What is science?

Jason Rosenhouse of Evolutionblog expounds:

Science is best viewed as an activity undertaken with a specific goal in mind. That goal is to understand the way nature works. We measure our understanding by the extent to which we can make nature's phenomena predictable and controllable. Any investigative technique that brings us closer to this goal can reasonably be considered part of science.

All of the standard pieces of the scientific method we learned about in high school - experimentation, hypothesis testing, inductive reasoning and so forth - have their role to play in bringing us closer to our goal of predictability and control. By contrast, hypothesizing the actions of ill-defined supernatural entities such as ghosts or poltergeists do not help us move closer to our goal. Consequently, the actions of supernatural entities play no role in modern scientific discourse. The day someone finds a way to use such an hypothesis to bring clarity to some confusing aspect of nature is the day scientists will embrace the supernatural.

Many of the terms that get thrown around in this discussion - such as testability, falsifiability, or methodological naturalism (MN)- are really just ways of saying that scientists care about predictability and control. Saying that scientists adhere to MN in their work is really just a shorthand way of saying that science is a very pragmatic enterprise, and that the naturalistic hypotheses are the ones that have historically proven useful to scientists. It is a phrase that accurately describes the way scientists approach their work, and it survives because the only alternative - methodological supernaturalism - has proven itself time and again to be utterly ineffective in bringing scientists closer to their goal.