What is science?

There are (at least) a couple of ways to define science. It's either a body of knowledge, or it's the process by which we gain that knowledge. Or both. Too often, especially when people who don't understand science try to teach it to others, the focus is on the body of knowledge. To the poor kid trying to stay focused until recess, science becomes a huge pile of seemingly random facts.

It's better to focus on the process. We add to the facts all of the time (and sometimes have to correct them), but the process stays the same. By learning how the process works, we can develop our critical thinking skills, which are always useful in today's crazy world.

Before going any further, I should say that many people brighter than me have spent their careers trying to tackle this one. My views on the method are based in part on the views of people like Thomas Kuhn and Karl Popper.

The goal of science is to understand the world about us. So basically science is a disciplined means of obtaining knowledge. It starts with two assumptions.

Combining these two assumptions, any conclusions we reach in our search for generalizing principles must be consistent with what we experience in common. And that's the key to the entire process. The goal of science is to find principles which can explain what we experience, and these must be consistent with the observable data. In order to separate the valuable ideas from the chaff, the process of science follows a standard method. The scientific method, while a very idealized version of what really happens, is a good starting point. Generally, you
(1) observe, or collect data,
(2) hypothesize, or make generalizations about your observations,
(3) make predictions from your generalizations,
(4) experiment to test the predictions, and
(5) draw conclusions from your experiment about your generalizations.

Most hypotheses fail the test, so then you'd go back to step (2) and try again. Sometimes, though, really good ideas keep passing the test, and they get accepted as a theory or law. An important point is that the quality of the conclusion is always limited by the quality of the data. Newton's Laws of Motion provide a good example. They worked perfectly well when developed in the 17th century, but don't fare so well now that modern technology allows us to move objects at speeds approaching the speed of light or observe the conditions around extremely massive objects. So Newton's Laws of Motion required some revision when confronted with this new class of data, and thanks to Einstein, we now have the Theories of Special and General Relativity.

As I said, the scientific method is an ideal. In geology, where most processes take millions of years, or in astronomy, where most objects are light years away, well beyond our reach, it's not easy to perform experiments. In astronomy, you'll typically see a paper presenting some interesting observation (step 1), and if it's really eyecatching, a few papers proposing explanations of the observations (step 2, and if they're really good, step 3, too), followed by papers presenting new observations to test the various hypotheses (steps 4 and 5). Lots of papers are combinations of the various steps, but few cover all of them by themselves.

Over time, if observations continue to agree with the hypothesis, confidence begins to grow that we're on the right track. But there's no way to prove the idea right. You can only prove it wrong. For example, I can't prove to anyone that Maxwell's Equations are the best possible way to explain electricity, magnetism, and electromagnetic radiation, despite the fact that they have been enormously successful in an amazing variety of applications. In fact, we have no way of knowing that there may be some superior means of explaining electro-magnetism, but we do know that in every situation we have encountered so far, Maxwell's Equations work remarkably well. That's a pretty good reason to keep using them, until somebody comes up with something better.

Karl Popper popularized the idea that a scientific statement had to be falsifiable. Since you can't prove something right, the best you can do is discard ideas because they're wrong. Popper argued that in order for a statement to be scientific, it had to be a statement which could be tested to see if it were false. An example of an unfalsifiable statement would be a story about something that happened to you or a group of friends long ago in the distant past. It may well be true, and all of your friends may agree that it happened, but unless each element of the story can be somehow checked independently, there's no way to prove that the story is true or false. It may well be a true story, but it's not scientific.

So somebody comes up with an hypothesis which is testable, and it passes every test proposed. How do we know it's right? We don't. We only know it has not yet been proven wrong. Thomas Kuhn has written a couple of books proposing his model for how science progresses. Somebody has a good idea, builds a theory, and over time as more and more data are collected, the theory has to be tweaked a little this way and that to fit the data, but by and large the theory survives.

The theory becomes, as Kuhn puts it, a paradigm. At some point, though, the paradigm shifts. A once-elegant theory may have required so many tweaks and modifications to keep up with new data that it's now an unrecognizable and cumbersome mess. Maybe someone came up with a simpler, more elegant way to explain the same thing. Or maybe some new phenomenon was discovered which is completely incompatible with the old theory. However we get there, this is a scientific revolution. The most famous example is the theory of Copernicus, which puts the Sun at the center of the Solar System, despite 2,000 years of astronomy which worked just fine with the Earth at the center. Using the theory of Copernicus, an astronomer could calculate the positions of the planets as accurately as before, and it was easier. What finally clinched it was the discovery by Galileo of things like the phases of Venus that were simply incompatible with the Earth-centered Solar System. Another, more recent, example is the rise of quantum mechanics a century ago to explain the behaviors of atoms which couldn't be done with classical physics.

Science isn't always right, but we have yet to find a better method. It's a self-correcting mechanism guaranteed to provide explanations that are consistent with current data. And it will modify the explanations as new and unanticipated phenomena come along. Sometimes it takes a while to get there, since scientists after all are just people with egos and pride like the rest of us. But even when some great name is able to prevent an entire discipline from rejecting his or her pet theory, sooner or later this person will fade and the evidence that they are wrong will accumulate to the breaking point.

The fact that you can't prove any explanation right is troublesome for some people who like everything to be more certain than that. But this is missing the point. By maintaining some discipline in how we approach problems and solve them, we can avoid many pitfalls along the way.

If you wish to read on, continue to the page on what we know.
Home Science

Last modified 1 September, 2006. © Gregory C. Sloan.