>There’s only so many ways to solve a problem and there’s only so many problems to solve in a particular field. That is true today but not tomorrow. The field

>There’s only so many ways to solve a problem and there’s only so many problems to solve in a particular field. That is true today but not tomorrow. The field often changes. New ideas generate new ways to think of problems which require new experimental designs to solve. New experimental designs become possible with new technologies offering more resolution or fidelity. Entirely divergent technologies sometimes emerge which offer new metadata to compare with in the analysis, or change the context for other forms of analysis. What was cost prohibitive becomes cheap. All of this means you cannot merely write a few pipelines and expect them to be cutting edge forever. You cannot automate the very creative act of hypothesis generation. You cannot reliably develop an automated solution without first developing some sort of a test over a subset of the data or a simulated dataset, to ensure your solution to a given problem even has the power to resolve what you are expecting it to resolve.