O.K. if the message is smaller,
you're going to have smaller
size documents.

So if a message is 1000 characters or
less, you're going to get essays instead
of complete works.

Tracy Johnson
MSI Schaevitz Sensors


-----Original Message-----
From: Roy Brown [mailto:[log in to unmask]]
Sent: Thursday, January 25, 2001 7:32 AM
To: [log in to unmask]
Subject: OT: Data Stopped, Encrypted, And Sent On Its Way


In message <[log in to unmask]>, "Johnson, Tracy"
<[log in to unmask]> writes
>As I was reminded last week, you could
>achieve any number of results for an
>encrypted item.  But you have no way of
>knowing whether results end up being
>the works of Shakespeare, Chaucer, or Tolkien,
>because without the key, you would end
>up with all results (actually an
>infinite number of results) from which
>to choose.

Iay on'tday inkthay osay.

Any encrypted message is likely to be of finite length. This places an
upper limit on the amount of information it can encode.

Further, the number of possible encryption keys is large but still
finite, and so, therefore, is the number of possible decryptions.

The decryption candidates can be considered by, say, a spell-checker and
grammar parser, to rapidly weed out those which are gibberish. This does
require a couple of assumptions (known language, single-encryption), but
these can also be dealt with in a larger problem space.

For instance, my encrypted message above, using a very old and simple
cypher, could perfectly well be an encoding of 'To be or not to be'.
But it couldn't possibly be an encoding of the whole of Hamlet, let
alone the whole of Shakespeare's works.

But I bet you can crack it. Is it 'To be or not to be'? And given the
theoretical difficulties you outline, how can you be so sure? :-)

--
Roy Brown        'Have nothing in your houses that you do not know to be
Kelmscott Ltd     useful, or believe to be beautiful'  William Morris