Army signals security & "Clansmen" series radios

[Brian Gladman]

From: "Owen Lewis" <oml@eloka.demon.co.uk>
To: "ukcrypto" <ukcrypto@maillist.ox.ac.uk>
Sent: Thursday, September 07, 2000 3:17 PM
Subject: Re: Army signals security & "Clansmen" series radios

[snip]
> Does one allow each individual user of such a system to generate their own
> PK pair, placing the public key on duplicated and mutually updating
servers
> at each communication node? Or should one better use cipher systems where
> the traffic is bulk encrypted and within which one-to-one superencryption
> can be employed as may be thought necessary?
>
> It seems to me that, in a system designed to function in a high risk
> environment, there would be a number of serious disadvantages to the use
of
> PK systems.

A big problem with the terminology of PK systems is that it encourages
people to assume that the 'public key' parts of keys have to be public.
This is most certainly not the case since they only have to be known by
those with whom the key owners need to communicate.  And in military
applications of this technology I don't expect the public keys to be public
(but it is nice to know that if they are lost it does not matter so much, a
feature that is not true of multiple use symmetric keys). And when the
public keys in PK systems are not public a number of their alleged
weaknesses go away.

Many of the problems attributed to PK technology are not a product of the
technology itself but rather a product of misinterpretation of the true
meaning of the terminology employed.

In consequence there are major problems with the term 'Public Key
Infrastructure' because it can be parsed as either 'Public <Key
Infrastructure>' or '<Public Key> Infrastructure' and this has led to
serious misunderstandings of the most appropriate ways in which the
technology should be used.  In particular it has led to the promotion by
governments (and certain parts of industry) of the need for some widespread,
open means of access to 'identity linked public keys' a development which
actually represents one of the least effective ways of deploying PK
technology in support of business applications.

In practice the most difficult issue in the effective widespread use of
cryptography lies in key management and key distribution and this is an area
where the cryptography and the business processes have to go hand in hand.
PK technologies offer major potential advantages here for both defence and
commercial applications but I don't expect there to be much read across
(except, possibly, at the technology level) because there is almost no
correspondence between military and commercial business processes involved.

[snip]
> Patently, it is possible to use PK systems for diplomatic/military
> communication functions. The question is, whether those systems have a
> balance of advantage that should promote such use or whether the balance
is
> one of disadvantage. If, as I believe, the balance may be strongly one of
> disadvantage, the hypothesised disadvantages perhaps provide a perspective
> to help evaluate the suitability of  PK systems in other environments.
> That's the game I was proposing, rather than to suggest that there is some
> fatal flaw in the PK per se. If weaknesses there be, it seems to me that
> these are for more likely to be found in either flawed or simply
> inappropriate use.

IMHO your last sentence is spot on.

Words are so important and the whole PK business is an interesting example
of how an early choice of terminology that is seen retrospectively to be
easily misinterpreted can have such a huge impact on the course of events.

   Brian