🚀 go-pugleaf

radford@cs.toronto.edu (Radford Neal) writes:

>This isn't what most people mean by "data compression". Suppose we are
>considering a compression method for high-definition television, for
>example. There is no requirement that these images be decodable by
>anyone with an IBM PC, even with no special software. Instead, we are
>willing to build special TV's with decoding devices, and accept that
>images will be decodable only by people in possession of such decoding
>devices. These devices might, in particular, include large amounts of ROM.

Sorry, you're wrong.
I know what you say sounds plausible,
but it's based on a misunderstanding.
We're not talking about the ease or otherwise of decoding.
We're talking about the informational content of the data.
If for example the data is random
then you will not be able to compress it at all,
regardless of how much ROM your recipient has.

(Incidentally, Chaitin/Kolmogorov entropy is only concerned
with compressibility itself --
it is not concerned at all with the length of time
the data takes to uncompress,
or indeed how much space this requires on the computer.)

-- 
Timothy Murphy  
e-mail: tim@maths.tcd.ie
tel: +353-1-2842366
s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland