Ever since I was a teenager, I’ve been a great fan of Isaac Asimov. Asimov was a science fiction writer who came up with many ideas that now seem more or less achievable. (His autonomic, intelligent robots aren’t here just yet, but who knows where we’ll be a few decades from now). I had to think of him in the aftermath of the presidential elections in the United States. An election fueled by big data, according to many. One article caught my eye in the aftermath: “Gay vote proved a boon for Obama”. According to this article, both candidates scored about equal among ‘straight’ voters, but the gay voters overwhelmingly voted for Obama and got him reelected. So perhaps next time, we should just ask the gay people to vote, since they determine the outcome? This is more or less the premise of the story ‘Franchise’ by Asimov. In a far away future, a powerful computer can calculate the outcomes of the election by sampling an extremely small portion of the voting public: just one person. The trick is of course to find this one person who can serve as this powerful indicator. (And in the case of the gay vote, it may not be such a good predictor for the next election). Some Science Fiction Fan Scientists have taken this notion into science and are using the term “Asimov Data set” to indicate the smallest possible dataset that contains all features necessary for analysis. Big Data is doing many wonderful things to this Asimov Dataset. When we find more correlations and patterns, we may derive more information from less data. This may sound like a contradiction but it’s not: if we learn that all cat-lovers prefer Beethoven over Bach, we no longer have to have access to a ‘musical preferences’ database to infer your taste in music. All we need to know is what type of pet you have. (.. to learn about your personality?) On the other hand, the more we want to know, the more variations and the more samples we’ll need to find all possible combinations. There is probably a golden rule hidden in human behavior that describes just how different and how much ‘the same’ we are. Everybody is unique, but people behave pretty identically. If only we knew which part is unique, and which part is identical? So how does this relate to business? We don’t call it by this term, but we are always on the lookout for our Asimov Dataset. We want to know which of our clients are the early indicators of changing preferences or new needs. We want to have this ideal cross-section in our customer boards and in our test-panels. These are the people we want to be our facebook friends, these are the people we want to be ‘intimate’ with (in a pure business sense, of course). So perhaps this should be the prime focus of all social media analyses: not to find opinions or sentiment but to find the ideal set of people who can ‘predict the outcome’. The people who can help you predict the future. All you’ll have to do is ask the right questions and then listen carefully. Which sounds very much like one of those science fiction stories of the past.