Re: VMs: RE: King Tut Word Game, or the EKT Hypothesis

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi Graham, Dennis,

At 11:35 10/07/2003 +0100, Graham Waddingham wrote:
> Interesting idea, but no cigar I think. I can read messages written in
> this code almost in real time, and this was on my very first attempt.
> Also theres just too much predictability, y is always followed by e, b
> is followed by u 50% of the time etc. This *could* be a basis, but I'd
> say theres more encoding going on, or it would have been solved long
> before now.

I'm not quite so sure: the VMS has lots of frequent Tut-like predictable 
sub-sequences (qo, ol, or, ar, al, dy, chy, o + gallows & y + gallows, 
never mind dain/daiin/daiiin, etc). Sure, this is my whole pairification 
thing (so I would say that, wouldn't I) - but they're there whether you 
like it or not.

However, it's the contrast between this strong low-level structuring and 
the weak mid-level (ie, word-level) structure which generally fails to get 
explained by VMS theories. IMO, the EKT hypothesis is strong at the 
low-level but breaks down at the mid-level - languages are inherently 
redundant and broadly self-similar, which gives rise to many of the 
statistical relationships evident in normal texts.

Could there be a transposition cipher involved in the VMS? It's certainly 
possible - having spent some time looking at them, my inference is that 
transposition ciphers were trendy as personal cipher systems in the 
Quattrocento (eg, Leonardo da Vinci used one occasionally in his notebooks, 
which rather weakens the idea that he used mirror-writing for encryption, 
IMHO): this is because transpositions (like bbrvtns) typically make most 
sense to the brain-wiring of the person who made them.

However, (transposition + King Tut) would seem to be a long-winded code: my 
guess is that two of the mechanisms involved are (shorthand + King Tut)... 
well, OK, (shorthand + pairification). Remember, the author has already 
flagged to us that he's conversant with shorthand systems, by appropriating 
both (what appears to be) a contemporary pre-Characterie single-stroke 
shorthand alphabet, and a handful of Tironian notae.

Today's observation du jour: that the author probably realised from his own 
code-breaking experience that the difference between a crackable monoalpha 
and an uncrackable monoalpha is simply a matter of scale - the more letters 
you have, the easier it is to see a pattern. AFAIR, something like 30 
characters is enough to crack most monoalpha ciphers (if you have a 
reasonable idea of what the underlying language is)... and I'm sure that 
our author knew this too.

Therefore, having multiple local mono-alpha ciphers per page - though it's 
a bit of a drag for the person encoding it - is simply what it would take 
to get around this. However, I don't currently see the statistical evidence 
that suggests a full-on changing-codebook-every-glyph polyalpha... nor do I 
see evidence of a global mono-alpha (even pairified). So: where between 
these does the answer lie?

Steve Ekwall's 8-way / 9-way folding key is (for me) pretty much exactly in 
the right place within this continuum... but as he currently describes it, 
it doesn't ~quite~ work for me. I wouldn't be surprised if it turns out to 
be part of the story... but I'm quite sure it's only part of the answer, 
not all of it. We have more work to do yet... :-|

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


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