What I have seen in my limited practice with machine learning and big data projects is that it is easy to fool yourself that your methods work. And the problem

What I have seen in my limited practice with machine learning and big data projects is that it is easy to fool yourself that your methods work. And the problem is that the people who are good at this get promoted, and those who find the mistakes are shunned.