Re: Request Key
Well here is a good rebuttal on the general point.
If you are referring to the "shared keys idea" that was put around, it was very well written. Competent mathematics. And terribly naive. Any component not in your direct control that can be used to expose the cleartext, is by definition a backdoor.
As it stands the state of mathematics requires some *large* measure of effort to break a properly encrypted message. If done *carefully* it is possible to say "No known method is efficient enough to decrypt a message before the heat death of the universe"*
An encryption algorithm is only effective if t(E(x)) << t(D(E(x))), where x is the cleartext, E(x) is the encrypt algorithm, D(y) is the decypt algorithm and t(z) is the time for the computational process z.
If any function B(E(x)) exists that has t(B(E(x)) <= t(D(E(x))), then E(x) is not considered an effective encryption algorithm, since D(y) is obviously obsolete, and cleartext x can be obtained more easily using B(E(x)). (B(y) is the backdoor function).
Currently, t(B(E(x))) >> t(D(E(x))) for algorithms such as RSA, they are considered secure*.
The problem is that the mathematically illiterate and politically expedient "wish" that t(B(E(x))) could exist and have the property ~= t(D(E(x))).
Which makes any encryption algorithm worthless.
Does that help?
P.
*Quantum computers not being real, yet...
PS. El Reg, MathML? Latex? Please?