VMs: Rugg Grills

[Koontz John E <John.Koontz@xxxxxxxxxxxx>]

I guess we should call them Rugg Grills, since, though inspired by Cardan
Grills they're quite a different thing.  Cardan Grills pick the
significant text out of a mass of irrelevant material, with the mass
intended ideally to appear speciously sensible of itself.  Rugg Grills
select characters out of tables for insertion in a text any part of which
is as relevant as the rest.  It's all gibberish or all encrypted text.

If you think about them, Rugg Grills are somewhat misleading.  The tables
have multiple triplets of columns side by side, but since you are supposed
to apply the grill within a particular triplet of columns you would get
the same affect if the column triplets were disconnected and laid out
randomly and you applied the grill to any of them at random.  Attaching
several sets of triplets side by side to make a table produces an ordering
of of the triplets that may or may not actually be used in Rugg's
experiments.

The reason the triplets are laid side by side in a table is probably that
this is the way that the source notion, the Cardan Grill worked - the
table derives from the obfuscatory host text.  In addition at one point
Rugg's idea was that the generation process was not a random walk of grill
across table, but a highly patterned one.

The triplets provide a mechanism for expressing something tantamount to
Stolfi's original tripartite word grammar for Voynichese - an analysis of
Voynichese words in the EVA character analysis into three parts that Rugg
calls prefix, midfix, and suffix.  Note that Stolfi finds this approach in
itself inadequate for generating Voynichese words and has a much more
complex approach which is still in adequate in itself.  In this sense
alone Rugg's work is in the nature of a preliminary experiment, using a
mechanism inadequate for generating the actual VMs.

In any event, a triplet of columns is a list of possible prefixes,
midfixes and suffixes.  A grill is a tool for selecting a specific triplet
of prefix, midfix and suffix out of one of these lists.

You can order the lists in a triplet as you like and you can cut the
windows in the grill where you want.  Nevertheless, all that happens when
you vary the locations of the holes in the grill is that you modify the
alignment of the columns in a triplet.  There are some edge effects,
depending on whether you restrain grill holes to locations within the
triplet columns, or treat a hole that falls above or below the end of a
column as, say, a blank, but, if you like, you can toss the grill, cut the
columns apart and slide them up and down next to each other instead and
just read your words across the resulting rows the columns.

You could also warp the columns into concentric circles, cut them apart
circularly and spin them relative to each other, as people have noticed.

So given any arbtrary set of tables and grills you can reduce them to an
equivalent list of tripartite words.  There may be repetitions in this
list.  In some sense they are the sometimes unanticipated results of the
choices you made in choosing your tables and grills, but in essence the
list and any repetitions in it are your deliberate choices and in
principle you could equivalently have written out the list by hand
instead, without resorting to tables and grids.

If you now want to produce a text using this list of words you can
randomly pick words from the list and write them down, one by one.  Words
that are more commonly repeated would appear more often.  This is
equivalent to randomly laying a random grill on a random table.

This approach is going to yield some repeated sequences, but not, I think,
in a way that will reproduce the patterning of words in the VMs.

Let's consider that we are only allowed to use a subset of grills and
tables.  This amounts to restricting ourselves to a sublist of the
possible words.  If we do this within any section of text, this should be
apparent to the extent that our subset differs from the whole list.

To get repeated sequences of words we can simply pick some sequences or we
can eliminate some of the randomness in our selection of words.  For
example, if you moved one grill in a repeated pattern around one table you
would be picking a sequence of words.  Or if you used your grills one
after another in order on the same table.  However, it's difficult to
pursue this sort of speculation without knowing more than I know about the
ordering of words in the VMs.

To a certain extent, it wouldn't matter to an analyst approaching our work
how we came up with sequences of words out of our word list, any more than
it would matter how we came up with our list of words.  The analyst would
care instead whether the words we included in the text and the order in
which they occurred actually reflected some such layered and random
process, or whether they reflected an encoding of non-random information.

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