absolutely nothing to panic over
ndk.clanbo at gmail.com
Sun Oct 25 11:36:05 CET 2015
Il 25/10/2015 08:40, listo factor ha scritto:
> enough, we now see the cracks in the basement: advances in
> computing technology are corroding the fundamental algorithms,
> one by one...
Unless you move to another family of algorithms based on
information-theoretic limits on what an eavesdropper can know. Some
methods I remember involve neural networks in the form of tree parity
machines with a hidden layer (mutual learning is provably faster than
learn-by-watching), others use noisy channels (say readings from a
distant radio-source in deep space), others put a limit on the amount of
data an attacker could store...
All those have in common is that they require quite large data transfers
(so they're quite impractical) and the success probability of an attack
is mathematically limited (though quite "high" compared to current PK
and SK crypto, but can be made as small as you like by iterating enough
times). *No* advance in computing power can break 'em, unless it makes a
brute-force attack possible.
If the problem is "just" the birth of quantum computers, then there
already are some practical algorithms that address the issue (NTRU and
McEliece, as already pointed out by others).
More information about the Gnupg-users