> > > On Mon, Aug 21, 2000 at 09:21:59PM +0700, Oskar Sandberg wrote:
> > > > And we better hope somebody invents a Quantum proof Assymetric
algorithm
> > > > within the next four years - so that the inevitable patent has time
to
> > > > expire before our current version grow completely pointless.
> > >
> > > I am not an expert on this, but I have a suspicion that quantum
> > > computers can crack *any* assymetric algorithm, present or future.
> >
> > probably true - a general purpose quantum computer should be able to
solve
> > exponential problems in linear time.  since as far as I understand
modern
> > cryptography is based on the infeasibility of exponential computation,
> > we'll need something completely different.

> Probably the only way to make encryption which cannot be broken by
> quantum computers is to design an encryption algorithm in such a way
> that trying to brute force attack the data will end up creating a
> large set of incorrect files - the encryption would be designed so
> that incorrect data would look like correct data.  What is really
> being attacked here is not the brute force decryption by quantum
> computers itself, but rather the heuristic algorithms used to
> determine which decrypted data is the real decrypted data.  In these
> circumstances, you want the heuristic algorithm to produce as
> recognize as much incorrect data as being "correct" as possible.

Yes, the same basic idea as one time pad; give multiple seemingly correct
choices the equals probability. One way to do this would to have the
encryption algorithm build a key that the encrypted text can be decrypted to
a single "correct" plaintext but has the probability of being decrypted to a
multitute of choices. So instead keys for D(E) would result in nonsense they
would result in some human readable plaintext (if that was the original
plaintext). So if you decrypted your plans for assassinating the president
and someone decrypted it with a wrong key he would get shakespare, and yet
another key would give you martin luther king.

One way might be to have an artificial intelligence create a few bilion
other possible messages, and to compress all these messages in a known
order. The key would simply be an index to the huge list of messages.
Ofcourse the bit annoying problem with this would be that the encrypted
message might be a bit.. urm.. big.


--typo



_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to