Expert opinions about the VMS

[nickpelling]

Okay.  Well, on that subject then, what I question is whether there is really a valid concept in the "wider definition of cipher" that is not already included in some other category.  It seems to me that you have two basic options that are compatible with the properties of the VMS text:

1) Essentially unencrypted language of some kind, whether of the "code book"/artificial language variety, or the unknown natural language variety.

2) Extremely verbose cipher with some kind of scrambling/transposition applied, or complete gibberish with a message hidden using steganography.  In these two cases the plaintext would have to be much shorter than the VMS text, and the labels would have to be meaningless.

I don't think there's anything else that can be concretely described that can't be ruled out.

It has already been pointed out that 1) ("unencrypted language of some kind") is exactly what Tiltman and the Friedmans didn't think much of. To try to get around all the statistical objections, William Friedman ended up proposing that Voynichese would have to be an artificial language of some sort... though with the dating evidence we now have, this would be a couple of centuries before the first completely artificial language of any real sort was devised and discussed, which makes the possibility highly unlikely.

As far as 2) goes, if your starting position is that Voynichese is too predictable on a character level to be anything but some kind of verbose cipher (look at ol, or, al, ar, qo, aiin, aiir, etc), I'd agree that you're seeing it clearly. If your secondary step in your argument is to point out that Voynichese words are generally too short to be compatible with verbose cipher, I'd agree that this is also correct.

However, the explanation I proposed a decade ago is that Voynichese combines verbose cipher with scribal abbreviation - truncation, contraction, and so forth. By this scheme, a Voynichese 'word' would be made longer (through verbose cipher) as well as shorter (through abbreviation) at the same time, thus leaving Voynichese words about the same net length as their plaintext equivalents.

For example, EVA 'd' might well be 'contractio' (say, a token denoting that a syllable has been removed), and EVA 'y' 'truncatio' (say, a token denoting that the word has been shortened).

For what it's worth, I don't think that even this explanation gets us all the way to the finishing line, but I don't believe you've yet ruled it out.

Cheers, Nick