VMs: Re: Is an "o" always an "o"?

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi Dana,

At 20:59 18/10/02 -0700, DANA SCOTT wrote:
> I am curious at to what the general consensus is concerning the Voynich 
> "alphabet"? Can we say that each "character" (symbol, glyph, etc.) always 
> represents the same "character" wherever it appears in the manuscript.

Here's how I think the character set is concealed.

In the only other place that "4" / "4o" appears on its own, it's used as a 
verbose mechanism to confuse decoders between glyphs, letters, and the 
underlying alphabet.

You see, in Tristano Sforza's 1450 cipher, "c" <--> "4", and "s" <--> "4o": 
so, if you then take a frequency count of the text (separating "4" and "o" 
into two apparently different letters), you'd find that "4" had a high 
frequency - but it actually corresponds to a lower frequency letter ("c").

You might then guess that "o" maps to "s", but that "4" was a null (which 
it isn't). Or that "4" maps to "s", and that "o" was a null (which is 
closer, but also wrong).

In the VMS' code/cipher, I believe that this too is the basic misdirection 
mechanism, only extended just a little further than the above. In the same 
way that its numbers are probably hidden (as steganographically concealed 
Roman numerals), so I believe its non-code-book letters are probably hidden 
as pairs of letters (whose partial alphabet overlaps).

This points to letter-pairs (like EVA <ar>, <as>, <or>, <os>, etc) quite 
possibly being individual letters (or perhaps pairs like "ch"). Certainly, 
this might comfortably explain many apparently odd features of the text.

I believe that this helped keep the VMS within one of its main design aims 
- which was that its apparent alphabet should have approximately 22-24 
characters. This would mean that the proto-cryptologists of the day would 
try to decode it as a simple monoalphabetic substitution cipher, when in 
fact it was much cleverer than that.

If you believe my dating (of 1450-1460), then you should also accept that 
few would have used a simple monoalphabetic substitution cipher then in the 
belief it was secure - everyone knew that these could be cracked without 
great difficulty, and the cipher "arms race" of the century (to build an 
uncrackable code) was in full swing. In the terminology of Thomas 
Homer-Dixon, privacy was a technological "ingenuity gap", crying out for 
ever-better mechanisms.

The genius of the VMS, then, is not that it *is* a simple cipher, but that 
it *disguises* itself as a simple cipher. :-)

So: to answer your question (finally)... I think that an "o" is always an 
"o", but that each probably represents only half a character in the 
underlying character set.

Cheers, .....Nick Pelling.....