Monday, August 20, 2007

The Blogosphere is a Mesh

I said
You say "Wrong both descriptively - it's not what the blogosphere actually looks like... What we are more like ... is a mesh, and not a hub-and-spokes network."
and
I'd be very interested to know the evidence for that statement.
My evidence is that this is what I see, and that if you looked at it from the same perspective, you would see it too.

Yes, you could measure it 'empirically' via a formal study, but (as I have commented on numerous occasions) you tend to find whatever you're looking for with such studies.

For example, you could do a Technorati sort of survey and list all of the blogs that link to each other. From this, you could construct a social network graph. And that graph would show what the link cited in this thread shows, that there is a power-law distribution and therefore a hub-and-spoke structure.

And thus you would have found what you were looking for.

And yet, from my perspective - as a hub - I see remarkably little traffic flowing through me. How can this be?

The edublogosphere - and the wider blogosphere - isn't constructed out of links. The link is merely one metric - a metric that is both easy to count and particularly susceptible to power-law structuring. Links play a role in discovery, but a much smaller role in communication.

We can identify one non-link phenomenon immediately, by looking at almost any blog. After any given post, you'll see a set of comments. Look at this post of Will Richardson's. There's a set of 25 comments following. And the important thing here is that these comments are communications happening in a social space. They are one-to-many communications. This forms a little cluster of people coimmunicating directly with each other.

Now look at any social network, say del.icio.us. This tool was ranked second on a list composed mostly of inputs from edubloggers. People link to each other on social networks. Each person keeps his or her own list of 'buddies'. Here's mine. Empty; I don't use del.icio.us much. Here's someone else's network. Edubloggers are using dozens of networks - Friendster, Bebo, Facebook, Myspace, Twitter and more.

But that's not all. A lot of the chatter I see going on between people I'm connected to is taking place via email, Skype, instant messaging, and similar person-to-person messaging tools. People put people on their 'buddy lists' that they want to call and to hear from. They collect email addresses (and white-list them in their spam filters).

Communications maps are typically clustered. Like so:



The result is also observable. You get a clustering of distinct groups of people with particular interests. In the edublogosphere, for example, I can very easily identify the K12 crowd, the corporate e-learning bloggers, the college and university bloggers, the webheads (ESL), and various others.

This diagram is well known: it charts linkages between books read by bloggers:



This chart is semantic; that is, it depicts what the people talked about. This tells you about the flow of ideas, and not just the physical connections. And when we look at the flow of ideas, we see the characteristic cluster formation.

The network of people who talk about engineering is, similarly, a cluster:



Another way to spot the blogging network is to look at conference attendance. You can again find these clusters. I don't have diagrams of the edubloggers, but this conference attendee network of Joi Ito's is typical:



If we focus, not on a single physical indicator, but on the set of interactions taken as a whole, it becomes clear that the bloposphere is in fact a cluster-style network, and not a hub-and-spoke network. Bloggers form communities among themselves and communicate using a variety of tools, of which their blogs constitute only one.

1 comment:

  1. Thanks Stephen,

    That's very clear. If I understand correctly, you're saying this: if we only take one measure of a network it is likely to be skewed towards a power law distribution.

    Yet there are so many different ways people can interact, that power law gets flattened - since the hubs in one interaction method are not necessarily the hubs in another method.

    I won't be revisiting my little research project on the basis of this -- but it will certainly appear in the write up (if I'm allowed to quote a blog post that is!)

    Many thanks.

    ReplyDelete

I welcome your comments - I'm really sorry about the moderation, but Google's filters are basically ineffective.