Re: Brute force question

From: Halldór Svansson (hallsva@hi.is)
Date: 03/11/03 

From: hallsva@hi.is (Halldór Svansson)
Date: 11 Mar 2003 07:36:16 -0800

"Michiel Buddingh'" <michiel@nospam.com> wrote in message news:<pan.2003.03.10.22.27.13.33835@nospam.com>...
>
> I wonder how effective an index of coincidence test would be in
> distinguishing `meaningful' binary data from garbage. Instinctively, I'd
> say it wouldn't be, as binary formats tend to have less redundancy than
> written text. Then again, my instinct is wrong about statistics 90% of the
> time.

True - such a test would would probably be rather ineffective for a
compressed file format such as JPEG, since compressed data is
generally pretty random-looking. Hence the need for a JPEG-encoder,
which of course slows down the process somewhat. Stick with text is
what I say - text is always reliably causal :)

Quantcast