User:Severoon/What Is Science?

What Is Science?

If I had to define science in a single concise statement, here's what I'd say: science is a system of reliably making predictions that is accessible to everyone. That's it. In my mind, this is precisely what separates science from other frameworks of thought such as religion. (Bear in mind that by accessible to everyone I don't mean "accessible" as a practical matter—not everyone can have access to a Large Hadron Collider, for instance—but as a matter of principle. In other words, there are no mystics with special access to an unseen higher plane of scientific knowledge or a science pope with a direct line to the god of science.)

Of course, there's a bunch of stuff I'm not addressing in that single short statement above about the scientific method and the philosophy of science, but in short, the quickest way to tell if you're dealing with science is to ask yourself if it allows you to make meaningful predictions.

A Disclaimer

Of course, none of the definitions provided here are meant to be complete or exhaustive. Rather, my goal is two-fold: to focus on the factors relevant to the most common misunderstandings about science and to disambiguate terms that are often overloaded with definitions from everyday speech. Many discussions about science and scientific topics quickly degenerate into arguments over semantics, which I aim to avoid by providing careful definitions up front.

Scientific Fact

The keystone of all science is the notion of scientific fact. A scientific fact is an empirically evident observation that is repeatable and reproducible.

Why such a complicated definition? Can't we just agree that a fact is simply a true thing? As it happens, because the scientific fact is the foundation of all science, we need to make sure we nail it down exactly so that the discussion doesn't get slippery, semantically speaking. So we can break it down, piece by piece, to understand what one means when one talks about scientific fact.

Empirical

In this case, empirical means guided by practical experience and not theory. In other words, a scientific fact does not depend in any way on a theory; a scientific fact is invariant with respect to the theoretical context in which it was observed.

This does not mean that a fact is totally context-independent, just that it is theory-independent. For example, if I drop a ball I might say that it is a fact that it accelerated towards the ground. Someone might dispute that fact by pointing out that, if I let go of it while it was resting on something, it could not have accelerated towards the ground. So I must express the setup of the ball's physical context before my statement of fact is meaningful; in this case, I must make sure it is well understood that the ball was not resting on something at the time, that I was near the Earth's surface, etc. So, if I'm going to report that I observed a bunch of facts, it is necessary for me to describe the relevant physical context of those facts. This is why every scientific experiment must take great pains to carefully describe the setup.

On the other hand, it is not necessary to describe theories of any kind if one wishes only to establish facts. Facts are facts—the hypothesis that happens to be under test in any given experiment or the canon of currently accepted theory at the time of the experiment has no impact on the existence of a fact. So, facts are empirical; they depend upon practical experience and not theoretical models.

Evident

A scientific fact must be evident. If the procedure that manifests an observation of the fact is carefully executed, the fact should be obvious to the interested observer.

Observed

A scientific fact must be observed. This is an often glossed-over property of facts because in the common parlance, any statement that is generally accepted to be true is often called a statement of fact. This is not so in science, however. A scientific fact refers to a particular event or characteristic of an object that was witnessed at some point in the past.

This is important for several reasons, but chiefly, this approach enforces traceability of the fact and ensures that it is a statement about something derived from actual experience—a specific instance of reality in action. When speaking informally, one might say, "It is a fact that if I drop this ball it will fall to the ground." Scientifically speaking, this would not be a fact; it is a prediction. In order to state a scientific fact, one would instead discuss a past event, "Last time I dropped this ball, it fell to the ground."

This is a very subtle but important difference.

Repeatable

A scientific fact must be repeatable. That is, a fact must be observable arbitrarily many times under the same relevant conditions.

Reproducible

A scientific fact must also be reproducible. That is, a fact must be observable by anyone who sets up the relevant context in which the fact was previously observed and perform the experiment.

Scientific Theory

A scientific theory is a model, restricted to a well-defined problem domain, based on scientific facts, that generates testable predictions.

In short, once we have collected enough facts, we may start to synthesize them into a model that suggests other possible facts to us. (One may guess whether a particular "suggested possible fact" will indeed turn out to be a scientific fact or not. This guess is called a hypothesis, and is tested by performing an experiment.) This "synthesis of facts" is the beginnings of a scientific theory, but it is not quite a theory yet. Before such a model can be called a theory, we must first know the domain of problems to which it applies. For example, I might develop a model that says when I release objects, they accelerate in the direction parallel to my body and towards my feet. But if we performed a series of experiments, we would find that this model only makes useful predictions if I'm standing upright on a massive body such as Earth when I release the object. Once we test several predictions made by this "proto-theory" we begin to get an idea of where it applies (standing upright on Earth, the moon, Titan, etc) and where it does not (empty space, Earth if I'm standing on my head, etc).

Because the boundaries of any theory must be well-defined, it is necessary to restrict the domain as much as possible, even excluding areas in which the theory applies until those areas are tested and can be included definitively. Naturally, science strives to make theories as general as possible, because the utility of a theory scales up with the size of its problem domain.

Good and Bad Vs. Better and Worse

Obviously, a theory that can make useful predictions in a superset of another theory's problem domain is a better theory, presuming its predictions are at least as accurate. Having said that, we come to another common pitfall in understanding science. And that is: while a theory can be "better" or "worse" than another theory, no theory can be objectively "bad" or "good" in and of itself.

A theory cannot be "bad" because, if it has attained theory-hood, that means it makes some kind of useful prediction in some kind of well-defined problem domain. No matter how restricted and small that problem domain happens to be, it is quite possible that the predictions of the associated theory are highly useful...and how could that be "bad"? (One could argue that a theory which has the degenerate problem domain of an empty set is a "bad" theory. To this, I would point out that such a theory makes no useful predictions and therefore could not have attained theory-hood in the first place.)

Similarly, a theory cannot be termed "good" in an objective sense. Placing value judgments on scientific theories is nonsensical—if a theory produces useful predictions then people will find it beneficial to some degree or another, but at what point does such benefit cross over into the realm of being "good"? This depends purely on how one defines the term good, which could legitimately vary from person to person based on whether the theory-in-question provides any kind of benefit to that individual. In other words, arguing about whether a theory is "good" or not is a highly subjective discussion that's really more about semantics than it is about science.

It might seem paradoxical, then, that one theory could be "better" or "worse" than another. But, if you think about it for a moment, it should become readily apparent. While one cannot objectively place value judgments upon a single theory, it is possible to objectively compare two theories and deem one better, or more useful, than the other. The main considerations are: (a) which theory makes more accurate predictions? and (b) which theory applies to a larger domain of problems? If one theory's problem domain completely subsumes the other's without sacrificing accuracy, then it is without question the better of the two because it allows for all of the same predictions and more. If the problem domains of two theories overlap in one area but also have areas exclusive to each, then it makes sense to simply keep both of them around. In some areas of the overlap, one of the two might produce more accurate predictions; in that case, that theory is the better one for those areas of the shared problem domain.