If you imagine the social web as a gigantic town meeting, you could presume that the number of inbound (trackback) and outbound links (blogroll) can define its size. Taking a short look at network theory, from its very beginnings to contemporary studies, will show how this interconnected city works. Networks can be found literally everywhere: in airports and airplane traffic, in our brain, in micro-organisms and even in social interactions. Network theory has evolved in a study proving that we are living in a connected age where ‘everything is connected to everything else’. Seeing the multitude of networks existing you can logically assume that there isn’t just one type of network. Depending on its structure and the behavior of its nodes, a network develops specific characteristics. So let’s introduce these networks, follow their expansion and consequently observe the characteristics of the blogosphere (and their meaning too!).

The type of network our town constitutes, can define the relationships and interactions in it. For example, if the members of the community are all randomly connected – each of them having on average one friend (link) – we result in a distribution where a large number of members (nodes) have almost the same number of friends (links).

Random networks

It is a theory developed by Paul Erdòs and Alfred Rènyi in 1959, called random graph theory. It relies on clustering, that is to say that most people’s friends are also to some extent friends of each other [Duncan J. Watts; Six Degrees]. It explains that the connection of small groups (clusters) will result into a single giant component. In a random network each node can contact any other node in very few steps and the more nodes the network has, the harder it is to find an isolated one.
If you want to translate random networks in attentiveness, the theory implies that drawing the attention of the public (namely the other nodes) is not such a complicated process. Everyone on the system receives more or less the same amount of links, namely interest. If you were to create a blog in a random network, sooner or later you would have roughly as many viewers as Google has. But we know (experience says) that the blogosphere - and the Internet in general - that we are not dealing with a random network. You do not randomly link to other blogs in your blogroll (or do you?), you choose them according to your interests and preferences. Therefore a new approach should be considered.
What random network theory did not take into account, was that most real networks (organisms, social relations, the web, etc.) are not static objects. New links are constantly created, while old ones abandoned. Apart from that, clustering takes place by virtue of individuals having a number of affiliations, which are not all equally alike, meaning that our new acquaintances depend sometimes from our old ones. Indeed our actions directly deprive from our characteristics and these actions may lead us to completely new connections, which have no relation to our previous ones.

 

 

[Both pictures taken from the book 'Linked' from Albert-Làszlo Barabàsi]

Scale Free Networks

In 1998 Albert-Làszlò Barabàsi with his co-workers were having their doubts, if the web was actually a random network (as thought until then). They started by creating a map of the internet (in reality a representational part of it), to research if its structure complied with the random model. They noticed that on the one hand 80% of the nodes researched had an average of four links while on the other hand 0.01% had more than 1000 links.
The architecture of the World Wide Web, Barabàsi’s group concluded, is dominated by few highly connected hubs (like Amazon.com, or Google). These hubs are visible to everyone – there is a link in most sites pointing to them. On the other side there are sites, which are on the borders of the map, receiving almost no attention at all. Such a network (named scale-free network) follows a power law distribution. It is a continuously decreasing curve, which shows that many nodes have a few incoming links, while a few hubs have a lot.

Attributes of scale free networks

A primary property of scale-free networks and the power law distribution is that the rich get richer. This happens because we usually prefer to link to webpages that are not ordinary nodes, but hubs. If you want to point to an online shop, the chances are you would point to Amazon. This means, that the more links a site attracts the more popular it gets and the more easier it is to find it.
But, this also implies that the oldest node on a network will always have the most incoming links, since it had the longest time to collect them. Still, we have seen examples of sites and blogs, which although not popular when first online, receive great amount of attention, sometimes over night. This happens due to the webpage’s ability to offer better content, updates, or services than its competitors. In such case of a competitive environment, each node has a certain fitness a degree, which defines how often the node will be linked. For example, between two nodes with the same number of links, the fitter one acquires links more quickly. While an old site might still draw attention of the users, a fit one will do so much faster. In other words, seniority is a great asset to draw interest, but fitness – which is independent of seniority - defines the speed at which nodes acquire links and therefore a senior node is not necessarily – or will eventually not be - the most popular.

[a good introduction to the subject was written by Michael Schuster; it helped me a lot // also a good article with nice recommendations was written by Jason Kottke]

What do scale-free networks tell us

• First of all the obvious. If you want to get heard in the blogosphere, you need lots of links - inbound and outbound. That the blogosphere is actually a competitive environment (whether we accept it or not) and we are all competing for links.That content can have a great deal of impact. And by content I don’t just mean text, but any type of services you could provide. May it be fancy CSS stylesheets, theoretical articles, beautiful pictures, groovy videos. If your fitness degree is high, you have nothing to worry about.