11 Uncertainty Is Inherent in Science

The deepest misunderstanding about science . . . is the idea that science is about certainty. Science is not about certainty. . . . The very expression “scientifically proven” is a contradiction of terms. There’s nothing that is scientifically proven.

—Carlo Rovelli¹Close

We have seen in the preceding chapters that there are many uncertainties when it comes to the details concerning various questions related to climate change, vaccination, human evolution, genetic testing, and forensic science. After learning of such uncertainties, one might begin to worry whether this entails that these domains are inferior to other areas of science. One might reasonably think that good science ought to be certain and that whatever is filled with uncertainties should not be considered good. Unlike the beliefs we hold in our everyday lives (see Chapter 1), one might think that science ought to provide us with knowledge that is not plagued by uncertainty. Since these domains are faced with many uncertainties, perhaps they are not really scientific?

Unfortunately, as mathematician William Byers has explained, thinking that to be scientific is to be certain is “more characteristic of a mythology of science” rather than how science really is.²Close In fact, scientists recognize that there is always going to be some amount of uncertainty in science regardless of the advancements we make in constructing better and better theories, developing superior technologies and devices, and so on. Recognizing that all science is uncertain is important for at least two reasons. First of all, it is simply the truth about science. We cannot really understand science unless we appreciate its uncertainty. Second, this recognition helps us see that the fact that there are uncertainties about climate change, say, does not mean that we should not trust what climate scientists tell us or that we should not accept that humans are significantly contributing to climate change. In this chapter we take a step back from the particular details that we have been exploring in the preceding chapters to consider general features of all science that make it uncertain.

The mere fact that science is done by human beings introduces uncertainty into the picture. All humans are fallible. Aren’t scientists better at reasoning than the average person, though? Perhaps, but they are still humans, and, as such, they are susceptible to the same biases that we all are. For example, cognitive scientists Hugo Mercier and Dan Sperber reported that studies have shown that scientists fall prey to “myside bias” just like everyone else.³Close Recall from Chapter 1 that this is the bias where we find it difficult to appreciate evidence that runs counter to our own opinions. As a result, we tend to interpret the evidence differently depending on our prior convictions. Scientists make this same mistake, and the famous physicist Max Planck once claimed that “A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it.”⁴Close While things may not be quite as bad as Planck suggested, scientists are humans and are therefore vulnerable to biases and errors. Thus, they face uncertainty in their conclusions just like the rest of us.

But there is more than being biased. Scientists may intentionally discredit and disregard the available evidence in order to reach particular conclusions for political, ideological, or other reasons. Sadly, there are numerous such instances. Historians of science Naomi Oreskes and Erik M. Conway have described how particular scientists misled the public about the connection between tobacco and cancer, ozone depletion, and more. In these cases, it seems that particular scientists were either influenced by monetary gain, their political leanings, or both. What happened in these cases was that a handful of men who were well known and highly respected by virtue of their earlier work in the weapon programs of the Cold War presented themselves as experts on topics for which they really had no expertise. And what they did was question the findings of the science of the time by highlighting various uncertainties. How can we say that tobacco smoking causes lung cancer if most people who smoke do not develop lung cancer? Why do some smokers develop lung cancer while others do not? Thus, the tobacco industry, for instance, hired some well-respected physicists, such as Frederick Seitz, to argue that the science was still unsettled about the issue. As Oreskes and Conway explained: “Industry doubt-mongering worked in part because most of us don’t really understand what it means to say something is a cause. . . . Doubt-mongering also works because we think that science is about facts—cold, hard, definite facts. . . . Doubt is crucial to science—in the version we call curiosity or healthy skepticism, it drives science forward—but it also makes science vulnerable to misrepresentation, because it is easy to take uncertainties out of context and create the impression that everything is unresolved. This was the tobacco’s industry key insight: that you could use normal scientific uncertainty to undermine the status of actual scientific knowledge.”⁵Close

But even if scientists were completely unbiased, there are other issues stemming from the fact that the natural world is extremely complex. Philosopher Angela Potochnik has identified four specific ways in which our world is complex: (1) there is a large variety of different phenomena, (2) for any given phenomenon there’s an enormous range of factors that influence it, (3) the various influences on a phenomenon might differ in important ways from the influences on similar phenomena, and (4) individual influences can affect phenomena in complex ways.⁶Close Let us follow Potochnik and consider the complexity of the causes of obesity as an example. Take a look at Figure 11.1. This is an attempt to map the various causes of obesity. Each of the nodes represents identified causal influences on obesity, and the lines between them represent different strengths of interaction. The key point to recognize here is that a single phenomenon such as obesity is exceedingly complex. In fact, the condition is even more complex than the diagram illustrates. As Potochnik has noted, in such diagrams only some causal influences are depicted, and each of the represented causal influences is affected by numerous other causal influences that are not shown. Additionally, even though various strengths of influence are depicted, the complications in terms of how these various influences interact are not illustrated. Other seemingly straightforward phenomena and conditions are similarly very complex.

Therefore, when we realize that science is something that is performed by humans who are cognitively limited in important ways and done in an effort to understand things that are tremendously complex, it should not be a surprise that science is filled with uncertainties.

The facts concerning the complexity of the world and the cognitive limitations of humans help to shed light on another essential feature of science: it is thoroughly collaborative. It is in an effort to overcome their own limitations in the face of great complexity that scientists collaborate quite extensively, but uncertainty is a limitation that is not easily overcome. Scientific research is seldom a solitary affair. It is often conducted by a team of scientists—in many cases several teams. For example, one recent physics journal article had 5,154 researchers listed as co-authors!⁷Close Of course, this is an extreme case, but the fact is that many articles in scientific journals present the findings of collaborative works. According to a 2016 article in The Economist, as of 2015, the average number of authors on articles in science journals was 4.4.⁸Close What is more, even articles that have a single author likely depend on the work of other scientists to at least some degree, which explains all of the citations of previous work in the reference sections of these articles.

Although one might be tempted to think that collaboration in science is a recent development, it’s not. Historian of science Kathryn M. Olesko has argued that even the most prominent cases of “solitary geniuses,” Isaac Newton and Charles Darwin, were not really solitary.⁹Close They both relied on data collected by others as well as on the scientific work of others in developing their own theories. This does not lessen their achievements, but it does reveal that collaboration is not a new thing in science, and it is at the heart of some of the greatest discoveries in the history of science. To illustrate this point, consider the work of Gregor Mendel; he is usually portrayed as the father of genetics, a lonely pioneer working in isolation, who eventually discovered the material basis of heredity. It is exactly because of this isolation, the story goes, that his work published in 1866 did not become widely known until 1900 when three other naturalists independently reached the same conclusions he did. As a result, his priority was only acknowledged posthumously, and the world was deprived of his knowledge for 34 years. Sad story, right? Well, it would be if this was all there was to the situation. However, a closer look at what exactly Mendel wrote and the related historiography suggests that there is more to the story. Mendel was not trying to develop a general theory of heredity. He was rather trying to understand the development of hybrids and how heredity occurs in them from a practical point of view related to the agricultural context in which he was working. At the same time, a number of scholars including Charles Darwin, Francis Galton, August Weismann, Hugo de Vries, and several others attempted to develop theories of heredity contrary to what Mendel was doing. It was their work, and advancements in other domains such as microscopy and cytology, that eventually created a new context that allowed for a reappreciation of Mendel’s experimental approach. Simply put, our understanding of heredity advanced due to the work of a scientific community, not of Mendel alone.¹⁰Close

Scientific progress relies in a significant way on collaboration among scientists. Of course, such collaborations require scientists to trust one another and the information that they share. The old adage that a chain is only as strong as its weakest link provides an apt analogy here. For a chain to be strong every link must be strong. Similarly, for a scientific claim (or any claim for that matter) to be epistemically certain, every bit of information that is relied on in supporting that claim must be epistemically certain. Is it really plausible that every single collaborator to a scientific project is in a position to provide epistemic certainty for every result and calculation they contribute to the overall findings? It seems not. If we cannot be certain that each collaborator is certain about all of their contributions, we cannot be certain about the final product of the collaboration. There cannot be any weak links if we are to achieve certainty.

Of course, as we emphasized in earlier chapters, just because we cannot be certain it does not mean that all claims are equally worthy of acceptance. Far from it! Some sources are reliable, and others are not. Fortunately, there are a number of social structures in place to aid in assessing the reliability of scientific claims. In order for a claim to become accepted by the scientific community it must undergo a peer-review process by experts in the same field who evaluate the questions asked, methods used, data collected, and conclusions drawn. It is because of these sorts of evaluation processes that science produces knowledge that is so well supported by the evidence. Nevertheless, nothing can guarantee certainty. Sometimes peer review fails to weed out results produced by poor methodology. Those conducting the peer review process are fallible, too, so it is always possible that a mistake may allow poor-quality research to slip through the cracks. Thus, despite the social structures in place, uncertainties of various kinds exist.

At this point, we might be tempted to appeal to the scientific method that many of us were taught in school. Doesn’t this method make it so that the sorts of uncertainties we have been discussing do not infect science? Cannot we simply follow the scientific method and be guaranteed to reach the truth? Short answer: no. The main reason this is so is simply that such a thing as the scientific method does not exist. As historian of science Daniel P. Thurs has explained, the so-called scientific method (at least when understood to be some sort of universal, discipline neutral method) is a myth. Even an agreed upon account of what the purported method is does not exist. As Thurs noted, “even simplistic versions vary from three steps to eleven. Some start with hypothesis, others with observation. Some include imagination. Others confine themselves to facts.”¹¹Close If the very nature of the scientific method is itself uncertain, attempting to follow it is not going to yield certainty. After all, we cannot be certain that we have followed the scientific method when it is not certain exactly what this method is.

Let us assume that there is such a thing as the scientific method and there is no dispute about the steps of this method. Most representations of such a method agree that it involves making observations, formulating a hypothesis, testing the hypotheses via an experiment, and then, on the basis of the outcome of the experiment, accepting or rejecting the hypothesis.¹²Close Does following these steps produce certainty? It is worth considering each step.

First, consider making observations. You might think that we can be certain of our observations, can’t we? One problem, as we explained in Chapter 1, is that our senses can deceive us. Beyond that, the “theory-ladenness” of observation also suggests that the answer is “no.” In philosopher and historian Thomas Kuhn’s terminology, scientists interpret observations they make through the lens of their particular “paradigm” (“beliefs, values, techniques, and so on shared by the members of a given community”).¹³Close As a result, how we understand or interpret a particular observation depends on the particular paradigm with which we are working. We focus on particular aspects of what we observe because our paradigm deems them important. For example, as Kuhn explained, astronomers in Europe did not “see” changes in the heavens (such as the appearance of new stars in the night sky) until after the old view that the heavens are unchanging was abandoned for Copernicus’s paradigm. This is in stark contrast to Chinese astronomers who observed the appearance of many new stars long before Europeans did. Why was this? The Chinese did not have better equipment or techniques. Presumably, they were not more studious observers either. It also was not because the stars were only visible in China. The difference was that the Chinese were not encumbered by a view of the heavens as unchanging. Rather, they were working under a different paradigm than the one guiding the pre-Copernican Europeans, one that prompted them to take note of the evidence for new stars.¹⁴Close The content of observations in science is at least in part determined by scientists’ paradigms, and paradigms have changed throughout the history of science. Hence, there is uncertainty as to what exactly has been observed and how the observations are to be interpreted.

What about formulating a hypothesis? This seems pretty straightforward. Perhaps we can be certain that we have formulated the hypothesis we take ourselves to be formulating. Admittedly, we may not have come up with a true hypothesis or even a very good one. But, this is where testing comes into the picture. We test hypotheses via experiments and determine whether to accept the hypothesis or reject it. Uncertainty arises here. What happens when an experiment yields a result that is contrary to the hypothesis? Is the hypothesis rejected? It depends on whether or not the particular experimental result counts as a mere anomaly (something that does not fit well with the hypothesis but is not deemed important) or as evidence against the hypothesis. However, whether the result is considered an anomaly or evidence for rejecting the hypothesis depends on the way the result is interpreted. Is the fact that Mercury’s orbit around the sun does not proceed as Newton’s theory predicts a mere anomaly to be ignored or a reason to look to another theory? Prior to Einstein’s work on relativity, Mercury’s orbit was simply taken as an anomaly, but after Einstein proposed his theory that same result was taken to be evidence against Newton’s theory.¹⁵Close Similarly, for results from controlled experiments: whether a particular result is an anomaly or a reason to reject a given hypothesis depends on how it is interpreted. And, of course, like observations, interpretations of an experimental result depend on the accepted paradigm employed by those evaluating the results. Again, we face uncertainty in how we are to interpret the results of experiments used to test the hypothesis.

What about accepting or rejecting a hypothesis on the basis of experimental results? This step is uncertain as well for at least two reasons. First, given our fallibility it is always possible that we make a mistake in our inference here. We might think that an experimental result warrants accepting or rejecting a hypothesis when it doesn’t. Second, as Kuhn pointed out, the history of science shows us that scientists do not tend to actually treat experimental outcomes that conflict with their hypotheses as providing counters that demand the rejection of the hypotheses. Instead, such outcomes are treated as anomalies. And, in the face of anomalies, scientists tend to “devise numerous articulations and ad hoc modifications of their theory in order to eliminate any apparent conflict.”¹⁶Close A clear example of this is the numerous epicycles that were added to the Ptolemaic model of the universe in order to make the theory fit with the observational data. Rather than abandon this model, a number of ad hoc modifications were tacked on to it so that it could yield the correct observational predictions. The prevalence of this sort of modifying may be part of the reason for Max Planck’s cynicism that we noted earlier. It may be that “a new scientific truth does not triumph by convincing its opponents and making them see the light” because ad hoc modifications are the typical response to anomalies.¹⁷Close Whatever the case may be, it does not seem that scientists strictly follow this step in the purported scientific method at all. If the step is not even adhered to, it cannot provide certainty.

None of this is to say that science is purely subjective or relativistic. Rather, this is simply to point out that there is no scientific method that can give us certainty. We have already discussed how our ordinary thinking leaves us with uncertainty. And, as Albert Einstein said, “the whole of science is nothing more than a refinement of everyday thinking.”¹⁸Close Scientific reasoning is the same sort that we employ in our everyday lives. As cognitive scientists Hugo Mercier and Dan Sperber explained, “Scientists’ reasoning is not different in kind from that of lay people. Science doesn’t work by recruiting a special breed of super-reasoners but by making the best of reasoning’s strengths.”¹⁹Close The way that science makes the best of reasoning’s strengths is by implementing the sorts of social structures that we mentioned earlier: peer review, discussions, independent testing, debate, and so on. We explore more fully how scientists reason in the next chapter. For now, it is enough to recognize that their reasoning won’t guarantee certainty.

Although science does not attain certainty by way of the scientific method, one might think that it does so through mathematics. Many contemporary scientific theories are expressed as mathematical formulas. And when we want a clear example of reasoning that is precise and certain, mathematics is a good place to look. After all, what can be closer to certain than math?

Unfortunately, for those set on science being certain, appealing to mathematics will not do the trick. Mathematics is itself uncertain and does not provide us with certainty either. It is worth taking a bit of time to discuss one of the classic illustrations of uncertainty in mathematics. According to the January 22, 1978 issue of the New York Times, mathematician and logician Kurt Gödel was the “discoverer of the most significant mathematical truth of this [the twentieth] century.”²⁰Close This was Gödel’s famous Incompleteness Theorem. The exact details of this theorem and the proof that demonstrates its truth are not important for our purposes. However, the upshot of this theorem definitely is. With his Incompleteness Theorem, Gödel showed—and it is pretty much universally accepted that he is correct—that some mathematical questions are undecidable: they cannot be shown to be true or shown to be false mathematically. More precisely, he established that “if there are no contradictions in mathematics, then there exist mathematical statements that can neither be proved nor disproved.”²¹Close This means that mathematics is incomplete—there are mathematical claims whose truth or falsity will remain uncertain. As William Byers aptly pointed out, this constitutes “the existence of fundamental limits” to what we can know.²²Close The presence of this sort of limitation in mathematics reveals that there will always be some uncertainty in mathematics. By extension, there will also be uncertainty when it comes to anything that is based on mathematics or formulated in mathematical terms.

Of course, as we have emphasized many times now, lack of certainty does not entail that something is unreliable or untrustworthy. Even though mathematics is not certain, it is still an incredibly reliable source of knowledge and very effective. As philosopher of mathematics Marcus Giaquinto noted, “we cannot be certain of the reliability, regarding finitary consequences, of much mathematics.”²³Close Nonetheless, “a high degree of confidence in the reliability of a large amount of classical mathematics has already been justified” by way of logical proofs.²⁴Close This means that, in spite of disagreements and uncertainty, it is reasonable to have faith in mathematics. We just cannot have epistemic certainty.

In some cases, uncertainty in science is so evident and profound that it is explicitly acknowledged, as in the case of the Uncertainty Principle put forward by physicist Werner Heisenberg:

At the instant of time when the position is determined, that is, at the instant when the photon is scattered by the electron, the electron undergoes a discontinuous change in momentum. This change is the greater the smaller the wavelength of the light employed, i.e., the more exact the determination of the position. At the instant at which the position of the electron is known, its momentum therefore can be known only up to magnitudes which correspond to that discontinuous change; thus, the more precisely the position is determined, the less precisely the momentum is known, and conversely.²⁵Close

Simply put, the idea is that the closer we are to certainty about where a particular electron is, the more uncertain we are about how fast it is moving and in what direction, and vice versa. The Uncertainty Principle entails that there are simply things that we cannot know about the natural world. Importantly, the limitation here is fundamental, just like the Incompleteness Theorem in mathematics. Improved technology, better theories, better techniques, etc. will not make a difference. We simply cannot know both the position and momentum of subatomic particles at the same time.

The Uncertainty Principle reveals an inherent limitation on what we can know, and it has important implications beyond just subatomic particles. Intuitively, if we cannot be sure of where an electron is, then even if we could be certain of how fast it is moving, we cannot be sure of where it will be in the future. Similarly, even if we could be certain of where an electron is at a particular time, if we cannot be sure of how fast it is moving (and in what direction), then we cannot be certain of where it will be in the future or its future velocity. This means that we are stuck with an irreducible uncertainty when it comes to predicting the future. (We discuss uncertainty in predictions more fully in Chapter 13). This sort of inescapable uncertainty further illuminates Carlo Rovelli’s remark in the epigraph that thinking science is about certainty is the “deepest misunderstanding about science.” We will always lack knowledge of some sort, and so we will always face uncertainty.

As we have been at pains to emphasize throughout our discussion, uncertainty in science does not denigrate science or scientists. Science produces some of our best-confirmed knowledge of the natural world. So what is the point? Why should we care that science is inherently uncertain? Rovelli got to the heart of things when he stated that “the core of science is not certainty, it’s continual uncertainty.”²⁶Close Once we appreciate that all science is inherently uncertain, discovering uncertainties in the science of climate change, vaccination, human evolution, genetic testing, or forensic science should not incline us to think that they are inferior to other sciences. It is crucial to remember that being uncertain does not mean being unreliable or unreasonable. Let us now turn our attention in the next chapter to examining the uncertainties present in the explanations provided by our case studies (climate change, vaccination, human evolution, genetic testing, and forensic science) as well as by science in general.