Re: Is the VMs Undecipherable?

[reeds@xxxxxxxxxxxxxxxxxxxx (Jim Reeds)]

About unicity distance.  The usual examples apply to
classical ciphers, where there is a fixed key space, and
a more-or-less given plain text probability distribution.
And, of course, a definite algorithm for converting
plain into cipher, and back again, under the control of a 
key. 

To apply this to the VMS problem requies a stretch or 2,
as each of the features mentioned above are subject to question.
If the enciphering process involves a continual investment in
random choices on the part of the encipherer, one should
either abandon the idea of a fixed key space or should contemplate
adding an extra source of noise into the enciphering process.

Consider the following sort of "cipher": the encipherer is
entitled to randomly drop 1 vowel in 3 as she goes along, and
to randomly insert one extra consonant in 3; everything is then
subject to a simple substitution.  Do decipher involves deciding
which vowels to supply & which consonants to drop, as well as
in figuring out the unknown substitution alphabet.   A unicity
distance can be calculated for this situation, I suppose, but
the simplest formulas seen in hobby magazine articles -- like
mine -- will not work.

-- 
Jim Reeds, AT&T Labs - Research
Shannon Laboratory, Room C229, Building 103
180 Park Avenue, Florham Park, NJ 07932-0971, USA

reeds@xxxxxxxxxxxxxxxx, phone: +1 973 360 8414, fax: +1 973 360 8178