Everywhere Science is often presented to kids as a litany of facts. I believe that litany misses some of the most important points. I’m concerned that we don’t teach about what science is, its strengths and weaknesses, and how it should be valued. The result is that we live in a society that accepts as dogma that science has a monopoly on truth (at least when not in direct conflict with religion). Even the creationists tacitly accept this by trying to make their arguments “scientific”. And society yields truth to science even though most of us don’t know how to think scientifically — or even know if there is such a thing as thinking scientifically. This uncritical use of scientific terms for any purpose is pseudo-science. It is everywhere and it is dangerous. (OK, I like to sound dramatic sometimes.)
This sort of monopoly of truth scares me, even if it isn’t complete. Feyerabend makes some great points in this direction (read How to Defend Society Against Science for a quick taste and you’ll have a better idea where I’m coming from). But rather than dwell on my fear, I figure it is best to make sure I know what science is and isn’t so I can explain it to my kids and to anyone else who might be interested. Yeah, yeah, I know, they teach the “scientific method” in schools. But that is a distillation and idealization of centuries of experience. Science existed long before the method. And there is plenty of science that doesn’t fit the mold today. Anyway, I don’t think a method can define science. Who cares how someone came up with a theory if it is good?
So here I am going to start writing about what I think science really is. As always, I’m writing to test and organize ideas, not to proclaim any great answers. Don’t take me too seriously; I always have my doubts about stuff like this, but it is fun to write confidently. If you want to know my punchline, I’ll give it here: I think science is data compression and art, not a final arbiter of truth. And it is beautiful and valuable that way.
Even if I’m wrong about that — even if science does have some special status — I’ll accept no tyrant of truth. If forced to choose between truth and freedom, I choose freedom. If forced to choose between truth and beauty, I choose beauty. I think science gives us truth and beauty. And I think a person can make those choices and still be a good scientist.
What is Science?
I claimed that the goal of science is basically data compression, here’s why. The scientist produces a model of some part of nature that we can store in our minds. We can manipulate this model with logic to produce an infinite number of statements about nature. If it is good science we expect “most” of these statements to be true in some sense. We call these models theories.
My concept of a theory comes more from Mathematical Logic than from the usual Philosophy of Science (an area where I’m not very well read). In Logic, a theory is a set of statements that is logically complete — if there is any statement that must be true whenever all of the statements in the theory are true, then that statement is part of the theory. Loosely speaking, if a statement “X” and the statement “Not X” aren’t in the theory then all of the statements that are in the theory don’t say anything about whether X is true or false. For example, the statement “God Exists” is not in the theory of general relativity. So when I talk about a scientific theory, I think of a few laws or hypothesis and every statement that is logically implied by those laws.
I vaguely recall that the standard definitions of a scientific theory require that it use some “theoretical” or “unobservable” quantity. This never made much sense to me (I probably need to read more) — isn’t matter theoretical? All we can really observe is our sense data, but supposing that matter exists offers an awfully good explanation of that sense data. So any statements involving matter are “theoretical”. Um, that pretty much covers all of science.
OK, enough of that, let’s get back down to earth. A theory is a model of the world that we can store and manipulate in our minds. Or in our computers, or in a group of minds, or any sort of system we want.
So now I can say what I think science is. I believe science is the business of producing “good” theories. A theory is good if
It is concise
It contains a lot of “interesting” statements
You have to go pretty far before you start deducing false statements from it
It yields “interesting” arguments
Notice that aesthetics seem a little more important than truth in these criteria. A theory can be good because of its practical or its entertainment value. Hopefully it has some of both. But this is why I say science is art. Aesthetics are important.
I’d like to see if we can quantify some part of this definition. Then we could rate theories. I’m sure the measure would be fairly artificial, but it would be cute to compare some famous theories to see if we can better understand why they succeed.
Some Strengths and Weaknesses of Science
Why are theories so important? Because our minds could never catalog the facts of nature. The facts are overwhelming and we can’t even describe the facts of nature without resorting to theory! How do you describe the state of a system? Through its quantum states? Through quantities like energy, momentum, temperature, etc? These descriptions all come from theories!
So maybe facts are really just observations of sense data. Imagine if we were able to record all sense data collected by all beings. What good would that do you without theory? Even if you could store it all in your mind, why bother? Facts are fairly useless on their own.
But theories are different. Theories are amazing things! They allow us to store an infinite number of fact-like statements in a very small part of our minds. This allows us to face the world without being surprised and to have accurate expectations about what is coming. We can call up that part of our infinite knowledge that is relevant whenever we need it. Theories have clear survival benefits, and with my definition I’d say even frogs and flies have scientific theories. I think those fiesty biologists out there are selling themselves short when they proudly say “Evolution is not just a theory anymore”. I’d be more impressed if they said “Evolution is not just a fact anymore”!
There is a catch. Looking at theories the way I’ve described, it is fairly useless to talk about whether a theory is true or false. Do we say a theory is true only when every statment implied by the theory is true? If so, then of course the theory is false! Find me an interesting and valuable theory that isn’t wrong, and I’ll show you a theory that hasn’t been around long enough.
But I want to dismiss the notion with a little more care because it yields some interesting points. The fact is that even for pretty simple theories, we can’t even decide which statements are in the theory! If we can’t do that, how could we claim that every statement in the theory is true?
Let me elaborate with an example. Genetic regulatory networks are capable of universal computation (with a few difficulties related to data storage). We can design one set of genes that can take it’s initial chemical environment as input and run any program that you can run on your laptop. If we do this, we can ask a question like this: given certain initial conditions, will our network ever synthesize adrenaline?
It is impossible to write a computer program or logical process that can answer this question reliably. If you require it to always be correct, there will be some initial conditions that make your adrenaline-checker program run forever. If you require the program to always stop, then there will be inputs where it gives you the wrong answer. The problem is undecidable.
You can play the same trick with kinase cascade reactions, billiard ball collisions, and just about any nonlinear phenomenon in nature. Universal computation is everywhere. Undecidability is everywhere.
In any case, this shows us that even for relatively simple systems we can’t even decide which statements are in our theory. How will we ever use a logical process to decide that all of the statements in our theory are true when don’t know which ones to test. All we can do is falsify a theory, and all of the good theories I know are already false at some level.
And who cares anyway? What matters is that it is true enough. Or beautiful enough.
