Looking Glass Worlds

Abstract

Ordinary science proceeds in the following way: once we have our ground rules, assumptions, protocols, and data, we are ready to make incremental advances in our chosen field. But how do we decide whether our paradigm is better than another? We know how to compare hypotheses, but how do we compare paradigms that involve the whole infrastructure of our research program, including the language we use? Statistical and logical tests often are of no help. Formal logic tells us that failed predictions and counter examples are not enough to falsify a proposition. As failed predictions and conflicting evidence accumulate, one might make continual adjustments to the theory. But it seldom occurs to a practitioner to jettison everything and start afresh with a blank slate; there is too much baby in the bathwater.

Additional details