30-05-2017, 07:40 PM
(This post was last modified: 30-05-2017, 07:42 PM by Psillycyber.
Edit Reason: errors
)
It appears that Rene and Torsten are arguing past each other a bit here. I think Torsten is saying that the VMS cannot be directly representing a language (i.e. a "simple substitution cipher"). I don't think Rene disagrees with this.
The real point of contention is, could there be a more complex cipher system that would explain the strange co-occurrence patterns found by Torsten in the VMS? I lean towards "no" and think it is more likely at this point that the VMS text is truly auto-copied gibberish, but in theory I could imagine ciphering systems that would give rise to and/or allow the strong self-similarity autocopy-like properties of the text.
However, the only sort of ciphering system capable of this that I can think of would be some sort of verbose cipher.
For example, we know that there is a "curve/line" system in the VMS thanks to Brian Cham. Just to pick this oddity out as a far-fetched example, what if this curve/line system was cryptographically significant? To imagine a very anachronistic example (for the sake of simplicity), imagine that each curve-character stood for a dot, and each line-character stood for a dash, and VMS vords thus spelled out Morse-code letters, with gallows characters and other oddballs being nulls to throw off deciphering attempts. In this one weird case, one could easily imagine the following thought process from the point of view of the encipherer:
1. I need to encode 5 curves in a vord. I randomly write "chedy" which fulfills my need (I could have chosen many other possibilities such as "dyeeko" or "toodoo")
2. For my next vord, I need to encode 4 curves. I could either come up with an entirely new random vord (such as, "Feoyd") or I could be lazy and just tweak an existing vord. I look over and see "chedy," so I just write "okedy" next.
3. For the rest of the VMS, I continue to be extremely lazy and never explore the rest of the vast combinatorial possibility space of Voynich characters ("deedo," "tollokedo," "forger," "rookeed" etc. would all work just fine) and instead I just make new vords from existing vords, leading to "qokedy," qokeedy," and all the rest.
Or, consider a giant lookup table (GLUT) idea:
1. The encipherer starts with a random vord (that just happens to be something like chedy).
2. The encipherer, being extremely lazy, just lists a bunch of ways that chedy can be tweaked: okedy, qokedy, qokeedy, etc.
3. The encipherer numbers these tweaks starting from 1, 2, 3, 4....etc.
4. The encipherer then using a book like the Bible and matches vord no. 1 (chedy) with the first letter in the Bible, vord no. 2 (okedy) with the second letter in the Bible, and so on.
Now, my lack of imagination isn't conclusive proof that a non-verbose cipher is incompatible with the autocopying of the VMS. In principle, there could be some non-verbose cipher that is able to work even though the vords of the VMS were lazily autocopied. But I just find it unlikely that there would be enough degrees of freedom in the non-verbose cipher to allow it to still work despite the lazy autocopying. In any case, it's impossible to prove a negative. So I'm happy to tentatively conclude that I find it very unlikely that the VMS is a non-verbose cipher. And as for verbose cipher with intended meaning vs. gibberish, I give gibberish the benefit of the doubt at this point, although my mind is still open to possibilities of verbose ciphers that would be compatible with the autocopying and provenance of the VMS.
The real point of contention is, could there be a more complex cipher system that would explain the strange co-occurrence patterns found by Torsten in the VMS? I lean towards "no" and think it is more likely at this point that the VMS text is truly auto-copied gibberish, but in theory I could imagine ciphering systems that would give rise to and/or allow the strong self-similarity autocopy-like properties of the text.
However, the only sort of ciphering system capable of this that I can think of would be some sort of verbose cipher.
For example, we know that there is a "curve/line" system in the VMS thanks to Brian Cham. Just to pick this oddity out as a far-fetched example, what if this curve/line system was cryptographically significant? To imagine a very anachronistic example (for the sake of simplicity), imagine that each curve-character stood for a dot, and each line-character stood for a dash, and VMS vords thus spelled out Morse-code letters, with gallows characters and other oddballs being nulls to throw off deciphering attempts. In this one weird case, one could easily imagine the following thought process from the point of view of the encipherer:
1. I need to encode 5 curves in a vord. I randomly write "chedy" which fulfills my need (I could have chosen many other possibilities such as "dyeeko" or "toodoo")
2. For my next vord, I need to encode 4 curves. I could either come up with an entirely new random vord (such as, "Feoyd") or I could be lazy and just tweak an existing vord. I look over and see "chedy," so I just write "okedy" next.
3. For the rest of the VMS, I continue to be extremely lazy and never explore the rest of the vast combinatorial possibility space of Voynich characters ("deedo," "tollokedo," "forger," "rookeed" etc. would all work just fine) and instead I just make new vords from existing vords, leading to "qokedy," qokeedy," and all the rest.
Or, consider a giant lookup table (GLUT) idea:
1. The encipherer starts with a random vord (that just happens to be something like chedy).
2. The encipherer, being extremely lazy, just lists a bunch of ways that chedy can be tweaked: okedy, qokedy, qokeedy, etc.
3. The encipherer numbers these tweaks starting from 1, 2, 3, 4....etc.
4. The encipherer then using a book like the Bible and matches vord no. 1 (chedy) with the first letter in the Bible, vord no. 2 (okedy) with the second letter in the Bible, and so on.
Now, my lack of imagination isn't conclusive proof that a non-verbose cipher is incompatible with the autocopying of the VMS. In principle, there could be some non-verbose cipher that is able to work even though the vords of the VMS were lazily autocopied. But I just find it unlikely that there would be enough degrees of freedom in the non-verbose cipher to allow it to still work despite the lazy autocopying. In any case, it's impossible to prove a negative. So I'm happy to tentatively conclude that I find it very unlikely that the VMS is a non-verbose cipher. And as for verbose cipher with intended meaning vs. gibberish, I give gibberish the benefit of the doubt at this point, although my mind is still open to possibilities of verbose ciphers that would be compatible with the autocopying and provenance of the VMS.