Mapping Internet Worlds

Mashups based on Google Maps are pretty common. For one, maps are cool. Second, the API is so convenient and powerful. And viewers (myself included) are suckers for the a panoramic view of the world.

I remember Twittervision as the original full-browser map with automatic jump-around showing tweets with approximate location. WikipediaVision applies the same idea to wikipedia page edits.

Both are cool, and serve as a great reminder of the world-wide, near-real-time flows of data.

Abstract visualizations on the other hand, focus on the activity in a meta-verse without regard to location. Most of the ones I know are related to news, such as the newsmap. Digg has several interesting ones, the stack and the swarm are my favorites.

But, really, I think there is an opportunity for richer, more interesting and informative visualizations that evoke and key off a sense of space. Such visualizations can create and reinforce a map of the Internet world. Eventually, the places on such an Internet map can become as familiar and evocative as those on a Mercator world map.

Several people I know would argue for a 3D map with rotating objects, perceived depth and simulated horizons. I believe that richer, more useful ways of escaping flatland exist and work better for 2-D display devices. Which is all I think we will have for the next five years, until we get to retinal displays or direct neural manipulation.

Google Map mashups bring in cartographic features (forests, oceans, towns, states) which are uninteresting in conjunction with the information streams that are being visualized. We need entirely new features for these kinds of maps. While, Google Earth and several other services provide a “layers” approach for mapping additional data it is tied to a coordinate system of latitudes and longitudes.

It is difficult to think of a new coordinate system in the abstract. But here is an approach which can lead us there. Let’s start from the geo-coordinate system, and create a projection into a defined hyperspace. For example, in the twitter vision case, we can combine a user’s geographic information with relationship neighborhoods (as derived from follows-follower graphs). Our transformative matrix and placement algorithms then creates a new twitter-map showing mega-twitter-cities of densely connected groups, rural homesteads of sparse followers and hub-cities of multi-networked people.

Over such a baseline of cities and neighborhoods, we can visualize transient data (such as links, tweets, etc.) along the temporal dimension. Or we can go beyond ephemeral representation, etching those patterns into pathways between the”twitter-cities” based on the flow of links between the clusters. These could be large highways (e.g. mainstream links) or a network of small roads for niche topics or obscure #hashtags.

If we build a few such concrete maps across multiple domains, we may be able to define a new abstract coordinate system of consensus.