From: Adam Back <aba@dcs.ex.ac.uk>
Date: Sat, 9 Nov 1996 07:23:59 -0800 (PST)
To: ph@netcom.com
Subject: Re: Why is cryptoanarchy irreversible?
In-Reply-To: <v02140b04aea905f722f8@[192.0.2.1]>
Message-ID: <199611091130.LAA00192@server.test.net>
MIME-Version: 1.0
Content-Type: text/plain



Peter Hendrickson <ph@netcom.com> writes:
> > I hide the relatively small amount of data within a very large
> > amount of data which makes it impossible to find.  Data from analog
> > sources, like the "real world" (images, sounds, etc) is noisy.  This
> > is a fact of life.  Because this data is noisy I can hide
> > information in the noise.  As long as the information I am hiding
> > maintains the same statistical properties of noise it is impossible
> > to pull the information out of the data file unless you have the
> > key.  If I am paranoid enough I can make this key impossible to
> > discover without a breakthrough in factoring.
>
> Where will you keep your secret key?  Remember, when they go through
> your house they bring 20 young graduates from MIT who are just dying
> to show how clever they are and save the world at the same time.

Keep your secret key in your head.

> > This is the essence of steganography and the nature of signal and
> > noise are fundemental principles of information theory.
>
> The concept of noise is not all that well defined, however.  There
> is no way to look at a signal and say "this is all noise."
> Sometimes physical theories may lead you to believe that it is all
> noise.  That is fine for many applications, but when becomes less
> convinced of things if the consequences are severe.

Your plausible deniability has to get quite low before it will stand
up as "proof" in court.

Your real challenge is keeping your stego programs safe.  Boot
strapping a stegoed encrypted file system while leaving no stego code
lying around isn't that easy.

> >> If you are not doing it by hand, you own terrorist software and will pay
> >> the price.
> 
> > Ah yes, terrorist programs like cat and perl and operating systems like
> > Linux which contain a loopback filesystem that I can hook a perl
> > interpreter into at compile-time (which is enough for me to rewrite the
> > program from scratch each time if necessary, unless things like math
> > libraries are also outlawed on computers :)  I think that the crypto
> > concentration camps are going to be very crowded places.
> 
> Can you elaborate on this?  I am curious to know exactly what you are going
> to keep in your head and what goes on the disk.  Please post the Perl
> code that you would type in from scratch every time.

My specialty :-)

rc4 in C:

#define S,t=s[i],s[i]=s[j],s[j]=t /* rc4 key <file */
unsigned char s[256],i,j,t;main(c,v)char**v;{++v;while
(s[++i]=i);while(j+=s[i]+(*v)[i%strlen(*v)]S,++i);for(
j=0;c=~getchar();putchar(~c^s[t+=s[i]]))j+=s[++i]S;}

rc4 in perl:

#!/usr/local/bin/perl -0777-- -export-a-crypto-system-sig -RC4-3-lines-PERL
@k=unpack('C*',pack('H*',shift));for(@t=@s=0..255){$y=($k[$_%@k]+$s[$x=$_
]+$y)%256;&S}$x=$y=0;for(unpack('C*',<>)){$x++;$y=($s[$x%=256]+$y)%256;
&S;print pack(C,$_^=$s[($s[$x]+$s[$y])%256])}sub S{@s[$x,$y]=@s[$y,$x]}

The other problem I see is that if you have a stego file system in an
audio file, your disk writes are going look strange.  The inaccuracy
of disk head placement, is going to ensure that someone with the know
how will be able to copy off the last dozen pieces of data you wrote.

If they are all the same data with the exception of the LSB, it's
goint to look fishy.  Solid state storage devices are better.

Adam
--
print pack"C*",split/\D+/,`echo "16iII*o\U@{$/=$z;[(pop,pop,unpack"H*",<>
)]}\EsMsKsN0[lN*1lK[d2%Sa2/d0<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<J]dsJxp"|dc`