From Pretty Pictures to "So What"
When I used to present academic papers, there are generally three ways in which my audience would critique me. I refer to these arguments as the “it’s been done before”, “it’s wrong”, and “so what?” questions. I have always been able to prepare for the first two points by knowing the area and methods well enough. Early in my career, however, I disregarded the “so what” critique. My thinking was that if my work was carefully done and relatively novel, I would be successful. In fact, I learned the hard way that answering the so what question is as, if not more, important than the others. My work was only important if it had a real impact on real-life people and the organizations they worked in. Put another way, if a train is new and runs well, but no one needs a ride to where it’s going, then no one is going to ride it.
To us at JANYS, “so what” is the most important question we ask ourselves. We know that, along with the excitement generated by terms like big data, social media, and social networks, there is a fair bit of skepticism out there about the actual value of this work to business. People want evidence (of accuracy) and relevance (to their business). For us, the best science, cool user-interfaces, and TED talk-like visuals are worthless unless we can answer the most important questions our customers are asking.
For the past three months we have been working with our strategic partner, IHS, in asking a single question of their clients: “So what is most important to you?”. We have sought to understand where there is a fit between their needs and our technology, with an eye towards answering needs that leverage the strengths of our technology. We have learned a LOT about a number of the best companies in the world, and how they are looking for meaningful answers that address their biggest concerns. Leveraging the years (and in some cases, decades) of experience of our IHS counterparts, we have been able to identify key ideas, data sources, and terms into our models. This industry knowledge (or priors, for the Bayesians among you), has given us a head start about what to look for in the mass of historical and real-time data we use.
Over the next few weeks we will be highlighting some of what we have found in our initial analyses of sentiment, social networks, and event analysis using our technology. We think that these are good examples of how our technology and capabilities can help answer some of your own so what questions. We look forward to hearing from you.