In one of my science textbooks, I make the statement that science cannot prove anything.1 I am always surprised at how controversial such a matter-of-fact statement is to some people. Almost every year, at least one student or parent will contact me simply aghast that I would write something like that in a science textbook. After all, science has proven all sorts of things, hasn’t it?
Of course it hasn’t. In fact, it is impossible for science to prove anything, because science is based on experiments and observations, both of which can be flawed. Often, those flaws don’t become apparent to the scientific community for quite some time. Flawed experiments and observations, of course, lead to flawed conclusions, so even the most secure scientific statements have never been proven. There might be gobs and gobs of evidence for them, but they have not been proven.
Karl Popper probably wrote the most important book related to this concept, which was titled The Logic of Scientific Discovery. Interestingly enough, he originally wrote it in German and then rewrote it in English. As a result, it is one of the few books that is published in two different languages but was never translated. The author wrote both versions. In this book, he argues that science should follow a methodology based on falsification. He shows quite clearly that while science cannot prove anything, it can falsify ideas that are currently thought to be true. He therefore argues that the test of any real scientific theory is whether or not it can be falsified. If not, then it is not truly a scientific theory.
There are a lot of scientists who disagree with Popper that falsification is the key to whether or not a theory is scientific. However, few would argue with his point that science cannot prove anything. Indeed, the journal Science seemed to forget this fact for a moment, but an astute reader chastised the editor, who admitted he was wrong.
The reader’s name is Charles L. Bennett, and he wrote a letter to the editor saying:2
The title of the 6 May News of the Week story “At long last, Gravity Probe B satellite proves Einstein right” (p. 649) made me cringe. I find myself frequently repeating to students and the public that science doesn’t “prove” theories. Scientific measurements can only disprove theories or be consistent with them. Any theory that is consistent with measurements could be disproved by a future measurement. I wouldn’t have expected Science magazine, of all places, to say a theory was “proved.”
Dr. Bennett is correct, of course. The editor, Colin Norman, admitted that in his response, which appeared right under the letter:
Bennett is completely correct. It’s an important conceptual point, and we blew it.
Unfortunately, as long as science magazines and teachers are sloppy enough to keep using phrases like, “science has proven,” it will be hard to teach children the truth.
1. Jay L. Wile, Exploring Creation with General Science (Apologia Educational Ministries, Inc. 2008), p. 38
Return to Text
2. Charles L. Bennett, “Science Title Misstep,” Science 332:1263, 2011
Return to Text
13 thoughts on “Science Can’t Prove Anything”
Thank you for saying this. I have so surprised people when I have mentioned that many science laws are ones that haven’t yet been disproven!!
Well put, Zaneta!
I agree completely – this is a fundamental principle that I managed to grasp early. What is sad is how many people don’t understand it – including many scientists.
Shevrae, that is so true. Even some scientists don’t understand this very important property of science.
Thanks you for this astute article and refreshing point of view.
Thanks for reading it, Carmen!
Okay, this is hilarious. I didn’t know Prof. Norman was an editor of Science, but I did know Prof. Bennett knew philosophy of science. But they’re both astrophysicists, so I won’t be taking any of their classes anytime soon.
Anyway, I’m not sure how much of hypothetico-deductivism I buy, but I don’t think induction works. And I don’t know how far I’d believe in falsificationism. The Logic of Scientific Discovery is on my bookshelf at the moment, but I don’t think I’ll have time to read it for a while.
So they are both at the same university? I didn’t know that. I definitely don’t buy into the idea that it’s not science if it can’t be falsified. I think both evolution and creation are science, for example, but neither of them can be falsified.
Yes, they are.
Interesting article! It reminds me of the book “I Don’t Have Enough Faith to Be an Atheist” by Norman Geisler and Frank Turek. In it they assert that “we can be sure [about something] beyond a reasonable doubt, but not sure beyond all doubt.” because we base our knowledge on inductive reasoning. The question that I immediately think of is: at what point in gathering evidence for our inductive reasoning, can we be sure enough of a hypothesis in order to work with it?
For instance, Special Relativity and Quantum Mechanics are at odds with each other in some ways, which means that either one or both are in some way false or incomplete. Yet, as you mentioned before, we are sure enough about them in order to use them in applications where a lot of property or even life is at stake. At what point did we become “sure enough?”
Another thing, correct me if I’m wrong, but it seems to me that in the history of science there has been a slow trend from direct experimentation to more mathematics used in order to test theories. Is this true? If so, is this good? It seems to me that no matter how mathematically sophisticated something is, it doesn’t necessarily mean it accurately reflects reality.
I would love hearing your thoughts on these matters.
Thank you for the article and your time!
Thanks for your excellent comments and questions, Enoch. It is really general relativity that conflicts with quantum mechanics, not special relativity, but I understand your point. It is an excellent point, of course, but you have to be careful of what you mean by “sure enough.” You are quite right that quantum mechanics and general relativity cannot both be 100% correct. They have mutually-exclusive views of the nature of both space and time. However, in their own realms, they each work quite well. When it comes to atomic scales, quantum mechanics can reproduce known data and predict new data with uncanny accuracy. Thus, when it comes to the atomic scale, we can be “sure enough” about the predictions of quantum mechanics to use it confidently. In the same way, on macroscopic scales, general relativity can reproduce known data and predict new data with uncanny accuracy. Thus, on macroscopic scales, we can use general relativity confidently. What we can’t use confidently is their assumptions about the nature of space and time.
In summary, then, you become “sure enough” about a theory when it can be used to predict and reproduce a wide range of data. However, even then, you are only “sure enough” about the theory in the range over which those data test the theory.
You are also quite right about a slow trend from direct experimentation to more mathematics when it comes to science. Aristotle, for example, thought it absurd to use mathematics in trying to understand the natural world. However, modern science as originated by the Medieval Christian Church started accepting a small amount of math in science. By the Renaissance, mathematics in certain fields of science (like physics) became the norm. Nowadays, nearly all scientific fields have mathematical models in them.
I would say that overall, this is a good thing, because it has produced more stringent tests of theories. If I am to have confidence in a theory, I want to see it predict data. The more precisely the predictions can be confirmed, the more faith I can have in the theory. Now where this is not a good thing is when the data are ignored due to the mathematical sophistication of the model. For example, those who claim that global warming is human-induced and a major threat rely on global climate models that are mathematically very sophisticated but do a lousy job of predicting the data. Many scientists are willing to “excuse” this inability to predict the data and continue to believe in the models, mostly because of the models’ mathematical sophistication. This, of course, is nonsense, but it is not the fault of the models. It is the fault of scientists who seem to have forgotten what science is rooted in.
Numbers are the Supreme Court of science. However Godel proved that we may not prove everything. There are Physics Foibles!!
Comments are closed.