Speculative fraud hypothesis

[magnesium]

As far as I can tell, for the self citation you only require a couple rules:
[...]
I think that is the conceit behind the algorithm, and it is certainly not a complex process.

It seems simple in the articles by Timm & Schinner but when you look at the code of the generator:

You are not allowed to view links. Register or Login to view. has the "rules" definitions: similarElements, combinableLigature, allowedFollowerGlyphs, allowedInitialFollowerGlyphs, allowedPrecursorGlyphs, finalGlyphReplacements, selfIngroupGlyphReplacements, selfFinalGlyphReplacements, etc.

The way the generator uses these rules is also quite complex: You are not allowed to view links. Register or Login to view. (1300 lines)

The result is far from convincing Voynichese, so more rules would be required to make the generated text indistinguishable from actual VM lines at a glance. The question remains: why would anyone choose to do it the hard way, by following a large set of (mostly useless) rules, instead of doing something much simpler that is more algorithmic and less rule-based? Mostly useless because natural languages are a lot less restrictive: there is no need for so many word-building rules if the aim is only to imitate natural languages.

I am not discounting the possibility, however, that the rules can be expressed in a more elegant and compact way. In a board game like Go the patterns are emergent, not hard-coded into the rules.

I think it would be interesting to sweep across a wider parameter space for this generator: randomly iterate the code’s various threshold parameters, sweep across a wide range of initializing lines, and re-run a given “seed” (initializing line + set of threshold parameters) multiple times to build up a broader representative population of generated texts. Are all these variations equally convincing, or are some more convincing than others? If there are, how does that relate to the algorithm or initializing line?