What do you do with great masses of data, if the task argues against paring those infini-bytes down to only what’s required for the problem you’re trying to solve? The answer is: You take the data as is, to learn what it can do.

The expansionist approach to data explores that data before funneling it into a nice, neat model. Specifically, I mean experimenting with lots of different ways of looking at the data before you spit out any results. Whether you call this experimental process, iterative analysis, data science, or something else, isn’t the point. What’s important is that, in the legacy view, the app decided what to do with the data. In the expansionist model, the data is in the driver’s seat.

It’s the difference between organizing data ahead of time for its intended use, and organizing it at the moment of use for the problem at hand.