Way back when we were Persuadio, in 2004 and 2005, our maps were very similar to the map described by Matthew Hurst as an accurate depiction of the blogosphere. In the map below, influential nodes are displayed with names (you can't see them in the map displayed above, because there is a dense cluster of influencers in the lower left center of the network). We knew it was complete and a complete mess that conveyed very little useful information, if any.

Other examples of our mapping efforts during those early years included showing very granular link-level relationships and the cluster-and-connector style maps shown below. At every stage, we've labeled influential nodes so that customers could see who they were looking at, but as you can see, those names can get lost in the blizzard of information.

Link-level display

Cluster-and-connector graph of Jerry Michalski's social network in early 2005.

When we showed this last type of map to Esther Dyson, she commented that although there was not much useful information in the map—except that there were many connections to follow and explore—they were beautiful. Based on that, Esther was kind enough to invite us to show our mapping tool as a debutante company in 2005. But the maps were not ready for primetime, because we still needed to show the relationships between nodes in the network for customers to begin to understand what action they could take. For marketers and folks generally interested in influence, the problem is in trying to find: a.) who may be a good introduction to an influencer, or "an influencer of the influencer," and, b.) what the conversation looks like so that an entry can be strategized and relevant to the participants. We had to go deeper.

It's here that I'd point to Lilia Efimova's diagram of her relationship with another blogger, seen at the right, as an example of the kind of information that contextualizes a conversation. In this case, the participants co-authored their history, something we can't expect everyone or even very many people to do, because they have other priorities in creating and maintaining relationships than reducing their experience to a map. Lilia, whom I met at a Microsoft Research social networking forum last year, and Andrea Handl encountered one another by linking through their blogs, but expanded their relationship by sharing bookmarks, commenting on one another's blogs, then email and Skype. Their relationship evolved, adding dimensions of communication while the specific subject of their discussions changed, building on their widening knowledge of one another.

Lilia's map is so specific that a marketer would be overwhelmed if presented hundreds of maps like this to cover all the relationships an influencer had. Instead, they need to know whether two influencers interact through their blogs or published articles in the press, whether one primarily listens and repeats the other, and so forth. This does give them a good sense of where they may want to try to engage.

In our current UI, we reduced every influencer's neighborhood to a map that shows who is influenced by them, who influences them, the type of interactions they have (for example, if the sites interact through postings or comments, the line between them shows a bi-directional relationship)--and that's just one small part of the user interface.

Is it a comprehensive view of the network? No, but at this level, where the marketer is planning to engage the market, these are the details that are important. If we showed a single influencer within a big cluster the intuitive conclusion would be that they are well connected, but because the clusters tend to be so dense, there's no possibility of divining to whom the influencer is directly linked to. In our maps, you can explore the second- and third-degree relationships by dragging a node into the center to see the links it has in the network.

David Armano posed the issue as one of accuracy and intuitive impressions, but it is really a problem of representation and meaning, as he writes: "The visual may not be completely accurate as it's drawn from my own personal perspective as I was learning how the social network operated." Drawing on my editorial experience, I'd say the challenge is to pick how to tell the story of a networked conversation at each level and design to those requirements within a system that allows different types of people with different information consumption styles to explore and learn what they want quickly. That means leaving out some information in one view that will be important in another, because the extra notes distract from what is being represented at a given level of the conversation.

You've always got to mix words and pictures. Understanding human actions is never simple, the best we can do is approximate the meaning of their actions, measure the results of our interactions with them, then learn and apply the lessons of the experience.