From JIM LESURF
I am fully aware of the points Ross Anderson makes (Letters, 14 March)
regarding ‘A spy’s guide to chaos’ (1 February) – as he could have discovered
by reading some of the things I have previously published. Unfortunately,
he seems to have misunderstood what I wrote. This may be because the first
part of the article gave a conventional introduction to the topic, followed
by a more specific mention of some of the work presently under way. This
entails using a non-linear analog system as a generator, not a ‘function’
which is ‘implemented in fixed-precision arithmetic’. There is, therefore,
no ’rounding error’, and the system is both simple and fast.
The main limitation of such a system comes from the ‘real’ random noise
which arises in all practical systems. This restricts the degree of deterministic
complexity we can synthesise in a given situation, but not in quite the
same way as the problems which arise in digital computation. Noise occurs
in digital computers, too, but the ’rounding errors’ are usually a lot worse.
Jim Lesurf University of St Andrews, Fife
