VX Heaven

Library Collection Sources Engines Constructors Simulators Utilities Links Forum

Computer viruses (BMJ 307-59e)

Patrick Harkin
British Medical Journal, vol. 297, p.688
ISSN 0959-8138
September 1988

PDFDownload PDF (664.19Kb) (You need to be registered on forum)
[Back to index] [Comments]
\text{T_EX size}

Computer viruses have recently become a popular topic for discussion (23 July, p 246; 6 August, p 432; 13 August, p 488), but I have not yet seen a full description in a medical journal of the many forms ofdestructive computer software, other than viruses, which also exist.

Although they have received much attention and are potentially serious, computer viruses are rare. Much more common are "bugs" or unintended side effects. Many programs have subtle coding errors which may come to light only after many years of satisfactory use and others are not compatible with certain forms of hardware. I have erased my personal computer's hard disk twice through this form of incompatibility. Secondly, there are programs known as "Trojan horses" which claim to perform one task but which actually do something else (possibly as well as their stated function). These programs do not spread themselves from disk to disk but rely on a human vector. Finally, there are the true viruses. Viruses tend to lie low before becoming active, and Trojans may do the same. It is this feature which makes them dangerous and which may cause Minerva problems.

Minerva sets out a good set of directions to protect her computer from malicious code (p 432) but fails on one important point. It is virtually impossible to ensure that a disk is free of viruses or Trojans, and that uncertanty extends to the backup disks, so her plan of "sterilisation and reseeding" may only reinfect her system. The only sure way to avoid viruses is to avoid the use of resident code store-that is, suspect software should be run on a computer which has neither a hard disk nor much battery backed random access memory, and this computer should be switched off (not just reset) after finishing with the program.

I would also like to take issue with Dr John Croall (p 488): although it is theoretically possible for a virus or any other program to arise by random substitution, the chances of such an event are astronomically remote. In an effort to understand viruses and vaccinate my system I attempted to write my own (benign) virus. This required considerable effort. I estimate the shortest possible IBM personal computer virus to be 97 bytes long. If everyone on earth had an IBM personal computer and seeded all its memory with random values once every nanosecond this virus would be generated once every 10^{212} years. Even ifthis figure is out by a factor ofa billion billion I will still sleep easily in my bed tonight. The assertion that viruses might arise de novo is guaranteed only to provide ammunition for various anticomputer factions.

Patrick J R Harkin
Department of Pathology,
University of Leeds, Leeds
By accessing, viewing, downloading or otherwise using this content you agree to be bound by the Terms of Use! aka