This document discusses the limitations and realities of big data and predictive analytics. It provides two key insights: 1) After three years of effort by data scientists, Netflix was only able to improve their movie rating predictions by less than 0.1 stars. 2) Predictions suggest that human behavior is inherently random, limiting the success of consumer modeling based on available data. While targeted advertising has improved prediction rates slightly, the best methods still only predict a small minority of behaviors. The document concludes that big data has its biggest potential in artificial intelligence applications rather than making highly accurate predictions about human decisions.