Three Myths About the Scientific Method

This is the third in a series on the scientific method.  This installment is a little bit of a tangent, but I wanted to publish it now because I've gotten tired of having to correct people about these things all the time.  I figured if I just wrote this out once and for all I could just point people here rather than having to repeat myself all the time.

There are a lot of myths and misconceptions about science out there in the world, but these three keep coming up again and again.  These myths are pernicious because they sound plausible.  Even some scientists believe them, or at least choose their words carelessly enough to reinforce them, which is just as bad.  Even I am guilty of this sometimes.  It is an easy trap to fall into, especially when talking about "scientific facts".  So here for the record are three myths about the scientific method, and the corresponding truth (!) about each of them.

Myth #1:  The scientific method relies on induction

Induction is a form of reasoning that assumes that phenomena follow a pattern.  The classic example is looking at a bunch of crows, observing that every one of them is black, and concluding that therefore all crows are black because you've never seen a non-black crow.

It is easy to see that induction doesn't work reliably: it is simply false that all crows are black.  Non-black crows are rare, but they do exist.  So do non-white swans.  Philosophers make a big deal about this, with a lot of ink being spilled discussing the "problem of induction".  It's all a waste of time because science doesn't rely on induction.  Any criticism that anyone levels at science that includes the word "induction" is a red herring.

It's easy to fall into this trap.  The claim that all crows are black, or all swans are white, are wrong, but they're not that wrong.  The vast majority of crows are black, so "all crows are black" is a not-entirely-unreasonable approximation to the truth in this case, so it's tempting to think that induction is the first step in a process that gets tweaked later to arrive at the truth.

The problem is that most inductive conclusions are catastrophically wrong.  Take for example the observation that, as I write this in April of 2024, Joe Biden is President of the United States.  He was also President yesterday, and the day before that, and the day before that, and so on for over 1000 days now.  The inductive conclusion is that Joe Biden will be President tomorrow, and the day after that, and the day after that... forever.  Which is obviously wrong, barring some radical breakthrough in human longevity and the repeal of the 22nd amendment to the U.S. Constitution.  Neither of these is very likely, so we can be very confident that Joe Biden will no longer be President on January 7, 2029, and possibly sooner than that depending on his health and the outcome of the 2024 election.

How do we know these things?  Because we have a theory of what causes someone to become and remain President which predicts that Presidential terms are finite, and that theory turns out to make reliable predictions.  Induction has absolutely nothing to do with it.

Induction has absolutely nothing to do with any scientific theory.  At best it might be a source of ideas for hypotheses to advance, but the actual test of a hypothesis is how well it explains the known data and how reliable its predictions turn out to be.  That's all.

Myth #2:  The scientific method assumes naturalism/materialism/atheism

This is a myth promulgated mainly by religious apologists who want to imply that the scientific bias against supernaturalism is some kind of prejudice, an unfair bias built in to the scientific method by assumption, and that this can blind those who follow the scientific method to deeper truths.

This is false.  The scientific method contains no assumptions whatsoever.  The scientific method is simply that: a method.  It has no more prejudicial assumptions than a recipe for a soufflé.

Even the gold-standard criterion for a scientific theory, namely, its ability to make reliable predictions, is not an assumption.  It is an observation, specifically, it is an observation about the scientific method: it just turns out that if you construct parsimonious explanations that account for all the observed data, those explanations turn out to have more predictive power than anything else humans have ever tried.  That is an observation that, it turns out (!) can also be explained, but that is a very long story, so it will have to wait.

The reason science is naturalistic and atheistic is not because these are prejudices built into the method by fiat, it is because it turns out that the best explanations -- the most parsimonious ones that account for all the known data and have the most predictive power -- are naturalistic.  The supernatural is simply not needed to explain any known phenomena.

Note that this is not at all obvious a priori.  There are a lot of phenomena -- notably the existence of life and human intellect and consciousness -- that don't seem like they would readily yield to naturalistic explanations when you first start to think about them.  But it turns out that they do.  Again, this is a long story whose details will have to wait.  For now I'll just point out that people used to believe that the weather was a phenomenon that could not possibly have a naturalistic explanation.

The reason science is naturalistic is not that it takes naturalism as an assumption, but rather that there is no evidence of anything beyond the natural.  All it would take for science to accept the existence of deities or demons or other supernatural entities is evidence -- some observable phenomenon that could not be parsimoniously explained without them.

Myth #3:  "Science can't prove X" or "scientists got X wrong" is an indication that science is deficient

I often see people say, "Science can't prove X" with the implication that this points out some deficiency in science that only some other thing (usually religion) can fill.  This is a myth for two reasons.  First, science never proves anything; instead it produces explanations of observations.  And second, this failure to prove things is not a bug, it's a feature, because it is not actually possible to prove anything about the real world.  The only things that can actually be proven are mathematical theorems.

Now, you will occasionally hear people speak of "scientific facts" or "the laws of nature" or even "scientific proof".  These people either don't understand how the scientific method actually works, or, more likely, they are just using these phrases as a kind of shorthand for something like "a theory which has been sufficiently well established that the odds of finding experimental evidence to the contrary (within the domain in which the theory is applicable) are practically indistinguishable from zero."  As you can see, being precise about this gets a little wordy.

The scientific method gives us no guidance on how to find good theories, only on how to recognize bad ones: reject any theory that is at odds with observation.  This method has limits.  We are finite beings with finite life spans and so we can only ever gather a finite amount of data.  For any finite amount of data there are an infinite number of theories all consistent with that data, and so we can't reject any of them on the grounds of being inconsistent with observation.  To whittle things down from there we have to rely on heuristics to select the "best explanation" from among the infinite number of possibilities that are consistent with the data.

Again, it just turns out that when we do this, the result of the process generally has a lot of predictive power.  Some of our theories are so good that they have never made a false prediction.  Others do make false predictions, but to find observations that don't fit their predictions you have to go outside of our solar system.  For theories like that we will sometimes say that those theories are "true" or "established scientific facts" or something like that.  But that's just shorthand for, "The best explanation we currently have, one which makes very reliable predictions."  It is always possible that some observation will be made that will falsify a theory no matter how well established it is.

Finding observations that falsify well-established theories does happen on occasion, but it is very, very rare.  The better established a theory is, the rarer it is to find observations that contradict it.  For less-well-established theories, finding contradictory data happens regularly.  This is also often cited, especially by religious apologists, as a deficiency but it's not.  It's how science makes progress.  In fact, the best established theory in the history of science is the Standard Model of particle physics.  We know that the Standard Model is deficient, but not because it makes predictions that are at odds with experiment -- quite the opposite in fact.  The Standard Model has never (as of this writing) made a false prediction since it was finalized in the 1970s.  The reason we know it's deficient is not because it makes false predictions (it doesn't, or at least hasn't yet) but rather because it doesn't include gravity.  We know gravity is a thing, but no one has been able to figure out how to work it in to the Standard Model.  And one of the reasons we haven't been able to do it is because we have no experimental data to give us any hints as to where the Standard Model might be wrong.  This is actually considered a major problem in physics.

That's it, my top three myths about science debunked.  Henceforth anyone who raises any of these in my presence gets a dope slap (or at least a reference to this blog post).