Guardian Unlimited | Life | Natural defences

Intellectual Diversity:

Alive and well in America... at least in the Department of Defense:

It all started last year when the US assistant secretary for defence and other senior officials within the Pentagon read In the Blink of an Eye, a book I wrote on the Cambrian explosion. It triggered a series of meetings in Washington and Britain, involving all manner of political and military figureheads, as well as defence analysts, computer programmers, tacticians and statisticians. Their hope was to see what a knowledge of evolution could do for national security. They emerged with a plan to create an extraordinary piece of software. Dubbed the 'Cambrian program', it will take perhaps the broadest overview of the world's social and defence systems, and use evolutionary theory to predict possible threats and outcomes. I and a team of experts at the [British Ministry of Defence]'s defence science and technology laboratory have already begun work on the program in Britain, and a similar consortium is planned at the Pentagon under Tony Tether of the Defence advanced research projects agency (Darpa).

If DARPA is involved, we can expect a brilliant but extremely complicated and expensive computer system to show up somewhere... ah, here it is:

At the heart will be a neural network, itself a piece of software, that must first be trained to handle the disparate information it will be fed. This is where the Cambrian explosion comes in particularly handy. The fossil record of the event documents how major advances, like eyes, and myriad minor changes in creatures, put pressure on other animals to evolve responses, be they new defences or different attack strategies. To train the Cambrian program's neural network, it will be fed data from the fossil and genetic record of life just before the explosion. It will then be given data from the very end of the explosion. As the program runs, the neural network will look at both sets of data and work out what connections lead from life before the great arms race to life afterwards. Once it has achieved this, the program can be fed hypothetical new data, for example the early emergence of an electrical sting as a weapon, to see what impact it might have on other creatures.

The notion seems to be that, by first plotting and then tracking how life evolved to deal with particular threats, we can predict probable routes for humans faced with similar threats. That can both be useful for defensive purposes (given a weapon using chemical X, what kinds of natural defenses can be employed?) and offensive ones (if we design a weapon using electrical bursts, what kinds of defenses can we expect to be raised against it that we'll need to overcome?).

Part of this sounds like a resurrection of the TIA program:

It would very likely be able to suggest obvious responses to the threat, but might also spot links between factors that humans might not have noticed and alert defence experts to weaknesses in their planned countermeasures.

The inquiry into the 9/11 terrorist attacks in the US showed that a breakdown in communication was largely to blame, there was no single chief who could make decisions using all of the data collected by various disparate security organisations. The Cambrian program, on the other hand, could manage all the data it is fed and provide constant updates on the size of the threat from different areas. That would allow officials to continuously modify their defence strategy.

Success depends, the author says, on information dominance: "[T]he system is dependent on receiving all the relevant data needed to reach a decision, and at the moment, we simply do not know what the data are. But we will get there."

That is a scientist's answer, confident about accomplishing the technical task, but dismissive of the ethical issues. Still, there are evolutionary pressures here too. The advent of processor capacity allowing this level of data mining means that someone is going to be doing this kind of thing. If it isn't the American and British governments, it will be someone else. Even a moment's reflection is enough to realize what will happen to the "little fish" if the next evolution is mastered by a predator state, instead of a sheepdog.