How does science work? Various arguments have been made for the logical validity of science. For example, there is the idea of induction. The idea is that a scientist gathers a lot of evidence, and then by hard work and thought, he works out a theory from the evidence that explains all the evidence. Then he goes out and finds more evidence. Each piece of evidence that matches his theory is support, and a piece of evidence that does not match indicates the theory must be revised.
It sounds pretty reasonable, but there are two problems. First, it does not work logically. Secondly, it’s not how science is actually done.
The logical problem is this: there is no way to logically derive an explanation from a set of observations. At best you can note common points between all the observations, perhaps even a mathematical pattern. But that is not an explanation and it cannot logically be extended beyond the original observations. There are an infinity of possible explanations which all fit any set of observations, and no way to choose between them. Also, additional pieces of evidence that agree with a theory do not, logically, support it. Again, there are an infinite number of explanations that could fit the extra data.
This may seem like logical and philosophical nitpicking. Does it really matter if we can logically justify science? It works doesn’t it? This is true, but it is somehow unsatisfying to say scientific theories are likely to be correct because science has worked in the past. In fact, one might note that scientific theories change, and use that as an argument that actually the scientific method doesn’t have a great track record. Since none of those earlier theories, now abandoned, resemble what we now think of as the ‘truth’ how can we have any certainty that what we have now is any closer to the truth?
Fortunately the second “problem” I mentioned comes to the rescue. Science isn’t actually done using logical induction. Let’s take a look at the supposed inductive scientific method again:
Evidence -> Theory -> More Evidence -> Theory is either falsified or supported.
Take a look at the first step, from evidence to theory. In general this is not how new scientific theories are invented at all. A scientist does not gather reams and reams of data looking for a pattern. Even if they do come up with their theory by examining a lot of data it is not derived from the data. People make up theories. A theory is an invention of a human mind, not an inference from evidence.
Theories come first, so the actual activity of science is more like this:
Theory -> Compare to Evidence and other Theories -> Repeat until a winning theory emerges.
It is important that a theory be compared to other theories as well as to evidence. This allows us to find the best available explanation. The best available explanation is the theory that conclusively fits the evidence better and “makes more sense” than all other available explanations. This eliminates the problem of an infinity of just as good explanations that we cannot choose between. We can never be certain that we have the “true” explanation in hand, but we can be sure that the explanation we choose is better than any other explanation we have considered.
Some would say that we should not use the best available explanation, because we cannot be certain that it is the True explanation. That, however, is simply ridiculous. Despite the fact that we cannot be certain a better explanation will not come along in the future, we don’t have that explanation now and it may never come along. It only makes sense to use the best explanation you have available to you. The alternative is not to try to explain anything at all.
Others may object that the fact that a theory comes first leaves science open to interpretation of evidence in the light of a “favorite” theory when it would fit a different available (but disfavored) theory better. This is true, to a point. Scientists will be human and will attempt to fit the facts to their favorite theory. The assurance we have that this will not totally falsify science is twofold. Firstly, science offers fame and fortune to the scientist who can convincingly overturn an established theory, and there is no limit on how many times you can try. The established theory will be endlessly under attack by competing theories until a better theory is found. This continuous attack is in large part what establishes a theory as the best available explanation.
The second assurance we have is that nature doesn’t care what we think. A theory that is propped up by ‘special’ interpretations of the evidence will fail to make successful predictions or match new evidence. No theory at odds with reality can survive for long if someone tries to use it.