Well I for one am very pleased to see work on primitives such as Helix
and SOBER-128 which try to provide authentication as well as
encryption. It seems that construction flaws typically creep in where
not-so-crypto-savy programmers make up ways to authenticate, and with
AES we have this whole messy scenario where there are multiple
competing modes which provide authentication and confidentiality --
some of the more efficient of which are patented, and most of the less
efficient of which basically amount to no more than just that: MAC the
mesage and encrypt the message (great).
One question about using SOBER-128. Greg mentioned in his
announcement free use of the source code or other implementations, and
on the web page it says "freely available", but then there is also a
patent section at the bottom of:
http://www.qualcomm.com.au/publications.html
What is the overall picture: can anyone use the algorithms without
negotiating a license even though there are patents which may be
argued to cover aspects of SOBER-128's design?
Short answer: yes. There's a license in the source code modules that grants
a free license to use the source, or to use any of our patents for an
independent implementation to interoperate. If you want to use your own
implementation, and want to do it "clean room" (that is, not even *look* at
our code), there's a separate web page that you can print and file. If
anyone wants more guarantees than that (like a signable license), I'm sure
we could work something out.