Despite the benefits, Mayer-Schönberger and Cukier warn of a "dark side":
Big data often tells us that two things are related without explaining the underlying cause. For example, data once revealed that orange cars were half as likely to have defects; while the reason was unclear, the correlation alone was valuable for predicting vehicle reliability. Transformation Across Key Sectors Big Data: How the Information Revolution Is Tra...
Companies like Netflix and Amazon use "data exhaust"—the trail of digital interactions we leave behind—to forecast hits and provide personalized recommendations. Secondary uses of data, such as using global transaction records to forecast GDP, often hold more value than the data's original purpose. Despite the benefits, Mayer-Schönberger and Cukier warn of
Predictive analytics are used to identify early warning signs of infection in premature babies before symptoms appear. Large-scale genomic sequencing is also enabling personalized medicine tailored to an individual’s genetic profile. Secondary uses of data, such as using global
Traditional statistics rely on small samples to represent a whole. Big data allows us to analyze nearly every data point in a set, which eliminates sampling errors and lets us "zoom in" on small subgroups without losing reliability.
In the past, data had to be meticulously cleaned because any error in a small sample was catastrophic. With massive datasets, a sense of general direction is often more valuable than knowing a phenomenon down to the "inch or atom".
Google demonstrated big data's power by analyzing search terms for "flu" or "cough medicine" to predict the spread of H1N1 faster than official government statistics.