Today marks the public release of Kindred Britain, a new interactive scholarly work that explores the role of family in British culture. Integrating geospatial, temporal, and network information visualization, this project attempts to demonstrate the genealogical interconnectedness of the British elite. In doing so it expands the notion of Britishness, and the notion of society and culture in general, such that through family ties we can see George Washington, Napoleon Bonaparte, Sigmund Freud, and Kevin Bacon all included among these kindred British. To achieve this took significant effort not only in writing code, but in envisioning how to present so much information of so many different kinds in a way that did not overwhelm a reader. I want to stress the concept of the reader here, rather than a "user" since to create a site like this one has to move away from the stereotype of the user with their short attention span and incapacity to grasp complexity. It doesn't pay to create works for users, and buy into the myth that no one reads anything on the Internet and that a website has to be designed for someone with a ten-second attention span. Works like Kindred Britain and ORBIS not only include much flashy and interactive content, but vast quantities of text. There is an entire sub-section of Kindred Britain, known as Notes from Kindred Britain, dedicated to explaining in long-form linear narrative how the site was built and how it was imagined. Integrated into the main site are various stories, about cultural topics and technical, that are wired into the information visualization elements (a topic I'll explore in detail at this years HTML5 Developer Conference). And each person in Kindred Britain is simultaneously displayed as a temporal signal, a geographic signal, a network signal, and an event signal, with this last aspect conveyed through text. Text is the critical component of Kindred Britain, and its integration not only into the site's design and display but its very information architecture is key to its success in relaying to readers its vision of Britain.
Taking nearly 16 months to plan, design, and build, Kindred Britain is the longest project cycle I've been involved with so far. Kindred Britain began as ORBIS was released in May of 2012, though some earlier explorations of the underlying network took place before that time. But this pales in comparison to the five years put in by Nicholas Jenkins to collect the various pieces of information that Kindred Britain draws from. The site's crisp and elegent design was due to our having the fortune to work with Scott Murray, who literally wrote the book on Interactive Data Visualization for the Web. Besides the prominence of text, the prominence of formal acknowledgment of authorship and collaboration is a key component of Kindred Britain, and reflects the changing nature of scholarly production at the university. The various long-form pieces that I wrote for Kindred Britain are not low-visibility line items on my CV, but prominent components of the project, which is necessary for success in digital humanities scholarship.
In fact, the exhaustive nature of the various pieces that I've written, not only about the cultural implication of the computational methods we've deployed but also about the general development path and nature of the digital objects created, leave little to cover in grey literature such as this. So, rather than recapitulate what is stated more thoroughly and in more detail in Kindred Britain, I'd like to focus on one point that is particularly topical given the attention given to a post by my colleague, Mike Widner, about the responsibility of digital humanities scholars to engage in social commentary about computational techniques used in espionage and otherwise to measure and predict the behavior of people:
In my two "stories" in Kindred Britain, I walk through how computational methods are used to measure importance and tragedy in Kindred Britain. In doing so, I point out that they are remarkably useful, if contingent, techniques for understanding patterns in such a dataset. That contingency, oft-alluded to by digital humanities scholars, is much more comprehensible and much more worrisome to me now, having learned how to deploy it, than it was when it was theoretical. My crude methods for measuring a limited and simple database are a pale imitation of the methods being brought to bear by the NSA or Amazon to measure and understand people. But they are of a similar kind. And in understanding them, like understanding other techniques like topic modeling, I better understand how they mis-measure and reinforce flawed assumptions and, most critically, produce random patterns of correlation that are so tempting to invest with narrative. I tried to explain how these methods were used in Kindred Britain because I thought it necessary and timely to educate not only a scholarly audience, but a public one, about the use and limitations of this kind of computational measurement of people. Just as our Tragedy Index misses tragic Jack Kipling, so does every computational method have significant blind spots. Kindred Britain has quite an audience--scholars and developers and designers and the public--interested in its many aspects--anglophilia, infoviz, data modeling--and it's my hope that along with learning a bit more about the nature of family in British culture, that this broad audience also learns a bit more about how they live as a node within their own database, a sort of Kindred Modernity, and what that means for society and culture.