Trying to use data to directly measure things that are better suited to intuitive understanding. Smart people often take for granted that the human brain is sti

Trying to use data to directly measure things that are better suited to intuitive understanding. Smart people often take for granted that the human brain is still far better at recognizing patterns than it is at mathematical modeling, and the human brain is still many orders of magnitude more powerful than any computer. Data-driven approaches are great for some domains, but are still poorly-suited for others. Common manifestations of this trap include using code coverage metrics to decide where to write tests and using measurements like cyclomatic complexity to decide when to refactor code.