expresses at least some confidence, that a model like his is tweakable in ways that could produce all of these effects.

I agree with Marco and René that frequent reduplication and one-legged gallows, respectively, are the hardest problems of the Voynich text. They stand out in how few satisfying explanations that involve a meaningful text they have generated. I've been doing some thinking about the one-legged gallows recently. I'm feeling hopeful about two potential explanations in particular for their strong preference for top lines:

1. [f] and [p] are ligatures, most likely of [ke] and [te] respectively. Either they proved too cumbersome to write without bumping into the line above, and/or their use is a deliberate indicator of the first line of a new paragraph. (You are not allowed to view links. Register or Login to view.)
   
2. [f] and [p] are used to encode a type of information that, due to deliberate stylistic and content formatting choices, is confined mostly to paragraph titles and labels. By way of analogy, consider the rare Greek letter sampi. It's almost wholly obsolete, except for its use in the native Greek numeral system, where it represents 900. Suppose a Greek legal text, which uses Greek numerals to index laws, were to be the sole surviving specimen of the Greek language centuries after the apocalypse. Decoders would be puzzled by the fact that the letter sampi only seems to occur in titles and top lines of paragraphs.
   

As for reduplication, that one still spooks me. What I've been trying to ask myself is, In what kind of recorded information, at what level of detail, is reduplication and near reduplication to be expected to this degree? I'm struck that if each vord encoded a single Roman letter of a text in a language that features doubled letters, one might expect to find the same vord twice in a row fairly often. But this possibility creates more problems than it solves, so I don't take it seriously, and proffer it here only as an example of what I mean.