I have been thinking that there is a possibility that a variant of "Alberti" style cipher was used that could produce Voynichese-like poperties. Or a similiar method was used to produce nonsense assuming VM is a hoax.
Overview of the method:
First all the spaces are ignored in the plaintext as is often a common practice.
1. A word length is chosen by random means
2. Length defines what encoding table to use
3. Each plaintext letter (and maybe some bigrams) are mapped to a VM character or a sequence of characters (with sometimes multiple alternatives) until the target length is satisfied. Some sequences may also contain a space so they can be used only on start or end of the word. For example " qo" and "dy ".
4. Goto to 1.
This might explain some VM properties
- Binomial distribution of word lengths
- Trailing characters at end of lines when plaintext runs out but length has to be satisfied because decoding depends on it.
- Repeating words
- Preferred location in word for many character sequences
In real life this encoding would be tedious to write and decode though. It would be also very verbose so plaintext would probably be only half of the cipher text.
Overview of the method:
First all the spaces are ignored in the plaintext as is often a common practice.
1. A word length is chosen by random means
2. Length defines what encoding table to use
3. Each plaintext letter (and maybe some bigrams) are mapped to a VM character or a sequence of characters (with sometimes multiple alternatives) until the target length is satisfied. Some sequences may also contain a space so they can be used only on start or end of the word. For example " qo" and "dy ".
4. Goto to 1.
This might explain some VM properties
- Binomial distribution of word lengths
- Trailing characters at end of lines when plaintext runs out but length has to be satisfied because decoding depends on it.
- Repeating words
- Preferred location in word for many character sequences
In real life this encoding would be tedious to write and decode though. It would be also very verbose so plaintext would probably be only half of the cipher text.