Page 187 - Visions of the Future Chemistry and Life Science

168 P. J. KOLSTON

signals to be combined onto a single beam, taking up little space and not
interfering with cooling air. In, say, 20 years a fully optical computer would
integrate lasers with optical modulators and photodetectors, and could be
1000 times faster than today’s computers.

9.10 The year 2050
Within 50 years we may see the ultimate combination of biology and mod-
elling, with finite-element models being implemented on carbon-based
computing platforms. Carbon shares silicon’s electron valency, making it
a viable semiconductor. But carbon’s real potential lies in its unrivalled
ability to form compounds of very high molecular weight, which has made
it suitable for the encoding and processing of the huge amount of informa-
tion required to construct a human being. It is a logical step to consider uti-
lizing DNA code and associated enzymes, which have been developed and
refined over billions of years of evolution, to construct a carbon-based com-
puter. Such a device could exist in a test-tube, into which DNA-like mole-
cules would be placed containing the input data, and recombinant DNA
techniques used to perform the processing function. The output would be
the resulting new ‘genetic’ combinations. A carbon-based computer would
have several attractive characteristics:
• Fast: trillions of strands of DNA would be processed in a single bio-
chemical operation, so that a computation that would currently take
one year to perform could be completed in one second.
• Compact: if a grain of sand represented each unit of information on
DNA, the information contained in the sand on all the beaches on Earth
would fit under a fingernail.
• Efficient: a computation that would currently consume all the output
from a large nuclear power station could be performed using the output
2
of a single 1cm solar cell.
Taken together, these performance levels represent a million-fold improve-
ment over present-day computers. This means that the current rate of
exponential growth in computing power will be sustained for another half
century if carbon-based computers were to become commodity items by
2050. It may then be feasible to implement a finite-element model of a
complete human at a cellular level.