Data science will fail to achieve its full potential until it lets go of this idea of “cheap and fast” analysis and embraces the world of the hard sciences with its balance of creative reanalysis of existing data with the (often expensive) creation of new data.