VMs: Re: RuggWatch

[Jorge Stolfi <stolfi@xxxxxxxxxxxxx>]

  > [Rene:] I believe that it will be difficult to _prove_ that
  > Kelly had nothing to do with the VMs
  
Well, if we could decipher the text...  
  
  > [Rene:] I furthermore think that it will not be possible to show
  > that the VMs *could not* have been generated using a Cardan
  > grille, but I am convinced that it should be possible to
  > demonstrate that the method of a Cardan grille does not really
  > explain the complete riddle of how the VMs was created.
  
The problem is that the "Cardan grille theory" has too many degrees
of freedom.  

That freedom of course makes it easier to fit the theory to the VMS
(by opportune table and grille changes and complicated grille
motions). However it also leaves unexplained *why* would the author
would have taken all those decisions.  In particular, why would he 
chose those prefixes, midfixes, and suffixes (which BTW have lots
of structure on their own)?

We are to believe that he invented some clever and complicated tricks,
with the express intent of reproducing several very subtle features of
*unencrypted* natural language with meaningful contents (like Zipf's
law and the local variations in word frequencies); but at the same
time he allowed many obvious weirdnessess (like the doubled and
tripled words) that would leave even a naive buyer suspicious. It does
not make sense...

The Grille Theory actually makes a prediction that one could test.
Even if you change the grille after each pass through the table, you
should get parts of the same sequence of prefixes (combined with
different midfixes and suffixes) at every pass. EG you should see
patterns like

   .. olXXX qoYYY arZZZ choWWW ... olAAA qoBBB arCCC choDDD ...
   
Ditto for prefixes and suffixes. Has anyone checked this possibility?

There must be correlation-like tests that one could use to detect
such repetitions, even if masked by noise...

To avoid repeating prefix sequences, it would be necessary to move the
grille in a different sequence at each pass -- e.g. first vertically,
then horizontally, then skipping seven places at a time, etc. Would
the author be able to invent fifty different motion schemes, without
repeating the same sequence even once? Would he be smart enough to
move the grille at random?
  
  > [Dennis:] You could easily construct tables of common English
  > prefixes, roots, and suffixes, and then construct grills to
  > generate random English phrases and sentences. An extraterrestrial
  > who could not understand English might use these to generate
  > apparent English sentences, and conclude that English is gibberish
  > - but he would obviously be mistaken!

Exactly!

All the best,

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