Re: VMs: Pairing

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi Ken, Dennis, & everyone,

At 01:45 12/08/2003 -0500, Ken W wrote:

Quite so.  Also - what about "words" with odd numbers

of glyphs?

I'm no cryptologist and this may seem obvious, but what about the idea of some sort of "mixed" substitution used in some stage of encryption.  I'm imaging a process where some letters are represented by multiple glyphs and others by single glyphs.

You're imagining what I'm actually seeing - use my list of pairs and
singletons (from my previous email) to look at the VMS and see it for
yourself! :-)

Also: perhaps the key cryptological strength of this kind of mixed-pair
system is that - unless you can discern exactly which are pairs and which
are not - first-order statistical effects become transformed into a
mixture of (much weaker) first-order and second-order effects. I used a
"pure" pair cipher as a deliberately bad example.

Again, I don't think that this will explain the whole of the VMS' coding
system (I'm fairly sure there's more to it than just a pair cipher, I
think that's just the back-end coding stage), but it could well be a good
step in the right direction. :-)

IIRC (it's a couple of years since I calculated this out), the VMS'
(0-context) entropy appears to be something like 70% of languages circa
1500, so we might therefore predict an average symbol length of roughly
1.5 (fake) characters for every 1 (real) character. This is only a rule
of thumb, though. :-)

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

PS: looking back at Vigenere's Alphabeth Northmanique - I don't quite
know how I failed to notice it before, but it's simply counting through
the alphabet using capitalised Greek letters and Greek numbering. Alpha =
1 = A, Beta = 2 = B, etc. Some pair cipher *that* is! :-o



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