Scientists and mathematicians are more likely than the average to be atheists, or at least to choose evolution over creationism to explain what they see around them. This is for a very good reason. Excluding evolutionary biologists, for obvious reasons, scientists are used to seeing complex behaviour emerge from simple systems.

Take Conway’s Game of Life as a perfect example. With just four simple rules this cellular automaton throws up all sorts of complex, emergent behaviour, including several repeating patterns. If you regularly deal with such impressive output from simple rules, then it’s not much of a stretch to imagine monkeys turning into humans.

Simple rules minutely affect each and every iteration of a system, which over time creates huge, directed change. It’s a hard concept to grasp, but once you do, it’s easy to see its impact everywhere you look.

Computer scientists are more likely than most to come across this sort of behaviour, thanks to the field of artificial intelligence. In particular, those in one specific area of AI: the study of genetic algorithms. These algorithms solve incredibly difficult problems by mimicking natural evolution.

The way they work is incredibly simple; define a problem and the characteristics of an ideal solution, then set it running. That’s it.

The algorithm will take a set of randomly created solutions, compare them to the optimum, kill the poor performers and breed the better ones with each other. Throw in a little random mutation, and you’re all set. Within a few generations some staggering designs can emerge, sometimes so strange that no human engineer could ever have dreamed them up.

It sounds too good to be true, but it is. Demonstrably, reproducibly so. If at this point you’re sceptical, good, that’s a perfectly scientific attitude. Take a look at chapter 10 of this document, which runs through an extremely simple example.

It doesn’t just work for word games, though; so far it’s been used to design fusion reactors, create better load-balancing strategies in communications networks and NASA has even used it to develop more efficient antennae than was ever thought possible for a satellite. These things work, and are in use now, all around you.

Still, though, some people don’t believe it works. That’s understandable – if you’re an ardent believer in creationism, and a sceptic of evolution, then the field of genetic algorithms presents some interesting problems. You can throw the occasional spanner in the works of natural evolution, and the vast timescales involved make it easy to cast doubt on the theory for some, but seeing a computer evolve a perfect design in a matter of minutes is far harder to dismiss – even though it’s working in exactly the same way.

This brings me to the blog post that irritated me so much, written by Casey Luskin at Evolution News and Views. It criticises another article which details NASA’s fantastic work on the previously mentioned antennae.

“The presumption of evolutionary biologists, of course, is that these “brilliant designs” evolved by natural selection preserving random, but beneficial mutations. Engineers operating under such presumptions have thus tried to mimic not only the “brilliant designs,” but also the evolutionary processes that allegedly produced the designs,” says Luskin. “Did they use truly Darwinian “evolutionary computing?” The article goes on to discuss how design parameters were smuggled into the simulation, such that it really wasn’t a truly unguided Darwinian evolutionary scenario.”

Nothing was smuggled. The only things that the algorithm requires are details of the set of current solutions – analogous to a population of animals – and details of what an optimum solution will be like; low power use, highest efficiency, etc – which is analogous to the environment, weeding out poor solutions.

I had always hoped that genetic algorithms would help to convince evolution-sceptics to take a more thorough look at the evidence, but it seems that it’s just being added to the list of “incorrect” scientific theories. The problem is that the evidence is right there, routing your emails and phone calls, and whizzing above your head in orbit.

Indeed, the cutting edge argument falters even more when you learn that Edd is not even free to roam the massive Second Life landscape, interacting with other avatars in a truly free way. Instead, he is restricted to going where the boffins send him, and that’s that.

Which rather dampens the comments by Edd’s creators that this could be the forerunner of avatars which will be able to interact within 3D virtual world simulations of city streets or underground railway stations and be used to train emergency service workers. Get a grip on reality here, the truth is that this kind of avatar AI is in the earliest possible of moments in time. Once a synthetic avatar, if that is not too much of an oxymoron, can be let loose to roam at will and interact as just another resident of the virtual world. Then, and only then, will I start to get excited about all this. I cannot see Edd, or any other Second Life avatar for that matter, cracking the Turing Test any time soon…