It is important to recognize that although science is a rule-based procedure, it is very much a creative process. A conjecture is a philosophical invention, cooked up rather mystically by the mind through the mental computation we call careful contemplation. However, until the hypothesis is tested against reality, it is not yet truly knowledge; it is just information that represents speculation. Knowledge is information that has demonstrated its usefulness. It is what is left over after cycles of experimental testing have eliminated false theories. As scientists continually test their hypotheses and modify their models to account for new and surprising data, a kind of "learning loop" emerges that statisticians call Bayesian updating. Based on Bayes' Rule, developed by eighteenth-century English statistician and philosopher Thomas Bayes, Bayesian updating refers to a mathematical process whereby an accepted theory or predictive model gets increasingly accurate through the repetitive testing of competing variants of that theory.