Can VM be written in vowelless Latin?

[MarcoP]

A general question is how would you evaluate a solution based on constructed language? What would be the technique there?

In my eyes, this scenario is not too different from evaluating a "natural language" solution. Bowern and Lindemann ("The Linguistics of the Voynich Manuscript", 2020) point out that historically documented constructed languages were based on the syntax of natural languages:

To our knowledge, the most extensive pre-modern conlang is the Turkish, Persian and Arabic-based language Balaibalan (Haeberl, 2015; Koc, 2005). Balaibalan was a Sufi esoteric language, represented in 3 manuscripts dating from roughly 1580 but likely a collective effort at composition over many years. There are two other well attested ancient conlangs, Latin and German-based Lingua Ignota, created by Hildegard von Bingen (Higley, 2007) in the 12th Century, and Enochian, an “angelic” English like language invented in the 16th Century by Edward Kelly and John Dee (Laycock, 2001). All three conlangs are heavily based on natural languages and consist of embedding made-up roots in the morphology and syntax of natural languages. Thus if Voynichese is a conlang, we might expect it to pattern morphosyntactically with other natural languages, but be anomalous at the root level.

So, I would check grammatical consistency according to the rules described by the solver. Of course, a solution which is not based on a grammar for the supposedly deciphered constructed language can be readily discarded. If Cheshire changed his strategy and claimed that Voynichese is not Proto-Italic but a constructed language from Ischia or something, his solution could still be dismissed for its lack of any hint of grammatical consistency (e.g. daiin=naus/food is arbitrarily translated as subject, object or genitive).

Though historically ulikely, Voynichese could also be "an artificial or universal language of the a priori type", as Friedman concluded. Also in this case, in order to be intelligible, the language must have a grammar: anyone claiming to have deciphered it must describe how the grammar works.