Thursday, January 10, 2013

Thomas Kuhn's "Scientific Revolutions"

The New Atlantis has a great essay commemorating Thomas Kuhn's book The Structure of Scientific Revolutions.
Kuhn held that the historical process of science is divided into three stages: a “normal” stage, followed by “crisis” and then “revolutionary” stages. The normal stage is characterized by a strong agreement among scientists on what is and is not scientific practice. In this stage, scientists largely agree on what are the questions that need answers. 
This is probably where most research lies.  A lot of work that follows the same outline of previous studies is done.  Much of it is great work that's important, but the design features of the studies follow a previous template.
A crisis occurs when an existing theory involves so many unsolved puzzles, or “anomalies,” that its explanatory ability becomes questionable. Scientists begin to consider entirely new ways of examining the data, and there is a lack of consensus on which questions are important scientifically.
The catalyst for this crisis can be a new discovery, like RNA interference, for example.  Or the discovery of RNA splicing.
Eventually, a new exemplary solution emerges. This new solution will be “incommensurable” — another key term in Kuhn’s thesis — with the former paradigm, meaning not only that the two paradigms are mutually conflicting, but that they are asking different questions, and to some extent speaking different scientific languages. Such a revolution inaugurates a new period of normal science. Thus normal science can be understood as a period of “puzzle-solving” or “mopping-up” after the discovery or elucidation of a paradigm-shifting theory. ...  But since every paradigm has its flaws, progress in normal science is always toward the point of another crisis.
 Taking the RNAi example further, recent years have seen many RNA interference papers that follow the same general design.  Now that the "RNAi paradigm" has been firmly established, what will the next big crisis point in biosciences be?