Re: VMs: More on Cardan grilles

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

On Fri, 3 Dec 2004, Jacques Guy wrote:
> Conclusion? The VMs was not produced by that method.

True, but what Ruggs calls a Cardan Grill is used, not to position
meaningful text into an end document, but to select meaningless text from
a source document for insertion into an end document.  Needless to say,
something similar could also be used to select an encryption for some
input string.  For example, if the alignment of the grill were controlled
by indexes.  (But that approach would probably be too bulky to work.)

So what Rugg described was not a Cardan Grill, but a way of using
something physically similar to a Cardan Grill.  As you observe, it's very
hard to see how a proper Cardan Grill could be used effectively without
giving away the game.  It would certainly require a good deal of labor,
not to speak of ingenuity, if a meaningful context text was to be
supplied.

A proper CG would not really work with the layout of the VMs.  At a
minimum a lot of the text would have to be "hors de grille."

It seems to me that if the same Rugg Grill were used repeatedly in
generating a section of text it ought to be possible to deduce the
underlying source text and the grill.  Stolfi's grammar would represent a
good deal of the work involved.

I'm not sure that anyone but Rugg would be obliged to undertake the full
process, but I guess one might start by assuming that each word or subword
in a compound word is the output of a single RG application.  I think it's
fairly clear that some words are compounds, in any event.  A RG employed
in this way would amount to a way to generate random words and the
possible interactions of the windows in the RG with the source text (more
of a table) would (in?)advertently impose a morphosyntax on the words
generated.  Rugg's point, of course.  But only producing the RG(s) and
source text(s) needed to generate a part of the attested text can
demonstrate the validity of the hypothesis.  And, if an encryption scheme
were being used, producing these would most likely reveal it, or at least
provide the information needed to attack it.


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