Re: Data encryption 360 degrees the nsa cannot break -- 01

From: Walter Roberson (roberson_at_ibd.nrc-cnrc.gc.ca)
Date: 12/26/03 

Date: 26 Dec 2003 20:34:01 GMT

In article <Xns945D7A3D34C6Djason.R.larue@216.77.188.18>,
Jason LaRue <aqdqmqiqnq@iqnteluser.no-ip.info> wrote:
:compression only works so far. for example, I would say that a
:_very_ good compression ratio would be 10000:1. You are suggesting
:a ratio of 59,029,581,035,870,565,171,200:1. This is not poosible.

An *average* lossless compression ratio of 10000:1 is not possible.
Not even an *average* lossless compression ratio of 2:1.
Examined over all possible input files, no compression scheme can
average better than 1:1.

But that's over all possible input files. If you have a more
restricted subset of input files that are "interesting" and want
to compress those at the expense of making other files come out
longer, then you can do that. A compression ratio of
59,029,581,035,870,565,171,200:1 is entirely possible -- but
only for 1 in 59,029,581,035,870,565,171,200 files.

Consider for example, this compression scheme:

  If the input is the number "1 googleplex", then output a single
  binary 0. If the input is anything else, write a single binary
  1 followed by the original number.

This compression scheme has a compression ratio of "1 googleplex:1"
for *one* 'interesting' input file, and grows all other files.
The method generalizes: if you want to nominate N different special
input files, you can compress -them- down to about log2(N) bits,
while growing everything else. [And if -every- input file is
"special" (i.e., a general compression scheme) then you see that
you end up with the same size of input as you started with,
so you can't save *on average* over all possible inputs.]

I don't understand what the original poster is talking about,
about folding and cancelling: I'm having trouble getting through
the English to understand the proposed algorithm. It looks like
a lossy compression scheme to me, not a lossless. From the part of
what I can make out, I don't think the original scheme has any chance
of succeeding -- indeed, my first reaction is that the posting was
some stegography (a message hidden in something posted in plain sight.)
So I am by no means saying that the scheme has a chance: I'm just
showing that your explanation of the failure is faulty. 

-- 
   IEA408I: GETMAIN cannot provide buffer for WATLIB.

Relevant Pages

  • Re: gzip basic
    ... Write output on standard output; ... If there are several input files, ... To obtain better compression, concatenate all ... Rich Teer, SCSA, SCNA, SCSECA, OpenSolaris CAB member ...
    (comp.unix.solaris)
  • Re: gzip basic
    ... Write output on standard output; keep original files ... If there are several input files, ... To obtain better compression, concatenate all ...
    (comp.unix.solaris)
  • Re: Gasoline grade BTUs per gallon?
    ... They have much higher compression ratios than gasoline ... built to use say, 87 octane gasoline, you will see no "preformance ... Higher compression ratio engines can make better use of it. ... If my engine requires higher octane fuel, ...
    (sci.energy)
  • Re: Why Cant A Fuel injected Petrol Engine be as Efficient as a Diesel?
    ... >compression ratios. ... This limits the compression ratio. ... diesel engines compress air only and add the fuel after that. ...
    (sci.energy)
  • Re: [fitsbits] Rice compression from the command line
    ... but beats gzip and bzip2 handsomely in compression ratio. ... Which raises the general question of constructing a figure of merit for data compression. ... Discussions like this usually focus on compression ratio, the speed to compress and the speed to decompress, but there are a number of important, less quantifiable, parameters: ... The FITS tile compression convention is one step toward greasing the rails. ...
    (sci.astro.fits)