Yet another Grand Unified Theory...

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi everyone,

Here's my current Grand Unified Theory, and why I unfortunately can't test 
it - as normal, all comments more than welcome. :-)

The basic building block of the code is the metaphor of the *mirror", 
similar to the Kabbalistic "albam" cipher, where you write an alphabet in 
two halves, and each half maps to the other.

These "mirror codebooks" would have certain properties:-
(1) Symmetry: the encoder process = the decoder process
(2) Rearrangement: each codebook is not so much a mapping as a list of pairs
(3) Compact: would fit on a small area of paper
(4) Readable: the encoder can visually validate what's been written
(5) Ancient: the basic idea is probably 2000+ years old

However, if that were the whole code, then it would (of course) be simple 
to crack: but I believe that the VMS' code is second-generation, built 
using 8-10 separate mirror codebooks, each one corresponding to a different 
state - and that each time a gallows characters (and perhaps EVA ch) is 
emitted, it triggers a switch to a new state.

[Though it's no longer *globally readable*, it's still *locally readable* - 
and that would have appealed to the code designer.]

This scheme would require 8-10 mirror codebooks (each consisting of a set 
of pairs), which I earlier predicted would be somehow encoded in the 
"astrological" section in the nymphs' poses (and stars etc).

Close examination of those poses shows that each circle's nymphs are 
locally consistent, but are globally inconsistent. This (in combination 
with the reduced dataset required to encode a mirror codebook) points to 
what I now think the answer is:-

To determine the mirror codebook encoded for a sign:-
(1) catalogue the nymphs on a page according to their distinct attributes, ie
	left-facing / right-facing
	arm-up / arm-down
	arm-behind-hip / arm-in-front-of-hip
	legs apart / together / crossed
	star exists? (y/n)
	star tail exists? (y/n)
	star colour
	number of points on star - typically 7/8/9
	star - dot drawn in middle
	star - coloured blob painted in middle?
	pattern on barrel
	perhaps even hair style?
(2) carefully analyse this to form (mirror) pairs of nymphs (some may be 
left over)
	ie, "nymph #2 is-paired-with nymph#28"
(3) compare each pair's labels (in an aggregated alphabet, as in the AAH)
	the first characters where the labels differ [probably] forms the mirror pair

Then, I believe that these codebooks are [probably] placed on a 3x3 folding 
key [much as described by Steve Ekwall]. There are likely to be other 
special-case codes (number-encoding, for example - probably encoded as per 
Hebrew, ie 9x3 letters), but I think that the above may well form the 
backbone of the coding system.

Unfortunately, the reason I can't test this GUT is simply that all the 
astrological images are black-and-white - so any information encoded in the 
colour or colouring ain't there for me to see. :-((((

And until we get a decent set of colour scans - or I take a week off work 
to go to the Beinecke to exhaustively catalogue the use of colour in the 
VMS - that's the way it'll stay. :-(((

This also yields a prediction about the first-generation code that the VMS' 
code would have been based on: a simple one-state mirror code, where all 
the right-way-up astrological symbols went above the mirror line, and all 
the inverted astrological symbols went below the line. This may well have 
been the original intent behind the partially inverted character set.

Overall, then: arguments for the GUT:-
	Explains astrological pages structure and why 29/30 nymphs
	Explains number of astrological pages (why not 12 otherwise?)
	Doesn't require Trithemius-style cryptographic techniques
	Based on an ancient technique, so fits dates
	Symmetrical, so easy for encoder to validate visually
	Compact keyset, so not too unwieldy for encoder to handle
	Partially explains choice of character set - consistent with mirror metaphor

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