The End of Theory? Unlikely!

Wired Magazine posted an essay online that proposes that:

1. We have a ton of data and because of how much information we can store (what Wired calls “The Petabyte Age”), we can always have it on hand.
2. The incredible amount of computing power we have allows us to sift through data and run every possible statistical test on it until something comes up significant.

From the article:

There is now a better way. Petabytes allow us to say: “Correlation is enough.” We can stop looking for models. We can analyze the data without hypotheses about what it might show. We can throw the numbers into the biggest computing clusters the world has ever seen and let statistical algorithms find patterns where science cannot.

The author of the article, Chris Anderson, is not a scientist, and it shows. Researchers already get enough false correlations without running thousands of statistical tests - five or ten is often enough to find some significant correlation, let alone thousands.

What Mr. Anderson is proposing is a reversal of the scientific method. First, collect a bunch of data. Then, run thousands of statistical tests to see what correlated. Then, invent a theory to explain that correlation. It’s so painfully simplistic and shows such an utter lack of understanding about the way science is done that it’s laughable. In psychology, how do we know what data to collect? If we go about research atheoretically, then we would need to collect every possible piece of data about every person so that we can throw it at the computer to see what pans out. Without theory, there is no place to begin.

Sadly, even the editor-in-chief of Wired does not understand science well enough to write an informed article.