Small changes make big differences
One of the issues of using so much data, is that no matter how carefully you choose your variables, no matter how you select for randomness or representivness, small changes in variables or leaving out a variable that wasn't a concern before can lead to serious changes in the outcome. All you have to do to see that is adjust the interest rate earned over 40 years in a retirement planning algorithm. A mere half percent can tip the scale from balanced to broke or to that round the world trip you always wanted to take. The best thing to remembe with big data, is that even though it's big, it's only as good as the people using it and their assumptions.