>  >From the IBM security site:
>
>  >In fact, most home-grown algorithms we have seen involve simply 
>XORing the data with some constant, and then performing some other 
>slight transformations to the text. This sort of encryption is 
>almost no better than no encryption at all.

I'm not sure that I agree with that statement. Surely, the inventor 
of PGP (Pretty Good Protection) didn't followed that dark-side logic. 
Also, PGP source code is available in C -- you can search for it via 
Sherlock.

One can come up with any number of "home-grown" algorithms that are 
difficult to break. It is how much time it takes to break the code 
that is the measure of the effectiveness of the algorithm. So, you 
have to design the scramble with who you want to keep your secret 
from. If it's from the Government, then it better be pretty good 
because they have the resources (time and money) to break just about 
everything -- I am not aware that they have broken PGP yet.

In any event, encryption can be pretty simple, but also fairly effective.

Now, what are you designing an encryption thing for?

tedd

-- 
http://sperling.com