Value of data has its limits

More thoughts on the idea of Experiment Design from the AIChE Annual Meeting.

Vassily Hatzimanikatis of Northwestern's Chemical Engineering department spoke on "Complexity in Biological Systems." While the talk itself was interesting, a quote from him really pricked my ears.

Our rate of understanding is inversely proportional to the amount of data we collect.

To clarify, as we first start learning about something any new data adds to our understanding of the area. Once our basic understanding is solid, new data give us nuances to the original theory and help clarify points that don't make sense. Eventually, more data is simply confirmation of our existing theories and do not add anything to our understanding.

This "understanding" can be in the form of theories or mathematical models. In either case, we need to know when to stop collecting data that tells us what we already know and move on to the next phase or aspect of our work.