Just a hoax?

[magnesium]

[... important descriptions skipped ...]

The positional preferences together with the vertical effects and other line first word anomalies in particular show that the text cannot be a continuous narrative where words just wrap to the next line at every line break. This and all the other anomalies are going to be difficult for any natural language / shorthand / cypher hypothesist to explain.

I agree with these issues, and find them hard to explain - much harder than the low entropy issue.

If there is a real plain text behind this, some encoding method that left a degree of freedom to the encoder might go some way towards explaining it.

I completely agree with Rene. To add something here, it would be more likely to see the VMS with these properties if the underlying method of text generation made it more likely for these properties to arise. If the VMS is enciphered meaningful text, for instance, then the mechanism of the underlying cipher should be capable of reliably generating the patterns we see. (The Naibbe cipher currently fails to reliably generate long-range correlations and line-position effects, for instance, so cannot be the exact text generation method.) The same is true for any other text generation proposal.