On Wed, 24 Dec 2003 21:38:28 -0800, "John A. Malley"
102667.2235@compuserve.com> wrote, in part:
Roger Wilco wrote:
[...]
Now see, this is why we frown on Internet access at company office
parties. Too much of that spiked eggnog and we end up with "encrypted
angels" and "two binary" codes and gads of spilt bits all over the place.
Thank goodness we don't get stuck with the meatspace aspects... :-)
The original post didn't show up on my server. Apparently the 'cipher'
being described is something like a Fourier transform. Like a Fourier
transform, being unkeyed, if the legitimate recipient can unscramble
it (a four-bit checksum of the original message, of course, is not
subject to unscrambling, and the post was ambiguous on that point)
then anyone else can too.
The trouble with XOR is that it doesn't have a good complementary
operation.
With addition, you can transform a and b to a+b and a-b, and you have
two independent items that both depend on both a and b. XOR doesn't
let you do that, at least not with single bits. (Using multiple bits
and Galois fields, however, I think you can manage.)
John Savard
http://home.ecn.ab.ca/~jsavard/index.html
