From: ab411@detroit.freenet.org (David R. Conrad)
Date: Mon, 12 Feb 1996 21:05:31 +0800
To: cypherpunks@toad.com
Subject: Odd rounds in Blowfish
Message-ID: <199602121225.HAA09222@detroit.freenet.org>
MIME-Version: 1.0
Content-Type: text/plain




[ This is an article that I just posted to sci.crypt ]

In describing Blowfish Bruce Schneier mentioned that it achieves avalanche
after the 3rd round, and after each two subsequent rounds (5, 7, 9, ...).
(I understand 'avalanche' to mean that each bit of the key has affected
each bit of the data being encrypted.  Do I understand correctly?)

He also mentioned the possibility of decreasing the number of rounds in
Blowfish from the current 16, in order to improve the speed.  My question
is, would it be sensible to do it with an odd number of rounds?  If 10
rounds were a consideration, then why not go on to 11 to get the 5th
complete avalanche?  Or perhaps it would suffice to stop after 9 rounds,
after the 4th complete avalanche?

Is a whole number of avalanches a desirable feature?  Would an odd number
of rounds throw a spanner into the key schedule?

Also, how would this relate to the modified algorithm, Blowfish-SK?

I think I have now made my lack of expertise amply apparent.  :-)

[ ObNoise: does it take Nukes to end the jim bell/Alan Olsen flamewar? ]
[ ObRating: this post is rated G because it does not contain the words
  shit, piss, fuck, cunt, cocksucker, motherfucker, or tits. ]

Hoping these are not stupid questions,

--
David R. Conrad, conrad@detroit.freenet.org, http://www.grfn.org/~conrad
Hardware & Software Committee : finger -l conrad@grfn.org for public key
Key fingerprint =  33 12 BC 77 48 81 99 A5  D8 9C 43 16 3C 37 0B 50
No, his mind is not for rent to any god or government.