Data Mining FTW

Wired editor-in-chief Chris Anderson is a pretty smart guy -- this article is pretty inspiring and nicely stated.

While I don't feel that model-based analysis is completely outdated, I do agree that having mind-bogglingly huge amounts of data available allows for statistical techniques to be used to give awesome and predictive results. The comment about the status of physics as a 'discipline starved of data' carries some truth, but it applies more to the string-theory family of models whose hypotheses reside at the Planck scale. There is plenty of space for models testable by current accelerators or those coming in the next few decades and hopefully they will receive more interest as comments like these demonstrate the public's view of current physical research. A really interesting development is the combination of statistical analysis and models as a unified tool rather than competing ideas.

In all though, the mathematical treatment of data is bread and butter now; it is the quickest way to draw (frighteningly) accurate conclusions about groups of people and all sorts of objects and can be used in so many different fields, not just large-scale systems like finance and sociological trends. Any businessperson so inclined could gather data and generate reports about his own business and market and reduce both his need for personnel and overall risk. Correlation isn't causation, but in many practical matters it serves just as well.