BODEN | 365
Our simple, or simplified, illustrations carry us but a little way, and only half prepare us for much
harder things . . . If the difficulties of description and representation could be overcome . . . we should
at last obtain an adequate and satisfying picture of the processes of deformation and the direc-
tions of growth.
The concepts provided by computer science, and the experimental platforms provided
by computer models, were a large part of what was needed to overcome those ‘difficulties of
description and representation’ in broadly mathematical terms. It is hardly surprising then that
Turing—the father of computer science, long fascinated by the biological questions that had so
intrigued D’Arcy Thompson—saw the opportunity and grasped it.
His 1952 ‘mathematical embryology’ paper is now recognized as an early essay in A-Life—
one that has inspired many later projects; in fact, it was his only published essay on A-Life,
although unpublished drafts were discovered after his death (as discussed later). But remarks
in some of his other writings presaged insights and techniques now thought of as characteristic
of A-Life.
Three such remarks appeared in the Mind paper that is usually remembered for introducing
the Turing test, but which was primarily intended as an outline manifesto for AI.^19 They con-
cerned the classification of computational systems, the possibility of evolutionary computing,
and the role of chemicals in the brain.
As for the first remark, he outlined the four-fold classification of complexity that is now
normally attributed to Stephen Wolfram.^20 A machine’s behaviour, he said, could be ‘completely
random’, or ‘completely disciplined’, or involve ‘pointless repetitive loops’; intelligent behav-
iour, he suggested, would require a ‘rather slight departure’ from the highly disciplined case.^21
Wolfram investigated such differences thoroughly. After experimenting with a wide range of
‘bottom-up’ computational systems called cellular automata (CA), he distinguished four types:
• those that eventually reach stasis
• those that settle into rigidly periodic behaviour
• those that remain forever chaotic
• those that achieve order that is stable without being rigid, so that the structural rela-
tions between consecutive states are varied yet intelligible.The final category, he said, included life.
Many A-Life researchers agree with Wolfram’s claim that a certain level of complexity,
involving both order and novelty, is needed for life in general, and for computation too: life is
possible, as they put it, only ‘at the edge of chaos’.^22 The Turing-inspired biologists, whom we
mention briefly later, accept this claim also, applying the mathematics of complexity to a host
of empirical examples.^23
As for the second remark, in his report to the Laboratory (NPL), where he designed the ACE
computer, Turing had already said that ‘intellectual activity consists mainly of various kinds of
search’, of which ‘genetical or evolutionary’ search—a technique widely used in A-Life today—
was one possibility.^24 When he wrote his Mind paper, the NPL report was still largely unknown.
But he hinted at the possibility of evolutionary computing, saying that evolution was analogous
to the use of ‘mutations’ and ‘random elements’ in a learning machine.^25
As for the third remark, he said that ‘in the nervous system chemical phenomena are at least
as important as electrical’,^26 but he didn’t elaborate. He didn’t hint that computers might one
day be able to model the action of neurotransmitters in the brain. Even more to the point, he