William H. Geiger III, <whgiii(_at_)invweb(_dot_)net>, writes:
Anyone unfamiliar with Misty should look at the CP archives for the
numerous posts by Nobuki Nakatuji. It has snake-oil written all over it.
This is exactly what is meant by 'pet' algorithm.
Misty is described in the proceedings of the most recent annual conference
on fast encryption algorithms. It is designed to be provably resistant
to linear and differential cryptanalysis. As a new set of algorithms
(a few variants exist under the "Misty" label), it is one of many where
a "wait and see" attitude is appropriate to see how it holds up. As a
patented algorithm, it may have trouble competing with alternatives that
are free of restrictions.
However your charge that it is "snake-oil" seems unfounded. It appears
to be a respectable academic development effort, within the mainstream of
cryptographic research, and has some reasonable-looking theory behind it.
As far as I know there has been no cryptanalysis or technical commentary
of any sort regarding Misty on the cypherpunks mailing list.