Social Visualization, Looking Inside Out

Over the last couple of weeks, I’ve been thinking about different aspects of data visualization (see an earlier post).

Patterns in the Ecosystem

Since my first job on the web as a graduate assistant information designer, I’ve been interested in web metrics, thanks in large part to my friend and mentor, Rebecca Bernstein, the founder of the web team at the University at Buffalo, and someone who possesses the rare combination both quantitative investigation and qualitative creativity. She spear-headed the use of metrics as both a research and marketing tool at our university. Web statistics were the unspoken user feedback that geeks, marketers, and the administration loved. Regardless of the initiative, if we had statistical data on the site’s usage patterns and visitors, we could convince the administration that the ideas we had to evolve and grow site features or make refinements were completely justified. We used these metrics to guide our design and development strategies. With a combination of frequent informal user testing and web tracking, we could check our solutions and tweak where needed to provide the best user experience possible.

Fast forward to 2007 and those metrics are still good identifiers of visitor habits, but the web has expanded to cover a great deal more than the “page.” I listened to Brian Oberkirch’s excellent Edgework podcast where Jeff Veen talks about the development of Measure Map, the web analytics tool for bloggers (direct mp3 audio). Veen describes the true innovation of Measure Map as a new approach to a technology that already existed, and built specifically to serve an audience whose needs weren’t being met by existing software. One simple example is the switch from using page views as the metric, but rather, focusing on posts and unique entry ID’s instead – a much more relevant and interesting statistic for blogs. By presenting data visualizations of web traffic and conversations related to one post, bloggers can see the impact of a single idea, and its conversational net effect. This slight shift in paradigm opened up endless possibilities for new solutions.

With my current work with clients on social applications like Stylehive, we’ve used data and metrics in different ways – adopting game playing techniques to make data fun. In most social applications, data grows exponentially through the ecosystem each time a member acts or plays: whether that’s posting a unique bookmark, copying someone else’s, adding tags, adding friends, importing contacts, posting comments… the list continues. As a member, you’re interested in your own stats (points, total bookmarks, number of followers, tags, comments) and everyone else’s (popular bookmarks, recent sites, active members, new followers). If we treat the application as a system of visualization and interaction, it’s easy to enable discovery and to provide incentive, identity and status. All of these interactions keep members engaged and playing as your system grows and evolves. Whether it’s data tracking or visualizations, these tools and techniques reveal the behaviors and relationships of live interactions on our sites.

Social Data Analysis

What we’re also seeing these days are two other types of tools: web services that allow you to upload your own data for graphing and charting; as well as conversations and experimentation online with public data sets, or “social data analysis.”

I recently added the visualization service, Swivel, to eHub and then saw Many Eyes, a new site by IBM.

Both Swivel and Many Eyes explore data visualizations by allowing you to upload your own data set to graph and chart with online tools. Many Eyes is beta and doesn’t let you save data that you’ve uploaded at the moment. Both sites encourage conversations among site visitors and the open exploration of public data sets.

When asked how they “compared themselves to Swivel” in a post by Tim O’Reilly, the team behind Many Eyes answered, “…Swivel seems to have some neat data mining technology that finds correlations automatically. By contrast, we’ve placed our emphasis on the power of human visual intelligence to find patterns. My guess is that both approaches will be successful because social data analysis is a powerful idea.”

Swivel allows data upload with four methods: copy and paste into a web form, spreadsheet (CSV, Excel), import from a web page with tabular data, or a document that contains a table (PDF). One you upload data, you can preview and make various configuration changes, add an image, and preview. Once you create a graph, it’s easy to share the link to the page, or embed the graph into your post with a snippet of HTML.

Here’s an example at Swivel of the US Government Budget for 2006:

Many Eyes excels by providing multiple visualization methods to create visualizations from new or existing data sets. The methods include the standard bar charts, pie charts, and line graphs, but also include more specialized views such as a network diagram, scatterplot, bubble chart, block histogram, and tree map. Any data set that’s already been added to the site is available for “remixing.”

Here’s an example of top sites that get the largest share of attention as a treemap:

Of course, many of these graphs aren’t compelling on their own and data visualization is an entire field of its own used in a wide cross-section of industries. For some examples, visit one of my favorite sites – visualcomplexity.com, “a unified resource space for anyone interested in the visualization of complex networks. The project’s main goal is to leverage a critical understanding of different visualization methods, across a series of disciplines, as diverse as Biology, Social Networks or the World Wide Web.”

But, what’s interesting about the new visualization services and trends online is that the technology is beginning to come into the hands of enthusiasts and individuals (eventually, en masse) rather than solely available to business or specialized industries. Just as publishing became easier with blogging tools, visualization services are going to become more accessible as well. What will happen when we can also be statisticians with our own and the “world’s” data?