Not so long ago, I read an interesting article in Weekend Financial Times on progress with the molecular computer. The gist of it was as
follows.
Until a couple of years ago, scientists thought that a molecular computer lay many decades into the future. That is, a "machine" that could store and manipulate chemical molecules rather than electronic impulses.
But a Leonard Adleman of the University of Southern California announced in November 1994 that he had built a primitive molecular computer and actually used it to solve a problem. he used DNA molecules. His paper to the US journal "Science" could turn out to be a most important scientific document with far-reaching consequences.
DNA computing will have two distinctive advantages:
Individual molecules will be able to hold millions of times more encoded information in a given space than the most compact electronic memory, and
A DNA computer offers the ultimate in parallel processing.
What this amounts to is that whereas an electronic system is considered to be superfast if it has 1000 chips processing data at the same time, a DNA system will have trillions of reactions simultaneously.
This has enormous implications for solving certain types of mathematical problems that even the latest very fast electronic computers would take impossibly long to solve, because the latter can only test one by one whereas the DNA computer has the simultaneous ability.
It is early days yet but Leonard Adleman has used his DNA computer to solve the "travelling salesman" problem of the shortest path between seven cities linked by fourteen one-way routes without retracing steps. This was all done by encoding strands of DNA to represent the cities and routes and mixing in a test tube, and pulling out the unique solution according to the techniques of molecular biology - a rather simplistic
sentence! A test tube could hold a billion billion strands of DNA, and although individual chemical reactions would typically take minutes rather than micro-seconds so many can be done at once that it makes the completion of the task many times faster than the electronic equivalent.
It could, therefore, be that we shall see development of hybrid electronic/molecular computers beginning to approach the workings of the human brain. We have already seen, this last year, a chip implanted and connected to the spinal nervous system of a young lady paralysed waist down in a car accident. Although this does not produce any feeling, she can make herself stand up and take steps by use of the little black box connected to the chip. The next fifty years are, I feel sure, going to see some remarkable developments, both in the speed of computing output and in the fields of application.
How about a chip in the brain to counter the loss of memory as one gets older? 21st century or later?
1st December 1996