strategy+business Winter 2013 : Page 73

B E S T B U S I N E S S B O O K S 2 0 13 / D I G I T I Z AT I O N (and other organizations and individuals as well) are recognizing a growing number of novel ways to apply it. Thus, Target can deduce when specific women have just become pregnant—or are likely to become pregnant—from the patterns in their purchases. Google Flu Trends competes with the Centers for Disease Con-trol and Prevention in predicting influenza outbreaks by tracking billions of Web searches for flu symptoms and related subjects with a half billion different algorithms. The SecDev Group can identify the location of prob-able cease-fire violations in geopolitical conflicts within 15 minutes. High-frequency traders can buy and sell stocks in microseconds, based on ultrafast analysis of all the stocks traded a microsecond in the past, a practice said to account for more than half of all stock trades and flash crashes. Scientists at HP Labs can successfully forecast the box-office success of films by look-ing at the rate at which relevant tweets are posted. The list of prof-itable applications of big data is far longer than this—and growing fast. Viktor Mayer-Schönberger, a professor at Oxford University, and Kenneth Cukier, an editor of the Economist, plumb this phenom-enon in their book, Big Data: A Revolution That Will Transform How We Live, Work, and Think . They base their bold subtitle on three assertions. First, big data is qualitatively different from sam-pled data, yielding insights that are possible only when the size of your sample is close to the totality of the ob-served population. The really, really big picture can re-veal details that were invisible with less than near-total sample sizes. Look through the right algorithmic lens and you can see things in big data that you can see only with big data. Second, big data enables valuable forecasting of a wide swath of phenomena through the use of correla-tion, even though causation may be unknown. “Society will need to shed some of its obsession for causality in exchange for simple correlation,” suggest the authors, “not knowing why but only what.” Correlations that can be acted upon profitably are good enough to justify the use of big data. Third, big data is messy and imprecise: “We don’t give up on exactitude entirely,” the authors write, “we only give up our devotion to it.” Theoretic understand-ing and precision might not be needed to profit from knowing just in time about the right messy, unex-plained correlations. Much of big data starts out as a side effect of hu-man activity, without much intrinsic perceived value. This is so-called data exhaust: the amount of time our cursors hover over icons on Web pages, the daily price of butter in a million grocery stores, or the locations of le-gions of mobile phones minute by minute. The authors of Big Data regard this as data ore: a store of potential value that can be transformed into tangible value (only) through the extraction of useful knowledge and its ap-plication, whether that is to sell more units of a com-mercial product or to more effec-tively deal with natural disasters. Mayer-Schönberger and Cuk-ier aren’t uncritical cheerleaders for big data. They know that process-ing some of this data ore—the data connected to what we previously thought of as privacy or anonym-ity—may have toxic results. Con-sider AOL’s release of anonymized data about millions of its users’ behaviors for legitimate social sci-ence research in 2006. By applying big-data analysis, a journalist was able to lift the veil of anonymity and identify specific people—akin to the way Iranian revolutionaries pieced together shredded documents when they invaded the U.S. embassy in 1979. So although the NSA’s leaked PRISM program did not purport to collect the contents of citizens’ communications, the metadata gathered could inevitably reveal an enormous amount about them. If “dataveillance” on this scale doesn’t give you the willies, consider that algorithms similar to those used to analyze flu trends could be used to predict which indi-viduals are likely to commit crimes. A society that stops crimes before they occur? There was a movie about that flavor of police state called Minority Report. The authors of Big Data caution that the dangers of pervasive data-veillance are as real as the opportunities they foresee. Harnessing Collective Action best books 2013 digitization 73 Big data involves computers and networks slicing and dicing the artifacts of human and machine behavior in

Previous Page  Next Page

Publication List
Using a screen reader? Click Here