Compression theory reference?

bear bear at sonic.net
Wed Sep 1 14:12:27 EDT 2004



On Wed, 1 Sep 2004, Hadmut Danisch wrote:

>
>I have often heard that example and I used it myself several times.  I
>do not use it anymore, because it is not that easy. There's a major
>flaw in this example, and therefore it is not really a
>counterexample. If the faculty found that flaw I'd be in a bad
>position.
>
>You could define some reversible bizarre function f that does exactly
>that job, i.e. for a given message m you need to apply f again and
>again and after some finite number of calculations you'll find that
>f(...f(m)...) = x where x = 0 or 1

For example, loading the message into a Linear Feedback Shift
Register and iterating until it produces one of two predetermined
messages 0 or 1?

For a nontrivial message, the last star will burn out before you
get that many iterations.  Also, as you say, in order to decode
the message you have to know how many iterations you made - a
number which will, on the average, be the same length as the
message.

It hardly seems a worthwhile example.

>They say LZW and MTF. I have already given an example for LZW.
>They don't care. I've told them to take any random string taken
>from /dev/random under Linux. They don't care. The german principle is
>that a faculty is always right by definition.

That is inconsistent with the advancement of knowledge.  Any
university relying on such a principle has abandoned its duty.

				Bear

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com



More information about the cryptography mailing list