Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site utcsri.UUCP Path: utzoo!utcsri!carroll From: carroll@utcsri.UUCP (Eric Carroll) Newsgroups: net.crypt Subject: Re: DES certification Message-ID: <1270@utcsri.UUCP> Date: Thu, 18-Jul-85 19:54:23 EDT Article-I.D.: utcsri.1270 Posted: Thu Jul 18 19:54:23 1985 Date-Received: Thu, 18-Jul-85 20:42:24 EDT References: <9028@ucbvax.ARPA> <3154@cornell.UUCP> <405@petrus.UUCP> Reply-To: carroll@utcsri.UUCP (Eric Carroll) Organization: CSRI, University of Toronto Lines: 18 Keywords: DES, electronic surveillance Summary: Nasty Russians are listening. In article <405@petrus.UUCP> karn@petrus.UUCP (Phil R. Karn) writes: > 1. To formally adhere to DES (i.e., to gain a certification from NBS), you > have to implement the algorithm in hardware on a special-purpose chip. > General purpose computer software implementations are not certifiable. I've > never understood this requirement... > > Phil One reason for this is that it is currently possible to monitor the electromagnetic 'noise' the machine emits, and analyze that to discover some of what the machine is doing. Another reason is that software can be changed; if I have access to a system with a software implementation of DES, I could conceivably hack it to log all the attempts at encryption into a local file, or out to the phone line. Both compromise the whole idea of an encryption, namely that, in the ideal case, no-one but myself and the sender can read the message. The US DoD has very strict rules on the shielding requirments of machine rooms that house computers with access to top secret information.