A lot of AI researchers pin their hopes on neural networks, computers that emulate the multi-tasking capabilities of the human mind where hundreds of separate processors combine to form a complex and powerful analytical system. At the same time, distributed computing causes what would be vast, glorious natural phenomena if the digital ecosphere could be visualised; the open source movement, Seti@home and Napster are diverse examples of its majestic power. A fundamental truth has been accepted, a truth which is known as Metcalfe's Law - that the power of a network increases exponentially with each individal node added to it. The implications of this truth for the field of AI research are that computers which are effectively networks will grow in speed and sophistication far faster than single-processor units. The chief example of this truth, on a global scale, is the rise of the internet.

You might enjoy considering the proposition that the internet is quite simply the largest neural net that will ever be built. Computer scientists would no doubt reject it angrily; there is a specific definition for the phrase "neural net" and the internet simply does not fit in. It's a network of individual machines. It cannot be instructed to compute specific tasks. It doesn't act in any way as a single, discrete entity. However, we can take the phrase "neural net" at face value, constructing our own definition, and apply it to this massive *net*work ordered in a fashion which is, in its complexity and scale, reminiscent of the *neural* patterns to be found in a human brain.

From this perspective the internet is an interesting prospect. Imagine a geologist spending years analysing small stones and pieces of earth from one patch of sloping, rocky ground, and all the time wondering how large the stones and pieces of earth might get; one day he stands up, leans back, and sees the cliff face tower above him. "Fuck", he thinks, reeling. "*That's* how big they get...". Obsessed with the particles, we're shocked to the core of our beings when one day we notice the structure. We've always been too close to it to really notice it's there, until one day we lean back and are struck breathless by the structure.

In the same way, I might make some propositions about networks and how we should perceive their scale. For example, we currently map networks in such a way that machines, or clusters of machines, are the nodes. However, each machine on a network will correspond to any number of elements in the sphere of human life. The machine might contain the web pages of thousands of individuals. It might spend all day continually serving up these pages while playing host to a non-stop stream of human e-mail communications, or it might be the point at which a number of distinct transactions converge. Fleshing out the stark roadmap of the network is the vibrant, smelly city of human memetic and economic activity.

If we were to try to map this smelly city, this "human network" in the way that we map computer networks, it would of course be far messier, far more complex and far less stable than its digital backbone. After all, the human network is organic. It's virtual, yes, and it's digital - information travels in digital form across the connections - but it's organic. Digital is the transitory state of its contents, while digital is the permanent state of computer networks. Information is transferred from the human network to the computer network via a process much like encryption; it is "compressed" into digital form by its originator and is then "decompressed" at the other end, where the receiver converts the digital package back into an organic, human format.

What might pass through the computer network as "742#89#528#574#957#824#759#847#987#582#097" is transformed into, say, a devastating message of rejection which then reverberates upon the human network close to the recipient. It could also be the cause for great joy on other nodes of the human network for whom it bears a much happier emotional package. The relationship between the computer network and the human network is one-to-many; one digital meme can mutate into any number of analogue memes once it hits the human network. The human network is inconceivably messier, more comples, more powerful and more organic than the computer network.

And still, it is the computer network from which we imagine intelligence might come! I don't wish to criticise the computer network, of course, but I think that it's wrong to consider it in isolation. The human network has been the sole cause of its growth, and without the human network the computer network would eventually grind to a halt. The human network had for millenia yearned to strengthen the connections between its nodes, and when the computer network started to make this possible it was seized upon, and expanded massively as the human network flooded onto it.

So bearing in mind that the human network exists around the computer network, we should also consider that perhaps it's the "human network", the network of meanings behind the data - the meta-network - that will display sentience. Perhaps it already does; each of its nodes is sentient, so why shouldn't it be so itself? It won't be able to exist without the computer network, of course, and this would lead some to believe that sentience is a product of "the internet". However, they'd be wrong.

Any type of "artificial" sentience would, I think, be more likely to come from the invisible network of human thought and activity that overlays the simple geometric skeletality of the computer network. And it would be extremely, extremely strange. It would be undetectable. It might therefore already exist. It might have existed long before the computer age, before the Industrial Revolution. It might never exist. Its existence, in fact, might be a moot point for us in the first place; our inability to comprehend sentience in the global networks might cause a Schrodinger-esque quantum quandary where, simultaneously, it both does and does not exist.

Will we ever be aware of a non-human sentience emerging from our increasingly collective network of experience? It's a funny question! And one of the worst kind - the kind that will only ever answer itself, and will take its time in doing so.