Methods and tools Here we unpack the five factors that both hold us back and drive us forward as we try to break through the frontier of science. In most areas of science today, a powerful statistical method, cutting-edge x-ray scanner or super-resolution microscope can generally reveal more than a million eyes or minds puzzling over the same question. Our best methods and tools do not just push the boundaries of discovery but define where we presently draw those boundaries. They determine our experiments and data. With each powerful new tool we have invented and re-invented, we have drawn and re-drawn the map of the unknown. We have built ultrasound devices that detect soundwaves otherwise out of our reach, developed massive datasets that work as external memories, created computer and machine learning technology that examine those data to find hidden patterns. Amplifying our toolbox greatly amplifies how we can grasp reality—to observe the previously unobserved and explain the previously unexplained. Our toolbox has opened our scope to a world of microbes, synapses in the brain, elementary particles, galaxies billions of light-years away and genetic mutations that cause disease. Progress in science is largely about upgrading our instruments so we can explore parts of reality that evolution never equipped us to see. But every tool has its limits. Each MRI scanner, supercomputer or statistical method has built-in boundaries that set what questions we can study and what findings we can make (Chapter 9). Each tool defines the scale, precision and kind of detail we can uncover—leaving deeper layers of reality hidden until new instruments reveal them. Our tools struggle to make sense of extremely complex systems where the whole is more than the sum of its parts. They fall short when trying to model emergent properties that do not easily reduce to their parts—think of weather systems with chaotic, hard-to-predict dynamics, or the intricate signalling networks inside living cells. Despite powerful tools like x-ray devices or high-energy particle colliders, there are layers of structure and interaction we cannot yet measure or image. Our best computational methods including supercomputers, quantum computers and simulators also hit real barriers to how deeply we can study, model and theorise about the world. These limitations of our tools also constrain the very physical laws we are able to discover—and those we are not. They restrict the very theories we can conceive— because theories ultimately rely, directly or indirectly, on the unexplained findings our tools reveal (Chapter 6). We cannot properly conceive or confirm theories if we lack the methods to measure the variables or compute predictions. In neuroscience for example, we cannot fully explain consciousness because our current tools face enormous measurement constraints. Even our best brain-imaging tools cannot yet map the brain's activity with enough resolution or completeness. Our mind Our mind is extraordinary, but it faces many built-in limits, blind spots and biases that shape how we see the world, how we reason and how we do science. As the physicist 268 T HE ENGINE OF SCIENTIFIC DISCOVERY Stephen Hawking put it, 'We are just an advanced breed of monkeys on a minor planet of a very average star. But we can understand the universe. That makes us something very special' . How we do science relies on our evolved methodological abilities to solve problems, test hypotheses and spot causes, but also abilities to remember details, use language, learn from one another, teach, abstract and imagine possibilities. Because our mind is the unpredictable product of evolution, we are tuned to a narrow window of the world: making sense of those parts of reality that mattered for our survival and our senses can directly pick up (Chapters 7–9). Even our basic concepts of space and time partly reflect the limits of human perception and the way our mind simplifies a complex world to make it manageable. These built-in blind spots shape how we model the world and even the kinds of answers that seem sensible to us. Our evolutionary history did not prepare us to see, grasp and mentally model the size and nature of the universe, the speed of light, climate systems, global financial markets, the tangled networks of neurons that produce consciousness, and huge timescales. Only in recent history have we learned to make discoveries and stretch our understanding beyond what our eyes can see or our intuition can handle: by inventing powerful new tools and methods. With our evolved mind, we gather information through our vision and senses of temperature, weight, time and speed. But our remarkable tools push far beyond those limits of human eyes, ears, memory and imagination. Science is mainly about identifying the blind spots in our perception and reasoning by devising new methods that enable breaking through those current limits. Each time we invent a more efficient detector, a faster algorithm or a more precise sensor, we expand the scope of what we can discover. This is why science is deeply human; it is a story of recognising our limits and refusing to accept them—with better tools. Our place and time How does the time and place we live in affect what we can discover? We live in a small corner of the universe, on earth, and even with our best space telescopes and spectrographs, much of the universe is out of our reach. We also live in one moment in history, in our current century, and we cannot run experiments or collect direct evidence on the distant past. Instead, we must piece together parts of a puzzle and extrapolate using the evidence we can get our hands on—from using space telescopes, to extracting samples of ice cores, to studying ancient fossils with radiocarbon dating. We cannot directly observe the Big Bang—we infer it from cosmic background radiation and galaxies moving away from us. We rely on indirect evidence and models not only to study the birth of the universe but also the birth and evolution of life on earth. We do the same to predict how antibiotic resistance may spread or how our solar system may evolve in the future. Exploring the deep past or forecasting the far future always hits the limits of evidence. Even trying to explore other nearby planets or moons in our solar system faces enormous instrumental challenges that limit what we can know—let alone the hope to collect rock samples from exoplanets one day. T After all, every tool we use, experiment we run, observation we make and measurement we collect happens in a specific place and time. We test drugs in specific medical clinics, find fossils in particular geological sites, experiment on animals within local ecosystems, and observe gravitational waves with detectors from earth. Where we are in space and time constrains the evidence we can collect and the discoveries we can make about most of the universe and most of the past. Discovery is tied to the evidence we can generate—here and now—with the current tools we have. When scientists develop theories about how life first formed, what dark matter is, how the moon was created or how conscious experience evolved, they are going beyond our niche of the world accessible to us with our tools. They hit the current edges of scientific knowledge. We generally run into our current barriers to what we can discover, the further we move away from our human domain: the farther we look back in time, forward into the future, down into subatomic particles or out to vast cosmic structures (Figure 10.2). Why? Because our evidence gets scarcer, more indirect and less reliable as we move away from our familiar human scale. When we study the deeper past—in geology, cosmology, palaeontology and archaeology—we face constraints at the borders of science. We cannot experiment on dinosaurs or observe our early solar system firsthand. Instead, we build models, extrapolate, cross-check evidence from different fields and make our best inferences. Even so, we can often build remarkably detailed explanations about the distant past. We reach the current boundaries of science when it is no longer possible to gather new evidence with today's best tools—or reliably extrapolate from what we do know. This is why it is so important to keep designing more sensitive detectors, faster DNA sequencers, and more advanced computational models that can better spot patterns we would otherwise miss. Computer models are key in helping us test and simulate scenarios we cannot recreate in the lab or see firsthand, extending our reach beyond the here and now. Yet when we study phenomena with greater complexity, or with more depth at the micro or macro level, the present limits of science are generally shaped by human limits—our method-making and cognitive limits. Yet when we study further back or forward in time, the present borders of science are also shaped by it getting more difficult to access data. Ultimately, where and when we gather data with our methods—our present vantage point on earth—shapes the parameters and content of our evidence and the discoveries we can reveal. Yet scanning science's major discoveries we find that most breakthroughs come from what we can currently study—offering more reliable evidence: from chemical elements, matter, physical forces and the climate, to molecules, our genes, the brain, disease and immunity, to our social and economic systems. Human perspective Our human perspective on the world—as the biological animals we are—also shapes what questions and goals we pursue, as we use our methods and our mind to do science. Just by being human and members of our species, do we naturally direct more attention to problems that matter to us than others? Indeed, we focus most on the 270 T HE ENGINE OF SCIENTIFIC DISCOVERY parts of the world that influence our survival and interests—the niche where our species evolved and lives today. Nearly all scientists study aspects of the world that affect our needs and wants: human health, human technology, human society, human behaviour and the everyday challenges we face. We do not just choose certain questions—we also build tools, run experiments and design theories that fit our human way of thinking and sensing. Our instruments extend and reflect the evolved limits of our senses and mind—like building x-ray devices and radar telescopes to see through soft tissue and through clouds, rather than something we did not evolve to detect at all. They are essentially enhanced 'vision' . We are constrained not just by our sensory limits but by how our cognitive architecture makes sense of information—and even the kinds of answers that seem reasonable to us (Chapter 9). Discoveries we make are shaped not only by our human needs but also by our human lens and choices. Even advanced AI systems are trained on human-labelled data—generated by humans—reflecting our choices of what is important to us and what is not. So the boundaries of science are not just out there in the universe—they are also inside us and our tools. The world's largest science funding agencies—like the European Commission and National Science Foundation—generally require scientists to explain how their research will help society before they fund it. Most science funding worldwide is spent on studying human beings. Over half of all public research funding in the United States—at 52%—is allocated to medicine, health, life sciences and psychology. The remaining share goes to all other fields—from engineering and physical sciences to social sciences, environmental science and computer science—that often aim to improve human life in some way. So why do tens of thousands of scientists focus on explaining and predicting illnesses, pandemics, mental health, population dynamics, financial markets, our behaviour and the weather but only a few scientists study dark energy of the universe, deep-sea creatures and insect minds? Because some questions have a direct impact on our lives, while others seem far removed—unless the questions can eventually help us. We humans are always our own point of reference. When we research in biology, medicine, psychology, economics or social science, it is typically to make our lives longer, healthier or easier—not to improve life for all plants and animals on earth, unless there is a benefit for us. Ultimately, we explore the world through our human lens—our anthropocentric filter that shapes what discoveries we pursue and value most. Yet this is not necessarily a drawback—our human focus has driven extraordinary advances in medicine, technology and understanding of our planet and society. Social context The size and diversity of our scientific community—along with the resources it has available—foster research at the frontier. A larger, more varied scientific community means more chances to develop more tools and accumulate more knowledge. No single person can invent on their own a complex method (from advanced statistics to randomised controlled trials), develop an entire field (from molecular biology to T nuclear physics) or create a complex theory (from quantum theory to a theory of the origins of life). Strategic labs can have vast spillover effects on scientific progress. Think of the Cavendish Lab and Bell Labs—the two most productive methods labs and incubators in history. Inventing an array of powerful tools, these two methods labs sparked an explosion of discoveries—even though they were not founded to innovate tools (Chapter 6). We foster cutting-edge research through economies of scale, reward systems, science policy and targeted funding. Yet when academic careers often depend on short-term publishing metrics, scientists may avoid riskier, long-horizon research that can lead to the biggest breakthroughs. Some reward systems can help motivate scientists. Giving priority to the first person or team to publish new findings or develop a new method creates competition. This winner-takes-all system incentivises innovation. The race to sequence the human genome was for example accelerated by competition between public and private teams. Public institutions also help plan and finance frontier research, and shape part of science's direction by steering more resources to some big questions. Governments set research priorities that fund at times space telescopes or fundamental physics experiments. They can support such research areas that may not pay off for decades. So for some fields and questions, our constraints can also be influenced by funding in basic research or costintensive research. But it is important to stress that hundreds of major discoveries come from low-cost methods and tools: statistical and mathematical methods, light microscopes, electrophoresis, thermometers, chromatography methods, centrifuges and the PCR method (Chapter 6). Yet science is not method-neutral: researchers and institutions can be slow to adopt new methods, because of inertia with older, familiar methods (Chapter 1). Ultimately, these five factors come together to determine our scope and boundaries of what science we currently explore. Our methods and tools are the common thread underlying each of these factors—uniquely able to both limit and expand the frontiers of discovery more so than other factors can . Here we see how our toolbox sets and extends—in an iterative cycle—what we are able to see, measure and discover at the edge of science (as mapped out in Figure 10.2). All in all, what we discover—and what we even can discover—is always shaped by the power of our current tools, along with our mind and senses. But it is also influenced—varying by discovery and context—by our history, society, economy and human lens. To push past today's barriers to scientific progress, we need new tools that overcome our technical limits but also our cognitive and human constraints that can slow us down—or hold us back. Beyond our extraordinary methods and what our mind is methodologically capable of developing, we have no other way to explore and discover the unknown. Science and its current limits move in a loop that begins and ends with our tools and methods. But we keep redrawing the edges of that loop through innovative toolmaking that stretches what is possible. To understand the borders of science we have to recognise that science is done by humans—and we face methodological and cognitive bottlenecks that draw our present boundaries. Deeper into the future Statistical modelling – climate change Deeper into the past Deeperinto the micro scale Electrophoresis – DNA Sequencing Electron microscope – atoms at quantum scale Hubble space telescope – size and expansion of the universe Advanced statistical methods – sources of global economic stability Our methods and tools More vast Laser interferometer – gravitational waves Molecular beam epitaxy – giant magnetoresistance More distant Seismometer – Earth's core Radio telescope – galaxies and pulsars More rapid or slow Interferometer – speed of light Seismometer – seafloor spreading Present limits of science Present limits of our methods Legend: Method or tool used to study – phenomena in science Our methods and tools Our place and time Social context Human perspective Our mind Present limits of our mind Our mind (observing, problem solving, experimenting) More complex Controlled, statistical experiments – societal behaviour and markets Psychological surveys – causes of mental disorders i like the evolution of our planet or species Deeper into the macro scale Mathematical field equations – modelling spacetime curvature and black holes Quadrant electrometer – radioactive substances Highly sensitive radio receiver – deeppast events like the age of the universe Radiocarbon dating – ancient fossils, artifacts and civilisations More intangibleG𝜇v + Λg𝜇v = 8𝜋G T𝜇vc4 Our mind Computational simulations – far-future events Figure 10.2 The limits of our methods and mind shape the current limits of science We provide examples here for each category, but many phenomena are also studied using other tools. Phenomena can also fit into more than one category—with for example ecosystems studied on a large scale and are also very complex systems. The factors are construed broadly to flexibly capture other features of what we explore. T Yet with new tools, we keep breaking through the frontier of discovery. Alone in the past few decades, nobel-prize breakthroughs in genetics, molecular biology and astronomy have rewritten what we thought was possible, but also in quantum physics and climate modelling. The pattern shows multiple breakthroughs clustered in fields where new tools enabled unprecedented observations—from the molecular scale (gene editing tools) to the cosmic scale (space telescopes) to the quantum scale (particle accelerators and quantum computers). We have made many major recent discoveries across fields—as shown in Figure 10.3 (even though it takes 21 years on average from discovery to Nobel prize recognition). Structure of ATP syntha se Bose-E instein cond ensates Aquaporin wat er channels Potassiu m ion ch annel Molecular logic of o lfactory perception Frequency combs Molecular basis of eukaryotic transcription RNA i nterfere nce Anis otropy of cosmic microwave background radiation Blackbod y form of cosm ic microwave background radiation Mapping of ribosomes Manageme nt of the commons Graphene LPS receptor Toll-gene Accelerat ing expansion of un iver se Reprogr ammation of m ature cells as pluripoten t Gri d cells STED micros copy Single-molecule micr oscopy LED light Neutrino osci llations Mechanism s for autophagy Bacteri orhodop sin at high resolu tion Obse rvat ion of gravitational waves/ LIGO detector Endog enous growth t heory Immu ne checkpoint thera py for cancer Direct ed evolution of enzymes Oxygen sensi ng mechanism i n cells Planet 51 Pegasi b CRISPR Supermassive compact object at ce ntre of g alaxy Contribut ions to labour economics Receptors for temper ature and touch Organocat alysis Bioorth ogo nal chem istry Clic k chemistry Genomes of extinct homi nins Danger model of immune system Human gen ome physics/astro.chemistrymedicine/biologyecon/social Field of discovery 1990 2000 2010 2020 Year the discovery was made Coronavirus vaccin e Meth ods for analysing causal relationships Compu tationa l th eory of mind Neurot ransmitt er r elease Exper imental approach t o alleviatin g pover ty Integrated assessment model of climat e change Figure 10.3 The expanding boundaries of science: discoveries since 1990 The data reflect major discoveries made since 1990—based on science's over 750 major discoveries.
boundaries of science So, where exactly do we reach the present borders of what we can observe, test and verify in the world? Our toolbox sets—at any given point—the current borders of what parts of the world we are able to get into contact with, and those still out of reach. Unlike any other species, we develop complex tools to uncover invisible chemical compounds, microbial life in our biosphere and infrared light. It is only by inventing tools that we—with our limited human senses—can reveal neurons firing in our brains and electrons inside atoms. Before the discovery of the human genome or elementary particles, these mysteries lay outside the periphery of science—invisible and 274 T HE ENGINE OF SCIENTIFIC DISCOVERY inaccessible. But we redrew the borders of biology and physics with breakthrough methods like gene sequencing and particle accelerators—that detected and decoded them. Our scientific boundaries flexibly evolve as we extend our toolbox. But our tools do not work equally well everywhere. As we move into the realm of the tiniest particles, or the largest structures, or the most complex ecosystems, or the furthest distant stars, or the future of our species and human societies, our tools and methods start to strain and are not as powerful. That is where our evidence and theories are not as reliable (Figure 10.2). Here we place our toolbox at the foundation of science and how we evaluate science: if we cannot test and verify evidence and theories using our toolbox, then they cannot be reliable—and remain uncertain. This is why science's core evaluation criteria—testability, verifiability and reliability—fundamentally depend on what our current tools can measure and reveal, and what they cannot. In short: what is—and is not—reliable depends on whether our current evidence and theories involve phenomena that are not yet observable and so not yet verifiable and reliable using our toolbox. So are we getting closer to hitting a wall in some fields, from theoretical physics and theoretical economics to philosophy of science? Indeed, most of their foundational questions remain unresolved. Progress in science has generally stagnated when progress in methods has stagnated—often independent of funding, researchers and imagination. Such fields plateau when they lack new applied methods to see, explore, measure, test and imagine in new ways. Research that largely, or only, relies on abstraction and imagination is most likely to stagnate without real evidence (Chapter 6). What about the limits of mathematics? It has been hailed as a universal language of nature—especially by physicists. Equations describe planets orbiting stars, and electrons moving around nuclei. But mathematics cannot capture many aspects of nature, when we move from predictable physics to dynamic, living systems. From tropical forests and biological systems to brains and societies, simple equations break down. Ecosystems and human behaviour are highly dynamic, and chaotic systems like weather or stock markets involve endless interacting variables—variables that we cannot control or predict well. Mathematical models can come at a cost: simplified, ideal conditions of a complex world. Statistical and computational methods help manage and model complexity. Machine learning methods can bridge some gaps by detecting patterns hidden deep in complex systems that other methods cannot—from diagnosing cancer using medical images to predicting protein shapes. But even these powerful methods cannot always provide exact answers. Consider a thought experiment: imagine what science might look like if we were beings with optimal cognitive architecture—not limited by our specific human niche? We would have a remarkable capacity to quickly process millions of observations— without cutting-edge computers and statistical methods. We would have astonishing eyesight to see through atoms and across galaxies—without advanced microscopes, telescopes and spectrometers. We would have an incredible memory of everything we have observed—without immense datasets. We would understand and predict the world with remarkable accuracy—transcending our human lens. Yet because we are a product of specific evolutionary adaptations within our niche, we do not have these superhuman capacities. But the extraordinary tools we invent do have these capacities—by stretching our evolved mind and senses in ways unimaginable before them.
The tool inventions we have made so far largely determine what we have discovered, what we can discover and what we cannot yet discover without better tools. This powerful principle of tool innovation flips how we think about science's limits: it is not just about mysteries or unanswered questions 'out there' in the universe, it is about tackling the actual roadblocks built into our current tools to explore it. It is about shifting the focus from the big puzzles to overcoming our tools' blind spots, resolution limits and design challenges—that have uncovered science's biggest puzzles so far. Because each of science's major breakthroughs has meant breaking the bottlenecks of what our instruments could detect or measure—whether it is mapping DNA or spotting distant galaxies. We can only extend our frontiers to the parts of reality that our tools open up and let us see. Yet our mind and senses but also our time in history, human lens and institutions also influence where we look and what we explore. But the key limit across all fields is what our methods and tools—from MRI devices to quantum sensors—can actually measure, test and reveal. Across fields, our toolbox is largely what redraws and redefines the outer borders of discovery—more so than other factors can. Our toolbox is the foundations and limits of science: setting what and how we can see, experiment and understand in the world. A key insight emerges here: if we want to understand the bounds of science and tackle them, we have to understand what our tools enable us to observe, test and verify—and what they currently do not enable us to do. The history of science is a history about all the ways we transcended the limits imposed by evolution and by the current tools we designed up to now. It is a history of building instruments that reach farther, see further, think faster and imagine deeper than nature alone ever enabled us to. Just as the limits of an engineer's tools largely define what we can and cannot build, the limits of a scientist's tools largely define what we can and cannot discover. Ultimately, science thrives on its limits and the exciting challenge of finding ways to push through them. Breaking the frontier means leveraging new methods—while everyday science means using conventional methods. So recognising the limits of science is not a weakness, it is the first step to breaking through the uncovered limits—the topic in the next and final chapter. Ultimately, every generation faces the same challenge: do we accept the current boundaries of what we can know and discover? Or do we create new tools and tool combinations—physical, digital, experimental—that continually expand the horizon of the unknown?