If you go to the right places you will see colorful signs that proclaim a creed which contains among its dogmas, “science is real.” Online, you will see many people arguing over “science” and “what science says.” It’s interesting that people proclaim their faith in, and debate about, something which they cannot define and which cannot be coherently defined.

The idea of “what is science” was debated throughout history, but very hotly in the late 1800s and early 1900s. It was called the “demarcation problem,” and it proved to be a failed endeavor. It is simply not possible to define “science” in a way that includes things that clearly are sciences and excludes things that clearly are not. (I will only give a brief sketch of why not, since it’s a big topic that justifies more than a little reading of books on the subject.)

First, we can dispense with the “scientific method” frequently taught in schools because (depending on how you interpret it) it either is generally not what scientists do or else describes what everyone does in all attempts to study the natural world. (See There Is No Scientific Method.) People who want to figure out how to read bird entrails¹ make observations, form a hypothesis, and conduct experiments to test the hypothesis. Paleontologists who study bones don’t conduct experiments².

As Lee Smolin put it in The Trouble With Physics, it can’t be that Science uses a method. Witch Doctors use a method. It can’t be that Science uses math, astrologers use math.

There have been many attempts at the demarcation problem, but I think that the two greatest efforts are Verificationism and Falsificationism. There have been various forms of these, but since this is a sketch I’m going to treat each as just one thing, since all of their forms have the same problem.

Verificationism defines science as those studies of the natural world whose theories can be verified. Falsificationism defines science as those studies of the natural world whose theories can be falsified.

The problem with both is that outside of some very simple investigations into the natural world which have already been done, you run into an infinite recursion problem in trying to decide what is a verification or what is a falsification of a theory. I’ll use another example from Lee Smolin’s book The Trouble With Physics, since it was so well chosen.

Suppose I have a block of marble and present the theory that there is a marble head trapped within it. I propose to test this theory by using a hammer and chisel to clear away the marble of the block from around the marble head. I do so, and lo and behold, there is a marble head. I made a prediction and then what I predicted came to pass. Did I verify my theory?

But suppose that I struck too hard with the hammer and the marble crumbled to dust. There is no marble head to be seen, only a pile of marble rubble on the floor. Has my theory been falsified?

(Another example of how falsificationism doesn’t work is the error in the orbit of Uranus. In theory, when Uranus wasn’t where it was supposed to be, Newtonian mechanics should have been falsified and the theory discarded. Except it wasn’t; if there was a Neptune-sized planet in Neptune’s orbit, that would also explain why predictions of Uranus orbit weren’t correct… and there turned out to be Neptune. So the falsification turned into an even better verification. The same thing with Neptune led to the discovery of Pluto, except that was by accident because the problem was actually with the estimates of Neptune’s mass, not the influence of another planet.)

Both verificationism and falsificationism fail, ultimately, because there is no way to instantly and certainly know that the verification or falsification means what one takes it to mean. There is always the possibility of experimental error, interpretational error, or some unknown factor which is influencing the experiments in unknown ways. Of the three, the unknown factor is the biggest theoretical problem, but I think that the issue of interpretation is the biggest problem in practice.

To see what I mean, consider things that used to be considered prestigious science which are now in disrepute. IQ studies that prove that white people are superior to everyone else, for example, or skull-measuring in order to classify races, or identifying criminals by the shape of their face. All of these considered themselves to be sciences. They all methodically measured things. They tested their theories and found their theories stood up to those tests. The problem, of course, is that they were grading their own tests.

But this is a problem everywhere. Physicists don’t call biologists in to decide whether their experiments legitimately give support to the theory they’re testing. Chemists don’t call in phrenologists to decide whether their experiments really show what they think they show. Biologists don’t ask electricians to judge their interpretation of fMRI data in brain activation. Physicists ask physicists, and phrenologists ask phrenologists, just as astrologers ask to be judged by other astrologers. This is what peer review means; it means that people who do not question the assumptions of a field judge particular ideas, experiments, observations, etc. under the lens of these shared assumptions. All peer review tells you is whether the work is, roughly, on par with the rest of the work in the field. It intrinsically can’t tell you whether the field is any good.

There are a great many people who are willing to grade their own tests and give themselves high marks; which of these people can we trust?

Well, that’s the problem. There’s no way to tell. All you can do is to look at their observations, experiments, interpretive framework, and judge for yourself. (Or you can find people you trust who will do this looking and judging, and have them tell you their conclusions. The number of people who do this out of the goodness of their hearts, rather than out of the desire for money or attention, which will bias their evaluations, is annoyingly small.)

So to come back to where this post started, it is simply not possible to hold that “science is real.” No one can say what the science is that might be real or fake. The word “science,” in that sentence, simply has no referent.

It would have been much better if the sign had simply said what it actually meant, which, IIRC, would have been, “anthropogenic climate change is real and is bad no matter what that change is because the climate is currently perfect.” That may be true or false, but at least it’s meaningful for those words to be put in that order.

---

1 I mean from scratch. Most bird-entrail readers, like most people who can administer a pH test to determine acidity, do so according to how they were taught by others with no attempts at observation or experimentation, but only following instructions that most couldn’t explain. But if someone wanted to resurrect the lost art of bird entrail reading, they’d begin by making observations—looking at bird entrails and also at what happened. Then they’d make guesses about what the bird entrails mean, then slaughter some more birds and see if these guesses were confirmed by fresh events. If they were, they’d gain confidence in their theories about what the bird entrails mean.

2 People will occasionally try to shoe-horn paleontology into this shoe by claiming that the “experiments” they do are predicting intermediate forms of animals then looking for them. The thing is, they don’t. They dig wherever there are bones and find what they find. Also, since the preservation of any given animal is overwhelmingly unlikely, the absence of anything in the fossil record is never taken to mean anything other than “we haven’t found it yet, if it’s there to be found.”