Harvard physicist's take on the VMS

[RadioFM]

A rambling, free-formed gibberich text could never maintain that kind of consistency and structure over 240 pages!

Glyph-to-glyph consistency you mean? What about word-to-word consistency? Yeah...

(...) the VMS has such a strange structure that it would have had to be a very complex “generator” if it weren’t based on an actual language. And that level of complexity would be completely anachronistic.

It would be easier to create a cipher that makes use of the principles already known at the time: spaces are not spaces, a homophonic system at the syllable level, different structures for different levels. And possibly also backward/rewerse writing.

You got it backwards: a cipher like that is anachronistic, meanwhile Gaskell & Bowern showed that certain individuals naturally gravitate to patterns seen in the VMS when penning gibberish. European cryptography well up to 15th century usually didn't cause horizontal/vertical reduplication patterns the likes of which we see in the VMS. You go ahead and try your best at a homophonic system, you'll soon end up in a spaghetti of rules, ambiguous parsing and perhaps even arbitrary anagramming before getting half as close as self-citation does. This is what everyone and their grandma have been trying the last 100 years, if not more.
The usual pipeline of meaningful solution goes: You make some homophonic system that kinda matches expected labels → You tackle a text page and get gibberish → You ammend the system with 99 variations to fit the text pages → You go to Rosettes and get gibberish yet again.