Re: computer virus

Stephen Danic (
Sat, 6 Nov 1999 14:24:43 -0800

From: "Stephen Danic" <>
To: <>
Subject: Re: computer virus
Date: Sat, 6 Nov 1999 14:24:43 -0800

> If a molecules is deformed, it will still probably bind to
> or act as a substrate to *something*. Hence, gene products are able to
> vary a lot and still function, while machine code tends to be much more
> restricted - the only code that can vary randomly and still work is code
> that doesn't get executed.

Even so, biological genetic mutations are usually fatal. So it is with
computer programs. Some GP researchers have suggested that mutations are
more likely to be succesful if the gene has been accidently duplicated
first. The product of the duplication (in biological systems) can then
mutate freely, because the correct gene is still being expressed.

I'm not sure how work is progressing in Computer Genetic Programming, but
some analogy should apply. As long as the original function is retained in
some form, the copy will be free to mutate without killing the system.

In a computer program, if a mutation occurs that causes a function to report
an incorrect value, the product may still be succesful. It shouldn't always
generate a system crash. I'm tempted to believe that computer programs are
More resilient to mutations than biological systems.

Virus API?

I'm surprised that computer virus programmers haven't introduced a way to
implement sexual recombination. Right now, virus authors see stuff they like
and reproduce it memetically, but the viruses themselves should be allowed
to do the same thing automatically. Simply borrow functions from other
viruses that they happen to come into contact with. The aspects of the
viruses could be encapsulated in genes, similar to java beans, and the
genotype would decide which beans to execute at which time. Then the genome
would be more able to change either through sexual recombination or
mutation. As long as functions, classes, objects are reoproduced and traded
in entirety instead of on the single instruction level, this should work.

Of course, that would generate very large, easily detectable virus files.
The way to get around this would be to create the programs as beneficial
worms so noone considers them to be viruses. Malicious code would only be
introduced later and at a very low frequency. Once people start utilizing
and depending on programs that are able to reproduce sexually, there's no
stopping the introduction of malicious code.

This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)