2 steps to a small and connected world

There was an interesting article here in the FT yesterday about geographic context providing the next generation internet, and there is already a lot of serious cash pouring in that particular direction as described in the article.

However my question is, will that be enough? This query is triggered because I’m also currently reading finally Duncan Watt’s now quite old book 6 degrees of separation where he describes something that had been puzzling people interested in networks for  a while, which they describe as the small world paradox.

Let me see if I can describe this as I understand it. By taking one step and then another, can never get you further than taking 2 steps (see the picture of the triangle for a mathematical illustration of this same point). However in network theory this is exactly what can happens. 2 steps through a network (or 2 degrees of separation) can take you very far indeed. The best way to describe this is if you have 2 friends who you know through very different circumstances who you don’t think would have anything in common. So 2 people who are potentially very different (and therefor far apart in their own social networks), are actually very close – in fact just 2 steps removed because of their affliation with you!

The logical leap here is that people usually connect with people they are similar to, however they also a) use a couple of different criteria to determine how people are similar such as geography AND occupation and b) people tend to emphasise similarities over differences. And it’s these conditions that make the world small, and therefore also searchable. If you use just one criterion, i.e. just geography, the network is very large and unsearchable. It’s also worth noting that if you use too many categories to label nodes in a network it becomes overly fragmented and again unsearchable.

Which brings me back to the FT article – geographic search is great and important but not enough in itself, so the wise bucks will go towards creating content and networks that are categorised and searchable by several but not too many different contexts.