When I worked in the Obama campaign’s new media department, one practice I particularly admired was the reliance on data in decision making. Which title should we use for this fundraising email? Run a test where one randomized group gets one title, one randomized group gets another title, and see which group donates more. Where on the homepage should we put the donate button? What color and shape is most likely to illicit a click? Which ad graphic is most likely to result in a sign-up from a woman age 18-35? For all these questions the campaign relied on the scientific method to decide which of a number of choices was best. Of course, a human being had to come up with those choices before they could be tested, but, whenever possible, a conclusion was tested before it was implemented. Intuition was never good enough.
If the Internet has given us any one thing in abundance, it is information. Every click, every friend request, ever video view can be logged and linked to other user behavior across the Internet and in our day-to-say lives. One day I gave a positive review to a commercial for Heifer International on Hulu and a few days later I received a fundraising letter in my mailbox. (People who like the ad are probably statistically more likely to donate.)
The Internet presents us with a bonanza of know-ability. We now have more recorded information about our actions than ever before in human history, especially if those actions are carried out online. In his 2003 book Six Degrees: The Science of a Connected Age, Duncan Watts of Yahoo! Research writes that, on the Internet, “data are recoded automatically…. By distributing the effort of data entry to the members of the network themselves… the main limitation to data recording is virtually eliminated, and the resulting data bases can essentially grow without bounds.”
One of the challenges of the Internet is actually dealing with the mass of data now available. This summer, Stanford University actually held a workshop on dealing with massive modern data sets with sessions on topics like “Massive-Scale Analytics of Streaming Social Networks” and “Geometric Network Analysis Tools”. Crowdsourcing tools like Ushahidi need specific applications simply to help them process the deluge of data that is submitted. Scientists now rent time on “clouds” of off-site computers because they need more computing power than their institutions possess.
But the answers to some of the most fundamental questions about the effects of digital technology are still unanswered. For example, what do we really know about digital activism? We define by anecdote, lean on pre-digital theories of social movements, and change the definitions as new cases arise. Yet the answers to the most interesting questions elude us: If digital technology is a tool in a cat and mouse game between government and the opposition (as it is in many repressive societies), who is winning now and what tactics can activists employ to give themselves the upper hand? Does digital technology simply increase the effectiveness of previous activism tactics or create new types?
Digital technology both automates and alters the status quo. It strengthens the power of current institutions (better-targeted corporate advertising through shared user data) and upends them in unexpected ways (corporations that sell hard copies of textual, audio, or video content are going out of business). Which trend is stronger now and which will be in the future?
Computing power offers us easy answers to simple questions: What color? How much? A or B? What is the likelihood that Sarah knows Tim? But where causation is diverse and leaks into the real world, computing power hasn’t helped us yet, particularly in the social sciences. Is this a limitation of our current methodology or a hard limitation of the power of technology to reveal itself?