Re: fwd: Re: RC5/6 patents

John Kelsey (kelsey@plnet.net)
Thu, 23 Jul 1998 00:23:13 -0500

• Next message: John Kelsey: "Re: DES & IBM's patents"
• Previous message: Adam Shostack: "Re: The Cost of Snakeoil"
• In reply to: Perry E. Metzger: "Re: The Cost of Snakeoil"

> From: Keith Lockstone <klockstone@cix.compulink.co.uk>
> To: CodherPlunks@toad.com
> Cc: klockstone@cix.compulink.co.uk; schneier@counterpane.com
> Subject: Re: fwd: Re: RC5/6 patents
> Date: Wednesday, July 22, 1998 9:00 PM

> It's ironical that a technique used in the key processing of RC6
> (adding a scaled version of the Golden Number into an accumulator
> to assist avalanche etc) first came to my notice in the TEA cipher.
> It's also used in LOKI97 - but they acknowledge the source, whereas
> Rivest does not.

Actually, RC6 uses RC5's key schedule.

TEA and RC5 were published at the same conference (FSE94, I think),
so it's not clear which of them used it first. Rivest's designs
(MD2,MD4,MD5) had previously used commonly-used mathematical
constants in various places, presumably to avoid charges that they
were installing trap doors. I don't know who did this first; it may
have been Rivest. I know that Merkle chose the Snefru, Khafre, and
Khufu (key scheduling only) S-boxes based on the RAND random number
tables.

Note that there doesn't look to be anything magical about the
constants used. They needed an arbitrary number with some reasonable
properties (reasonable Hamming weight and bit distribution, etc.) and
chose one.

> K.

--John Kelsey, kelsey@counterpane.com / kelsey@plnet.net
NEW PGP print = 5D91 6F57 2646 83F9 6D7F 9C87 886D 88AF

• Next message: John Kelsey: "Re: DES & IBM's patents"
• Previous message: Adam Shostack: "Re: The Cost of Snakeoil"
• In reply to: Perry E. Metzger: "Re: The Cost of Snakeoil"