Re: VMs: RE: Word pairs

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Dear John,

May I suggest you briefly consider the statistics of pair ciphers in your 
study?

Quick background:-

If the VMS is written in, as many suspect, an encoded European language 
from circa 1500, then its stats (its low entropy etc) would seem to point 
to a typical glyph containing *less information* than a typical glyph in a 
comparable plaintext (say, Latin). Anyone who claims that the VMS is a 
monoalphabetic cipher then has to answer the [admittedly comp-sci-centric] 
question - where does this extra information come from? How do you propose 
bridging this entropy gap?

The traditional defence of the monoalphabetic cipher hypothesis is that of 
symbol "choice" - ie, allow symbols to code for several different 
renderings, and choose whichever one looks "nicest" (somehow). This is not 
very impressive, and - if you try it out - you very quickly start to 
project your expectations onto the text. Unsurprisingly, this hasn't proven 
satisfactory to date.

As an alternative, Shannon's classic experiments demonstrated the low 
underlying entropy (or rather, the high predictability) of English 
plaintexts: and several recent data compression techniques (based on 
"ranking" techniques) have begun to exploit this. However, while it's a 
*possibility* that the VMS' coding system uses some kind of ranking-based 
data compression to reduce the effective entropy of its code-stream, this 
is pretty unlikely (and I suspect it would give a distinctive curve anyway, 
which we don't see).

Alternatively, the solution to this conundrum might simply be that the 
average glyph length of a encoded character is longer than 1 - the paired 
cipher hypothesis. There are so many pairs that recur with extraordinary 
frequency, it shouldn't (if true) be a big surprise: and numerous ciphers 
circa 1450 appear to have used the mechanism.

So, before drawing a line under your study of the VMS, may I suggest it 
might be an idea to analyse a pairified transcription? You might get 
interesting results... I recently posted a bitrans script to the list, but 
please feel free to email me for more details.

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

PS: good luck with your final throes! :-)


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