Re: VMs: A comment on Jacques Guy's table

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi Jeff,

At 21:58 17/12/2003 +0000, Jeff wrote:
> Notice the high peaks and number of zero entries. This cries out to indicate
> a rolling linked sequence.

Well... no, actually. What this means is that there's a lot of internal 
structure to Voynichese (but without indicating what that structure is). If 
you look again at the tables in Jacques' email, you'll see that different 
(EVA) letters have very different frequency curves.

What kind of language would have so much internal (letter-to-letter) 
structure? One strong candidate (clearly visible in many of the Tranchedino 
ciphers recently scanned and posted so kindly by Petr Kazil) is what we 
call "verbose ciphers" - ciphers which try to mislead code-breakers by 
emitting groups of letters (in a fake alphabet) to code for individual 
plaintext tokens.

Many of these go further still, and transform the plaintext into a more 
dense stream (using syllables, Tironian-style abbreviations, or 
sender-specific symbols) before encoding that. All this was standard 
practice circa 1450 - so why should we reject this kind of cipher and look 
for a more complex system?

Banging on an old drum, it should already be clear that "ch" / "sh" / "cfh" 
/ "ckh" / "cph" / "cth" function as single glyphs - perhaps it should also 
now be apparent to you that "ee" / "eee" / "ii" / "iii" / "iiii" function 
in much the same way. If you look beyond those, you might also see that 
"qo" / "dy" / "ol" / "or" (and arguably "eo" etc) appear together in ways 
that far exceed chance.

So, no: high peaks and number of zero entries (for EVA) cry out that the 
stream has some kind of highly predictable internal structure... but 
convert the kind of high-frequency EVA groups mentioned above into single 
tokens (which would be consistent with a simple verbose cipher) and you get 
quite a different class of distribution.

> Gordon Rugg seems to be the nearest to a solution and what is he using? A
> table lookup.

As just about any old-school programmer will tell you, you can encode a lot 
of "smarts" in a table - but that doesn't mean that two parallel 
table-based approaches necessarily have anything in common.

Gordon has diligently constructed one possible explanation for part of the 
VMs - but there are many confounding factors which his explanation fails to 
explain. As Philip Neal points out, beyond a certain point of complexity, 
such a technique would have been overkill. If Gordon's suggested mechanism 
is sufficient to generate plausible VMs-like text on its own, why on earth 
would the hoaxer need to make it yet more subtle and complex? Surely 
hoaxers are (almost by definition) lazy people, trying to get money for 
(next to) nothing - why bother to make the code closer to a work of art?

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


______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list