I love science and the search for truth. This month, as part of my more-conscious effort of sharing the things I love, I’m sharing a video that gets at a core concept that is widely misunderstood about The Scientific Method: that it is ultimately and fundamentally a human enterprise. First, the video, which is required viewing for the rest of the post.
So, there’s a lot to unpack here, and I’m not sure that a blow-by-blow analysis will really enlighten anyone. Instead of that, I’m going to respond to a comment posted on the video and then revisit a question Brady asked during the video. First, the comment from YouTube user calvarycome:
I think we need to distinguish clearly between the messy creative process of coming up with scientific theories, a process which depends a lot on social factors and chance, versus the more structured and systematic process of testing them, which doesn't. The latter is what we normally mean by 'the scientific method'. The discovery of graphene is no exception. No one would have believed that graphene existed without a systematic followup to that messy cellotape business.
So this is one of the more upvoted comments on the video, and almost all of the actual replies to this comment are positive, saying things like, “I think you’re right on the money.” I’m sorry to report that, in fact, calvarycome has missed the point entirely.
It’s understandable. It seems clear that calvarycome is a scientist, and the realization that The Scientific Method itself is a fallible, human enterprise is not obvious or straightforward, especially for scientists. Most scientists have loved science since they were children, and their view of the scientific method as an inevitable, infallible march to capital “T” truth is an important part of why they love science. Professional scientists have, as Upton Sinclair would point out, their salary and livelihood pushing them to not bother with this understanding. Thus, in a strangely ironic way, most scientists have also never seriously studied the philosophy of science or thought deeply about why and how The Scientific Method really works.1
So what, specifically, is calvarycome getting wrong? They suggest that there’s a “messy creative process” for creating scientific theories that “depends a lot on social factors and chance.” This is a false separation. There is no part of The Scientific Method that is not influenced by social factors and chance.
Showing that The Scientific Method is influenced by social factors and chance is most easily accomplished by examining scientific scandals. The Schön scandal is a solid example of this. Jan Hendrik Schön was an award winning scientist credited with numerous breakthroughs in semiconductor physics before it became obvious that his results were unreproducible and fraudulent. Researchers have faked studies on acceptance of gay marriage, the possibility of an AIDS vaccine, and done all manner of unethical, untruthful things in the name of science. Retraction Watch is a great place to follow if you’re interested in seeing some modern examples of this sort of problem.
To which, you might say (or calvarycome might say), so what? They were caught. Their work couldn’t be replicated, and replication is part of the scientific method.
First, how much replication is “enough” for truth? In the video, Professor Moriarty talks about how we don’t really know what that 101st replication would produce. And in the context of scientific fraud, we would have to replicate every finding however many times you decide is “enough” before we can declare we’ve caught all the frauds. Of course, we may have a pretty damn good guess given the previous 100 experiments, but we cannot know in an important sense what the result will be. That is, any answer to the question of how many replications is enough to be certain we have finally reached capital “T” truth relies on an inductive argument: that the previous experiments can be “generalized” into truth. From a logical perspective, this is not certainty. It is probability, even if that probability is very nearly certain.
Second, how precise do our measurements have to be? Accurate measurement of our observations is crucial to conducting science, but when do we know we’ve gotten an accurate enough measurement to have arrived at capital “T” truth? This is why Professor Moriarty talks about precision in the video. Newtonian mechanics might seem like the one and only truth if you didn’t have access to some pretty spectacular equipment. What scientific “truths” do we currently hold that only appear like truth because we don’t have the ability to measure or observe some important, latent piece of reality? From a scientific fraud standpoint, we can point to things like P-hacking or other abuses of statistical measures to demonstrate a “finding” that may actually be meaningless or actively wrong. In some sciences, this sort of statistical abuse is so widespread that half or more of the published results in some fields are not replicable. This is what scientists refer to as a replication crisis, and it’s something I’ve complained about previously.
Third, how much “true” knowledge is rejected by the scientific community? The graphene example that calvarycome uses to bolster their argument is telling in that it was rejected twice by major journals prior to winning the Nobel Prize. What if the authors had just given up? How long would it have been before someone else attempted to publish something similar? How do we even know that someone would have attempted to publish something similar? All the “science” that John Snow employed in creating his map of the 1854 Broad Street cholera outbreak didn’t matter all that much. The community at the time rejected his theory and replaced the handle on the pump after the outbreak ended. How do you even measure this sort of “lost” science that is proposed, reviewed by The Scientific Method, rejected, dropped, and then never (or at least not yet) revisited?
In short, the existence of known scientific fraudsters or bad-but-published methodological investigations raises the simple question: Are there others who haven’t been found yet? If so, who? If not, how do you know? What science can we trust? Rejected “good” science raises the question of whether we’re generating all true knowledge about the world by using The Scientific Method. So, here we are right back in the messy world of humans, not just in the “creative process” used to generate hypotheses, but throughout The Scientific Method.2
So, ending on this note would be rather sad, and luckily, we don’t have to end on this note. We can trust science, and I want to talk about how in relation to Brady’s question during the video at around the 10:30 mark. Brady says, and I’m paraphrasing, “It sounds like you’re just stoking the fires of climate change deniers who are saying that you can’t trust The Scientific Method. And you’re just saying, ‘Well, yeah, that’s right.’” That is, how can we trust science if we know and accept that The Scientific Method is not flawlessly leading us to actual truth?
The modern world is ridiculously complex. No one will ever be able to survive in it without trusting other people, even with their own lives. I trust the engineers who built my office every minute I spend inside it. I trust the teachers at my daughter’s school to teach her well when she goes there. I trust accountants, lawyers, civil servants, and scientists with matters great and small every day. This notion of trust was made explicit in the information security realm by Ken Thompson’s Turing Award Lecture, titled “Reflections on Trusting Trust.” Essentially, his argument is that ultimately we have to trust the people who built the software we use, not the software itself. Otherwise, we would have to personally write every line of code on the systems we use, from the assembler, to the compiler, to the operating system, and on up to the cat video player itself. So when a climate denier says that they don’t trust the scientists who claim that climate change poses a threat to civilization, they are not methodologically wrong because the exercise is ultimately an exercise of trust.
Still, I believe climate change deniers are factually wrong. Why? Well, it’s not as though they have somehow accessed capital “T” truth where the scientists haven’t. Their belief structures ultimately rely on trust as well. Who are they trusting? Oil and gas companies? Opinion pieces from political think thanks funded by oil and gas companies? Their own guts? Where are they getting their information? It is not as though we somehow “knew” the climate wasn’t being dramatically and negatively affected by human actions and that scientists are “disproving” this either. All knowledge depends on trust, so what you think of as “true” ultimately depends on who you trust, which is a fundamentally human process. This is why philosophers define knowledge as “true, justified belief.” It requires justification, and The Scientific Method has a great track record. Personally, I’ll trust the people who have presented reams and reams of compelling scientific evidence that climate change is real, vaccinations work, and so on. But I won’t make this mistake of positioning this knowledge as somehow beyond criticism because it was produced by The Scientific Method.3
To end this post with another comment from the video, Ashley Johnston says:
I suggest that it is more about differentiating the scientist in his role as scientist and his role as messy human. To state it clearly, ‘A man who does science is always a scientist and everything he does is science’, this is absurd and frankly dangerous. With this method of understanding the idea of science is lost and it just tracks the practices of a community of so called ‘scientists’, whatever they may end up doing. Which would assuredly end up being the opposite of science until they had vamipred out the last of the value in its reputation.
This demonstrates a much more accurate understanding of how science really operates than calvarycome’s earlier comment. The Scientific Method is simply a mechanism that produces trustable results, not infallible results. The long, successful history of results generated from The Scientific Method bolsters that trust. You don’t have to look far to see that The Scientific Method has materially improved your life. But we could still be wrong, maybe even about important things. If we stop questioning knowledge simply because someone labels it as “science” or because it came from a “scientist” then we have failed fundamentally at understanding the point of The Scientific Method.
My favorite introduction to some of the philosophical problems is to ask a scientist how they know that The Scientific Method leads to truth. Ultimately, there’s an assumption that if we keep generating hypothesis, testing them through observed, careful experimentation, and refining our hypotheses based on the results of the experiment that we will reach capital “T” truth. How do we know with certainty that this assumption is itself true? Should we test it with The Scientific Method? ↩
And we haven’t even gotten to the funding of science and how that affects the “truth” we get out of it… ↩
Of course, you’re going to have to have some seriously compelling evidence to criticize these findings. ↩