We cannot design an experiment to measure what our tools cannot detect. In science, the tools we use largely determine how and what we can study—and the very questions we can ask using them. Powerful new tools—from super-resolution microscopes to the gene-editing method CRISPR—unlock entirely new types of questions, problems and discoveries that have not ever been imagined before. Here we take a step back and look at the bigger picture, pulling together evidence from earlier chapters and expanding on the insights with new perspectives. This enables us to further develop the new methods-driven discovery theory. This general theory of science is backed by five different types of evidence from diverse domains: Statistical analysis of science's major discoveries reveals that behind our major leaps in science there is a powerful new method or tool that made them possible. Evolutionary biology explains our evolved method-making capacity. Cognitive science highlights how we overcome the limits of our mind and senses with powerful tools that outthink, outsee and outimagine our own minds. Scientific practice reveals how our available tools largely define how and what we explore and understand. Philosophy of science illustrates how each method has built-in assumptions, and filters how we understand and interpret the world. Together, these five lenses create a powerful new understanding of how science actually works. While the search for grand explanations of how science progresses has largely been abandoned since Thomas Kuhn's concept of paradigm shifts, we expand this new general theory of scientific progress here. Integrating the five different explanations, this unifying theory offers a roadmap for discovery—and a foundation for the fields of Science of Science and Methodology of Science. Ultimately, shifting our attention to designing new discovery tools would trigger a tool revolution in science: where we can vastly accelerate new breakthroughs and where labs across science begin competing to design discovery tools, not just to use them.
Discovering vaccines, anaesthesia and the structure of DNA transformed medicine and how we understand human biology. Discovering the universe is expanding reshaped our understanding of our minute place in the world. Developing quantum 242 T HE ENGINE OF SCIENTIFIC DISCOVERY mechanics and relativity theory redefined how we comprehend physical reality and explain the world on both the smallest and largest scales. Yet scientists have not yet solved the puzzle and there is still no general theory of how we trigger such extraordinary breakthroughs that is grounded across different disciplines. The current explanations of science and discovery—whether from scientometricians, historians, psychologists, computer scientists or philosophers—describe individual factors that affect some but not all discoveries. (We covered these in Chapters 1–6). Yet the best way to know if our evidence or a theory in science is reliable is commonly to examine if it is backed by independent methods from different fields. Take the theory of evolution: it is well established with rich evidence using methods from molecular biology, palaeontology, primatology, biogeography, cognitive science and archaeology—all telling the same story. The more diverse the methods that establish a finding, the stronger the finding is—and a general theory emerges. So, how does the new methods-driven discovery theory developed here hold up? To further test its reliability, we explore how well the theory is grounded in multiple sources of evidence: statistical data on science's major discoveries, observational methods from biology and cognitive science, analysis of actual lab practices and research methods, and conceptual analysis. Strikingly, we find a single principle running through all these lines of evidence: our powerful tools are the driving force behind science and discovery. What emerges is an integrated theory of how we develop new ideas, breakthroughs and fields through new tools we build—tools that make new ways to think, experiment, theorise and imagine possible. Here, we lay out these five different explanations that come from diverse domains across science. But together, they tell the same method-driven story, forming a unified theory of how discoveries emerge: • Before-and-after (statistical) explanation: Science's major discoveries and fields consistently follow the creation of new methods and tools that make the discovery or field possible—and provide the entirely new perspective (Chapters 1–5). • Evolutionary explanation: Our minds evolved with methodological abilities over our species' history to observe, solve problems, experiment and create ever more complex methods—used to gain our vast knowledge about the world. Being cumulative method-makers is what makes us human. Remarkably, it has been both the driver of our survival but also scientific progress. Our methodmaking species is a product of evolution, and science is its most powerful outcome (Chapters 7–8). • Cognitive constraint explanation: Our mind and senses face many constraints, biases and blind spots—from not being able to see atoms, to overlooking patterns in data. But science progresses when we invent tools that bypass these flaws and tackle these blind spots. Tools enable us to outsmart—outsee, outthink and outimagine—our own minds (Chapters 7–8). • Scientific practice explanation: Every experiment, study or discovery is largely defined by the tool it relies on. The method we use—whether an electron microscope or computer simulation—does not just enable but largely determines what and how we can study. It fundamentally shapes what questions and hypotheses T we can ask, what experiments we can design, run and analyse, and how we measure and collect data. It largely sets which conclusions we can draw and how others can even confirm the evidence using the method. • Philosophical explanation: Every method comes with built-in assumptions: how we define phenomena (ontology), how we measure them (measurement system), what causes them (causation), whether we study the process, outcome, property or system (level of evaluation) and what kind of explanations and theories we count as valid (epistemology). Each method is a lens that fundamentally frames how we view and interpret reality (Figure 9.1). A critical insight emerges here: there is no alternative explanation for how we directly develop major discoveries without leveraging the needed tools that make them possible and provide the new perspective. Here, we will tackle three questions: what exactly do these five explanations reveal about science and discoveries? How do they combine to back the new method-to-discovery view? And most importantly, how can we spark a method revolution—one that moves science faster and more strategically? Before-and-after (statistical) explanation Methods drive science's new discoveries and fields E = MC2Scientific practice explanation Methods largely define how and what we can study and experiment Evolutionary explanation Methods are developed using our evolved abilities to observe, solve problems and experiment Philosophical explanation Methods shape how we define, measure, evaluate, assess causation and so our knowledge and understand ing of reality cause → effect Cognitive constraint explanation Methods enable us to tackle the limits of our mind and senses and expand our lens to the world methods Figure 9.1 The new methods-driven discovery theory: How new methods enable us to develop, conduct, improve and advance science
Scientific progress thrives not just when we ask better questions but when we innovate new methods that spark entirely new kinds of questions and answers—and challenge existing knowledge. The 'before' stage is marked by unsolved mysteries and 244 T HE ENGINE OF SCIENTIFIC DISCOVERY existing tools that are not powerful enough to solve them. The 'after' stage marks when we devise the right tools that enable uncovering answers to those mysteries and trigger a new perspective. Take particle accelerators. Before we invented them, we did not know many subatomic particles that make up the world even existed. But after we created these extraordinary instruments, we could study their fascinating properties and behaviour. They did not just expand our understanding of physics, but they also helped fuel cutting-edge industrial technologies. Before electron microscopes emerged, we could not see the ultra-fine structural details of cells, but after we built this ultra-high-resolution tool a hidden universe opened up—from the complex structure of neurons to the inner machinery of bacteria. Examining science's over 750 major discoveries using statistical data, we find this striking pattern across science: we made them directly after creating the powerful, discovery-triggering method or tool—and typically soon after creating them (as we found in Chapter 1). What is particularly remarkable is that many new tools do not just solve one puzzle, they spur clusters of breakthroughs unforeseeable by the original toolmakers themselves. This ripple effect often stretches across different fields beyond the ones in which the tools were designed. While scientists generally apply available tools, frontier science that produces groundbreaking research applies new tool innovations (Figure 1.1 and 1.2). In fact, new tools not only spark new discoveries but also launch entirely new fields: from the electron microscope opening modern cell biology and particle accelerators and detectors giving birth to high-energy physics (Chapter 5). But how did we gain the extraordinary capacity to invent these striking tools?
What made us outstanding method-makers and tool-makers, able to reach far beyond the limits of our own eyes, hands and imagination? Why is it that only humans can build tools that reveal atoms, simulate galaxies and edit our genes? Our ability to do science did not just appear recently but evolved over millions of years. Our mind's methodological abilities evolved by gradually improving our ability to survive and adapt over time: vision needed to make observations and scan the world, memory to store what we observe, imagination to extrapolate beyond the present, and reasoning skills to solve problems, experiment and develop ever more effective methods. Only by using these evolved capacities can we perceive, gain knowledge and make discoveries. Other animals share parts of this toolbox too. But only we turned these abilities into cumulative method-making. Throughout history, our earlier ancestors leveraged our evolved methodological mind to create methods and tools: the stone tools of the Palaeolithic, the planting strategies and crop experiments with seeds and soil that sparked the agricultural revolution, and ultimately the fascinating tools that extended our perception and mind and sparked a snowball effect—triggering the scientific revolution and then industrial and digital revolutions. Each new major method in history marked a method revolution (as we saw in Chapter 7). Our evolving method-making capacity transformed us: those early ancestors who could invent more and more effective hunting strategies, better track animal behaviour or reduce illness with medicinal plants had a survival advantage. From T constructing pyramids by experimenting and applying geometry and engineering, to creating systems of mathematics and astronomy, it was method-makers who engineered the world around us and changed the game. Today's scientists build on these method-making skills, but with far more powerful tools (Chapters 7–8). Let us take a step back. Our mind—like everything in nature—is the result of evolution. It developed, within our environmental and cultural niche, responding to challenges we face. Our gradually evolving method-making capacity enabled us to develop ever more useful tools that increase our survival and vastly expand beyond what evolution alone gave us. Take our vision: it is our and other primates' most used sensory input to make sense of the world. That is why we make so many vision-enhancing tools to tackle the natural limits of our eyesight: microscopes, telescopes, x-ray methods, spectroscopes, particle detectors, etc. (Chapter 2). These vision-extending tools do not just set how much further we can see—they enable us to imagine and understand the world much further than with our unaided mind. In science, vision is still our most relied-on form of evidence to explain the world around us (Chapter 6). In short: our method-making mind is the deep result of evolution, and every tool we use today is an extension of our ancient mind. So we can trace the origins of science to contemporary science by tracking how we evolved to perceive, measure, theorise and invent more complex methods. But our mind faces limits imposed by nature and evolution. These help set the bounds within which we can do science—and generate new discoveries when we learn not to accept those limits but tackle our mind's constraints.
Our mind and senses—as extraordinary as they are—come with many built-in flaws, blind spots and biases to what we can know about the world. From our eyes that cannot see ultraviolet light to our ears that cannot hear ultrasound, the tools we develop using our mind enable us to overcome these limits. At its heart, this is how science moves forward. Our minds and senses evolved in a narrow window of experience and reality—the part that mattered for survival: things close to us, slow enough to track and visible to the naked eye. But most of the universe does not fall into our species' small sensory window. Most phenomena in the world fall below or lie beyond what we can observe. We are increasingly constrained the further we move away from these evolutionary conditions—from the earth's surface, from our ecological niche, from our unique context. Our mind is not able, without tools, to access what is beyond our senses: minuscule atoms and photons at the quantum level, vast magnetic fields and gravitation, extremely fast phenomena like the speed of light, extremely distant planetary systems and the earth's core, intangible global economic markets and political systems, and the effects of climate change over decades. But with the right tools, we can suddenly access and pull them into view. Today, most of what we study in science we cannot directly see with our bare eyes— from subatomic particles, to proteins and neural signals, to complex financial systems. It is the tools we create using our flexible mind that explain most of the expansion of science by uncovering what would otherwise lie far beyond our brain's natural reach . 246 T HE ENGINE OF SCIENTIFIC DISCOVERY So how we understand most of reality is determined by our tools of discovery. We can—with our naked eye—only detect a very limited range of the electromagnetic spectrum. But that narrow window widened suddenly once we constructed infrared sensors, x-ray devices and radio telescopes. In medicine, a powerful toolbox made much of the field possible and continues transforming it: optical microscopes, centrifuges, x-ray devices, chromatography, gel electrophoresis, electron microscopes and hypodermic syringes—as we uncover when examining medicine's major discoveries. Computer-generated images inside the human body were simply unimaginable with our bare senses. With magnetic resonance imaging (MRI), electrocardiograms (ECG), computer-assisted tomography (CT scans) and cardiac catheters we can see inside a patient's body: scan our brain, track heart rhythms, map blood flow and catch disease—without a single incision. Without such powerful tools, our unaided mind could not diagnose and treat most life-threatening diseases. These tools have not just vastly improved healthcare, they have added quality and years to our lives. Computers are vastly speeding up science and changing how we do science. Today's computing power, automation and machine learning tools enable scientists to run thousands of experiments, analyse vast data and make complex predictions at speeds and scales our brains could never imagine without them. Imagine trying to find patterns across millions of patient records by hand. Or simulate the earth's changing climate a hundred years from now. Or track a fast-mutating virus around the globe in real time. Without the massive boost of computer-assisted science, these tasks would be impossible and unthinkable. When we mapped the entire human genome, it was only possible because we combined computing systems with automated methods to process vast databases of DNA. That combination also powers automated DNA sequencing, real-time weather forecasting and measuring vital signs in hospitals. We think in science more with our tools than with our limited minds. Powerful tools do not just make science more efficient, but give us entirely new eyes and open up new worlds—the microscopic world, the statistical domain, the quantum sphere and the galactic universe. To understand how tools work and shape discovery, we break scientific methods and instruments into three categories here: • Direct tools: these are analogous (similar) to our mind and senses but enhance them—low-power microscopes, telescopes and arithmetic improve what we can already naturally do—see, measure and calculate. • Translational tools: these are partly analogous to our mind and senses but stretch them further—x-ray detectors, infrared telescopes and ultraviolet spectrometers convert invisible wavelengths and signals further beyond our sensory range into something we can perceive and detect. • More abstract tools: these are not analogous to our mind and senses but go vastly beyond them—quantum computing, network analysis, blinded randomised controlled trials and particle colliders do not resemble what we can intuitively grasp. They do not have a clear reference point with our mind or senses, but operate far from human intuition —on a level completely unfamiliar to our T everyday experience. Commonly, we need visual models just to interpret what the tools are doing. How much our tools mirror human cognition is on a spectrum: from more direct, then partially direct, to more abstract tools—from sensory-enhancing tools, then signal-translating tools, to more abstract, conceptual and modelling tools. Classifying tools by cognitive distance from our evolved perception reveals an important insight. The category our tool falls into shapes how we understand—whether we rely more on our external tools or our internal mind. While all three types of tools commonly produce very reliable evidence, we often feel more confident in what we see. So consensus often forms more easily with evidence from more direct tools. To conceive and test more indirect tools often requires a radically different approach—and is often more difficult. With indirect tools, we often rely more on computational models, statistical inference and AI algorithms—and that can make consensus at times harder to reach. As we develop more indirect and abstract tools—like quantum simulators that model materials before they exist or AI systems that run thousands of experiments—we may find ourselves, in part, depending less on direct observation and more on mediated, tool-led understanding. That shift raises important questions: how do we increase confidence in tools that no one fully grasps intuitively? What does it mean for such tools—in the domains they are used—to become the principal driver of our imagination and understanding? How do we keep science grounded in our sensory world when part of science happens far outside of that world? Think of phenomena with the least consensus like the evolution of consciousness, multiple universes and the historical origin of life. It is because we do not yet have the needed tools to collect more direct, reliable evidence—a topic we return to on the limits of science (Chapter 10). In short: we come into the world with a narrow range of what we can see, sense and understand—but with each new tool we push those boundaries. Computers, radio telescopes and statistical methods do not just extend but multiply the powers of our mind beyond our imagination. Most of scientific practice is distinguished by developing and leveraging tools that enable us to study parts of the world that would otherwise lie beyond our cognitive reach. In many fields today—from neuroscience, genomics and climate science, to astrophysics, economics and AI—the next major leaps will not come from thinking harder, but from designing tools that think and imagine for us. The deeper we reach beyond our brain's borders, the more our future discoveries depend not on our mind—but on our tools. Science is, in many ways, about outsourcing experimentation, pattern recognition and imagination to advanced microscopes, x-ray machines and supercomputers.