One of the Internet’s preoccupations recently — at least in the parts I frequent, which admittedly probably aren’t that representative — has been the physics, if you like, of data. There’s a lot of writing out there, much of it (like this piece from Tom Armitage) excellent, about treating data as a material; understanding it as a physical thing with seams, grains, symmetry and cleavage planes. This is true, but I’m beginning to suspect that it’s incomplete. Data’s not solely a static thing. It bounces, usually somewhere behind your eyeballs.
I don’t want to get unnecessarily tree-in-the-forest about this, but if no-one sees a piece of data, it doesn’t exist. Data only exists inasmuch as it supports either communication or decision-making. When it’s not being looked at, whether by a person or a process, it’s as if it had never been at all. So if it’s a material, it’s a profoundly weird one. It’s not even as material as fields are; anything with mass bends space-time. Unobserved data is truly weightless.
So right now, I prefer not to think of data as, primarily, a material. In fact, I prefer not to think of data as having any independent existence at all. Instead, I think about processes. How do people make decisions? How do they communicate and support their viewpoints? These processes are supported by technology, but they’re rooted in psychology and economics. When you view data mining this way, it looks a lot less like physical mining and a lot more like the newer, nerdier brother of user-experience design. User experience design is, loosely speaking, applying the lessons of graphic design to the design of intellectual processes. Data science is, loosely speaking, the application of mathematical and information-theoretic ideas to the design of intellectual processes. They’re both part of a wider field; that of service design.
Half the battle’s already won; people, like me, with science backgrounds are reading the design literature and trying to follow along. We’re waiting in the middle. When recently-graduated designers start talking about dimensionality reduction, though: well, then we’ll really be moving.