Re: VMs: Re: possible solution

[Gordon Rugg <g.rugg@xxxxxxxxxxxxxx>]

In answer to your questions:

Stolfi proposes a three-syllable composition for Voynichese words (e.g.
qo+tee+dy). Other compositions are possible, but his is the simplest to
demonstrate. You can populate a large table with a series of columns, consisting
of prefix,midfix,suffix, prefix,midfix,suffix, etc, and then "pick and mix"
syllables to generate words. If you use stochastic methods or do it "by eye"
then the results aren't similar to Voynichese.

If, however, you structure the table to some degree (e.g. every fourth prefix
cell contains "qo") and use a modified Cardan grille, then you can end up with
something very similar to Voynichese. For instance, suppose the first few cells
of your table look like this:

qo tee dy
ol che y
y shek y
ol kee dy
qo tee y

The modified Cardan grille is simply a piece of card with three slots cut in it,
so that it shows one prefix, one midfix and one suffix at a time when you lay it
over the table. Suppose, for instance, that the grille has the slots in a
diagonal pattern from top left to bottom right. If you lay it over the table
above starting at the top, it will show "qo" plus "che" plus "y", giving you the
word "qochey". If you move it down a row, it gives you "olshekdy", and so on.

The interesting bit is that although "tee" is more common than "che" in this
table fragment, the pattern of slots in the grille will prevent "qo" and "tee"
from co-occurring in the same word derived from this fragment of table, whereas
"qo" and "che" can co-occur.

This is pretty similar to the restrictions you get in Voynichese. You can use
this method to generate pretty much any permutation of restricted combinations.
It can happen deliberately, or as an accidental by-product of how the table was
filled in. If for example you fill in every fourth prefix cell with a "qo", and
every eighth midfix cell with a "chek", then all the "chek"s will occur on an
adjacent cell to a "qo" and would never be selected together with a "qo" by a
grille with diagonally spaced slots. If instead you filled every sixth midfix
cell with a "chek", then half the "chek"s would occur next to a "qo" and half
would not, making "qochek" a less frequent combination than would be expected by
random chance.

You'd need to change your grille design after each pass across the table, and
there's a limit to how many permutations of slots you'd be able to use; after
that you'd need to generate a new table of syllables, and I think that's where
Voynich A and B come from. This method can produce pretty much all of the
statistical features of Voynichese in principle; I'm reasonably sure the LSC
scores could be duplicated given the right table structure, table composition,
etc, and I'm working on a software test of this with my student at the moment.

This method can be used for reversible encryption in at least two ways, but I
haven't found any encryption method based on this which (a) gives a signal:noise
ratio better than about 15% and (b) produces text with the same word
distributions as Voynichese.

Best wishes,

Gordon

PK#01 wrote:

> > You might be interested to know that it's possible to replicate most of
> the qualitative and quantitative features of Voynichese by using a modified
> Cardan grille in combination with text generation tables.
>
> Arrgh. I don't like that at all :-)
> A substantiated hoax theory would mean most of the fun is over.
>
>  > My article on this will be published in Cryptologia in January 2004.
>
> Will you post a copy of the article somewhere for us to read?
> I don't think it is a magazine that's easy to get.
>
> > The use of a modified Cardan grille suggests a date after 1550.
>
> Because it needs a too good understanding of crypto statistics (frequency
> counts et.) ?
> Is "Cardan grille" a well known term? It doesn't ring a bell with me. Where
> can I read more?
>
> > (if the LSC scores turn out to be replicable, which appears likely).
>
> Can you give me a pointer to that one. Where can we find the Stolfi article?
>
> ______________________________________________________________________
> To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
> unsubscribe vms-list

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