While an individual hypothesis or theory of a scientist can be tested, challenged and even abandoned, the broader methods and scientific fields represent our extensive bodies of knowledge consolidated over time. We discard many of our ideas, hypotheses and even some theories, but we do not abandon our major methods applied across fields or our major fields. Testing the paradigm shift hypothesis with over 750 major discoveries including science's major theories, we uncover that the few abandoned discoveries were mostly theories—not experimental or methodological breakthroughs. These discarded theories often lacked robust evidence, reinforcing the fact that scientific progress fundamentally relies on the right methods to collect robust evidence. So the history of science tells a cumulative and unified story—one driven by our collective method and tool advances. Shifting our attention from isolated hypotheses and select theoretical discoveries to science's major discoveries, methods and fields is a more systematic way to assess the cumulative nature of scientific progress. For they make up the foundation of science and how we do science. So cumulative knowledge exists on a spectrum: from unestablished ideas and hypotheses, then experimental findings and theories (experimental and theoretical discoveries), and ultimately to well-established methods and fields. Within the vast landscape of science's major theoretical, experimental and methodological discoveries, Kuhn narrowed his lens on theories. Within that scope, he zoomed in mainly on physics. And he honed in even further on its early development. He finally zeroed in on a handful of theories often spanning centuries where evolution would be more likely. Yet if we want to draw general insights about scientific progress across the history of science, we cannot be selective but need a comprehensive approach spanning the history of science itself—the aim of this chapter. Kuhn's landmark book The structure of scientific revolutions popularised the idea that science undergoes dramatic paradigm-shifting upheavals—like political revolutions. Yet Kuhn did not establish a universal structure across scientific shifts, and they are generally slow and incremental rather than abrupt and revolutionary. A more accurate title, given Kuhn's select case studies, could have been An account of theory change mainly in early physics . Yet a more fitting description of science would actually be Cumulative scientific progress embodied in scientific methods and fields. We keep reworking the details of our knowledge as we generate new methods and collect new evidence. Evidence and explanations are works in progress—valid until we update them using better methods that provide better evidence and explanations. Science is about creating new tools that enable us to revise our best theories and explanations of the world in light of new evidence. This is the nature of scientific progress. Science and scientific methods thrive on iteration: they are a bootstrapped (correcting) process of constant improvement, refinement and synthesis. Quantum theory and the mechanisms of evolution, statistics and microscopes, chemistry and computer science are all continually refined over time as we come across new methods and challenges. Ultimately, we tend to measure a discovery's significance by its impact—how useful it is in helping us solve problems and better understanding the world. Some discoveries and methods seem timeless, like discovering the electron, the theory of evolution and the sun-centred theory of our solar system, and statistics, microscopes and x-ray methods that will be around in the future. For they form the backbone of entire fields. Others seem less timeless, like discovering the Coronavirus vaccine, partition chromatography and cardiac catheterisation (a technique for inserting a catheter into the heart). Yet each piece contributes to our ever more cumulative body of methods and knowledge that we build on—and account for science.
For many people, studying the evolution of science over history does not seem scientific—after all, we cannot run controlled experiments on the past. But by applying systematic scientific methods to examine science's major discoveries and methods across fields, we can uncover how science advances. Scientific methods, and discoveries they enable, are highly cumulative across fields and time. In fact, no complex scientific methods or tools (like mathematics, lasers and particle accelerators) and no complex scientific fields (like biomedicine, earth sciences and atomic physics) would even be possible if they were not deeply cumulative. We do not discard or disprove major methods, tools or fields. Instead, we abandon preliminary theories that are not grounded in rigorous methods and are often more provisional and speculative than our established methods—that largely lay the foundation to establish our theories and fields. If we instead scan the history of science and focus on a few select theories (like some historians and philosophers of science like Kuhn) or explore a random sample using article citations (like some scientists using big data), we can find what looks like discontinuity between those individual cases. But testing this fundamental question with rich evidence provides a very different, cumulative answer. The idea of grand paradigm shifts, which govern science and overturn central theories with entirely new theories, applies to only about 1% of major theoretical, experimental and methodological breakthroughs. If revolutionary science were pervasive, one may even expect the over 750 biggest discoveries in history—spanning all nobel-prize and major non-nobel breakthroughs—to serve precisely as the strongest evidence of disruptive breaks. Yet we find the opposite: the most groundbreaking of discoveries in history have been extended over time, reinforcing the deeply cumulative progress of science. New and continually improved methods and tools better explain scientific progress than new revolutionary, paradigm-changing ideas and theories that generally result from these improved methods and tools of discovery . Scientific progress is, at its core, fundamentally brought about and grounded on the tools we develop to explore, test and refine our understanding of the world—not just by changes in theories. The common emphasis on theoretical shifts reflects a final output but overlooks the crucial methodological process we take to create, replicate and refine the output. Traditionally, many think discoveries and theories are the heart of science, while methods are a temporary bridge that, once we develop the discoveries and theories, no longer receives our attention. We need to also place methods at the heart of science, with the discoveries and theories we develop using methods seen as the temporary output of science, until we update them with new evidence using our expanding and cumulative methods . Most theories thus remain provisional, evolving as we improve our tools. This explanation of methods-led science offers a more realistic picture of how knowledge advances—a more coherent alternative to the idea of winner-takes-all paradigm shifts. It aligns with actual scientific practice and enables us to better understand and accelerate scientific progress. Science advances cumulatively regardless of how we theorise about scientific change, so the debate among historians and philosophers of science on whether some exceptional theories may evolve through paradigm shifts has little impact on our scientific advances. Our vast and interconnected web of cumulative methods, R knowledge and technology keeps growing regardless of the theoretical debate. For some, the debate on paradigm shifts is, at best, a negligible or non-existent problem. Yet Kuhn would insist that we are not progressing towards the truth, in the philosophical sense—for example from Ptolemy to Copernicus. But we uncover a key insight from the debate: we can in fact speed up the pace of cumulative scientific progress by shifting how we conceive scientific change—to expanding our cumulative scientific toolbox. In the final section of this chapter, we turn now to the evolution of the classic scientific method.
scientific methods that extend our mind The classic scientific method—observing, experimenting and testing hypotheses— has remained unchanged for centuries; so can it be compatible with the deeply cumulative nature of science and its ever-evolving methods? Does it accurately capture the way we gather knowledge—and then revise our knowledge as we get our hands on new evidence? Or do we need to rethink and expand this classic method into a broader, overarching method of science that better reflects how science actually works? The classic scientific method is defined in science textbooks and dictionaries as 'the collection of data through observation and experiment, and the formulation and testing of hypotheses' . Many science textbooks present the scientific method to students as a simple sequence of steps: observe, experiment and test hypotheses. A study of major science institutions like the National Science Foundation (NSF) and National Institutes of Health (NIH) also revealed that they primarily endorse this hypothesis-driven method rather than exploratory methods that lack predefined hypotheses. So this approach, as a unifying method of science, is embedded in science dictionaries, textbooks and institutions—often stating that we follow, and should follow, the method. It is commonly traced back at least to Francis Bacon who popularised the concept in his book Novum Organum in 1620. Bacon emphasised that what we know comes from evidence and experimenting (in his words, 'twisting the lion's tail' and observing what happens); and is the only way to actually do science. His book not only laid the foundation for philosophy of science but also fundamentally shaped how generations of scientists conceive the practice of science. But science has advanced far beyond its early history, so does this traditional view still hold? Before hypothesising about what science's general method is and what it should be, we need to first take a step back and examine the evidence on how science is actually conducted. Surprisingly, the classic scientific method has not been systematically analysed using scientific methods themselves, as many assume that it cannot be subjected to scientific study. But by studying science's major discoveries, we can tackle the basic question: to what degree is the classic scientific method actually applied in making science's groundbreaking research? Think of Einstein's theory of special relativity that reshaped physics in the 20th century and how we understand space and time. Darwin and Wallace's theory of evolution by natural selection transformed biology and how we comprehend the historical origins of our species. Franklin, Crick and Watson's discovery of the double-helix structure of DNA redefined genetics and how we conceive the way genetic information of living organisms is stored, copied and passed along. These scientists fundamentally changed the way we view the world, but they themselves did not always directly carry out experiments to uncover these path-breaking discoveries. Examining science's major discoveries, we uncover an unexpected finding: 25% of breakthroughs since 1900 did not apply the traditional scientific method (all three features)—with 6% of discoveries made without observation, 23% without experimentation and 17% without testing a hypothesis. Expanding our lens and analysing science's over 750 major discoveries over history, we find that the traditional scientific method (all three features) is applied in making 71% of discoveries (with individual shares seen in Figure 3.3). Some hypotheses are tested through systematic experiments, while others rely solely on observation—such as an astronomical observation or in observational studies. So science does not always fit the textbook definition, with hundreds of groundbreaking discoveries not following the classic scientific method. The evidence thus challenges this long-standing concept of science. What about differences across fields? Surprisingly, the classic scientific method was not applied in making about half of all nobel-prize discoveries in astronomy, economics and social sciences. Why? Experiments are not always possible—astronomers cannot manipulate stars in a lab and economists cannot run controlled experiments on entire economies. Some discoveries are made through open-ended exploration, without testing a pre-established hypothesis, while others are more theoretical. Even in physics, about a quarter of breakthroughs broke from tradition. This reveals a key insight: the traditional scientific method does not capture the full reality of discoveries and, more importantly, it overlooks the fact that science's major discoveries depend on applying sophisticated methods (like statistics and randomisation techniques) or R tools (like centrifuges and advanced computers) (Figure 3.4). Science is about getting our hands on the right tools to break new ground. When we evaluate science's major discoveries, what is the fundamental method we apply to be able to do science and spark discoveries? We identify a common feature we can reduce the method of science to: science's major discoveries have relied on using sophisticated methods and instruments . These external resources—from lasers to chromatography—extend our mind and senses, allowing us to see, measure and analyse the otherwise unobservable that makes up most of science today. Unlike observing, hypothesising and experimenting—largely internal cognitive abilities— scientific tools are material artifacts that can be shared, refined and built on by others (Figure 3.4). Applying sophisticated methods or tools is a necessary condition for discovery. Without them, discovery and scientific progress is not possible—this reveals a universal principle of scientific methodology. The sophisticated scientific method is actually more unique to science than the classic method; after all, the most used scientific methods and tools—like particle accelerators, electrophoresis methods and x-ray crystallography—we largely only use in science. But we often make observations, test hypotheses and experiment in business, industry, public policy and even in our everyday lives and they are not just prototypical or distinct to science. Recognising the enormous importance of complex tools adds an essential element to understanding science and how science has evolved. In its early origins, science was often grounded only on directly observing, hypothesising or experimenting—while today, these activities are only possible for making discoveries by applying and refining complex tools (a topic we explore later in Chapter 7). The classic scientific method —or one or two of its three features—was more likely to be applied in its traditional form, without complex scientific tools, when early scholars like Bacon described it in 1620. At the time, the first two major tools that shaped 17 th-century science, the microscope and telescope, were recently invented, and tools did not yet dominate science in the same way. Bacon's vision of science aligned with the prevailing understanding of science and was limited by the tools of his time. But since then, sophisticated scientific methods have transformed how we observe, experiment, test hypotheses and solve problems—in much more diverse, complex and efficient ways than ever before. Just as science itself has evolved and expanded, so too should the classic scientific method.Its broad and general description is better understood as a basic method of reasoning used for human activities (non-scientific and scientific alike). Let us look at some of the nobel-prize discoverers who did not directly apply or generally could not apply the classic scientific method in making their groundbreaking discovery. Einstein did not himself run traditional experiments when he developed the law of the photoelectric effect in 1905, nor did Franklin, Crick and Watson when they uncovered the double-helix structure of DNA in 1953 using observational images generated by Franklin. The British Roger Penrose did not himself make direct observations when he formulated the mathematical proof for black holes in 1965, nor did the Russian-Belgian Ilya Prigogine when he created the theory of dissipative structures in thermodynamics in 1969. The Danish Niels Jerne did not directly test a hypothesis when he developed the natural selection theory of antibody formation in 1955, nor did the Canadian James Peebles when he created the theoretical framework of physical cosmology in 1965. If we were to abide by the classic scientific method, Copernicus, Einstein, Franklin, Crick and Watson and many others would not be seen as applying it, as they did not directly conduct experiments to trigger their seminal breakthroughs. Yet these scientists became iconic figures of science. Many nobel-prize discoveries have been awarded for breakthroughs that lacked theoretical underpinning, from radioactivity and x-rays to viruses and scientific tools. While individual scientists and breakthroughs at times bypass steps of the traditional scientific method, the method can be seen as often eventually applied at the collective level by the scientific community over time. Scientific progress, in this sense, is distributed across scientists over generations, making it a cumulative endeavour where one picks up where another left off. Yet the sophisticated scientific method is in fact implemented across science's major discoveries. By reframing the scientific method not as a rigid, linear sequence of steps a scientist applies but as a more flexible, tooldriven process that scientists actually apply, we gain a critical advantage. We shift our focus from formulating hypotheses to identifying the most effective tools and methods to solve complex problems—with or without a pre-established hypothesis, as many of the greatest discoveries highlight. So we need to reframe the (classic) scientific method from mainly inside our heads—observing, hypothesising and experimenting—to the (sophisticated) scientific method mainly in the world outside that we explore and measure with our cutting-edge scientific methods and tools. For tools extend the very limits of our mind, generate completely new kinds of data and uncover insights about the world far beyond what we could have imagined. Scientific tools, like scientists themselves, come in many shapes, sizes and levels of sophistication. To do science, we do not just observe; we combine mathematics with precise measurement instruments, merge statistics and AI with controlled experiments, link x-ray crystallography, spectrometers and particle detectors to systematic observation, and make hundreds of other powerful combinations. Think of the incredible diversity of methods used in immunology, oceanography, neuroscience and astrophysics, or chemistry, agronomy and behavioural economics. Our tools make it possible, for most phenomena in science, to observe, experiment and test hypotheses in the first place and do exploratory research—and in new and innovative ways otherwise out of our reach. The sophisticated scientific method integrates these features into our central methods and tools, creating an adaptable system of discovery (Figure 3.5). Even replication, a main feature of science, is deeply tied to sophisticated methods (not the classic method). Using advanced tools like x-ray devices and statistical methods, researchers can replicate discoveries (while simply observing, experimenting and testing hypotheses is too broad and too susceptible to each researcher using them differently). But with sophisticated methods, science becomes more accurate and reliable and less prone to human error, while enabling us to much better evaluate the quality of research. Ultimately, with the classic scientific method, we would not be able to label many of our greatest scientific discoveries as scientific—despite their profound impact on our lives. This traditional view, seen as a golden principle connecting the scientific community together, can be misunderstood as universal. It is an idealisation that can at times be confusing and misleading, especially for students and less-experienced researchers, when learning about science and realising that it does not always apply or focusing their attention at times in the wrong direction. In fact, adhering to it as a guiding principle can stifle innovation and constrain us from developing new ideas and breakthroughs. How we do science and spark major breakthroughs thrives on our diverse methods and tools, so we can best view the method of science as leveraging our sophisticated toolbox. We need to reform the classic scientific method, integrating and redefining it as the sophisticated scientific method. Since methods and instruments— from mathematics to microscopes—extend our mind and senses and are essential for doing science today, we lay out a definition that better reflects actual scientific practice: Scientific methodology is the use of sophisticated methods and tools that enable us to observe, experiment, test hypotheses, solve problems and do exploratory research. Sophisticated methods are general-purpose, meaning we can apply them to different questions and domains. This definition offers a more accurate understanding of scientific methodology. It shifts our attention to upgrading our sophisticated toolbox—the very engine that enables us to push science forward. Ultimately, the best path to discovery is not the classic scientific method but its extension: the sophisticated scientific method. At this point, we have gotten through enough evidence to return to the related but broader question: what is science? While science is generally defined as the study of the 'world through observation, experimentation, and the testing of theories' (Oxford English Dictionary), this view misses something crucial. At its core, science is powered by methods and tools we create—from telescopes to statistical simulations— that enable observation, experimentation and testing theories. Without them, science today is not possible; so science needs a better definition based on actual evidence: Science is the study of the natural and social world using methods and tools to observe, experiment, solve problems and develop theories—in order to describe, explain and predict the world. As a final thought, to develop new tools and ideas, we use not only observation and measurement but also imagination and abstraction that go beyond the periphery of the measurable scientific method. Take Emmanuelle Charpentier at Umeå University and Jennifer Doudna at Berkeley for example, who imagined and devised a unique new genome-editing technique CRISPR. Their method functions like genetic scissors that precisely cut a DNA molecule at a specific location. This method has transformed the life sciences and our understanding of how to change the DNA of animals, plants and even humans. Take Allan Cormack at Tufts University and Godfrey Hounsfield at EMI Laboratories, who conceived computer-assisted tomography (CT scans) by imagining the different pieces and devising the necessary computing system. How a CT scan works is that clusters of x-ray beams pass through our body from multiple angles, generating detailed internal images of our body with computer calculations. The instrument marks a vital advance in the medical world and some of us who have had injuries have taken a CT scan. Tool discoveries rely on imagination— from Wilson conceiving the first particle detector that visualised particle tracks and Ruska inventing the first electron microscope, to Gilbert and Sanger designing DNA sequencing.
It is tempting to think of science as revolutionary, as sudden paradigm shifts make for appealing and fascinating stories. But science—its scientific methods and the discoveries and fields they enable—is best described by (cumulative) evolution rather than (paradigm-changing) revolution. Our methods and fields would not even be conceivable if they were not deeply cumulative. New and expanded tools better explain scientific advances than new paradigm-changing theories that arise from those expanded tools. By recognising this, we can accelerate the pace of cumulative progress by reconceptualising scientific change: rather than thinking of elusive revolutions, we need to systematically expand our cumulative toolbox as the engine of science and discovery. Ultimately, because science evolves does not mean that it is not cumulative—it is precisely what makes science cumulative, by building on and refining our collective methods and knowledge of the world. By standing on the shoulders and cumulative methods of giants, we keep seeing further and developing new discoveries, life-saving medicines and transformative technologies. To reflect how science actually works, we need to also extend the classic scientific method (which we do not always apply) with the sophisticated scientific method (which we do always apply). So far, we have explored the two most influential and cited explanations of science—and then sketched out the method maker's new perspective. Examining science's major discoveries, we reveal that science and discovery actually follow a logical pattern, unlike Karl Popper's explanation of science that rejected this possibility (Chapter 2). We also find that the history of science is highly cumulative, unlike Thomas Kuhn's explanation of science that is shaped by non-cumulative paradigm shifts (Chapter 3). Both the logical and cumulative nature of science are deeply embedded in the ever more powerful tools we use to spark new advances.