policies needed to accelerate scientific progress What if we treated the design of scientific tools across science as a field of research in itself that speeds up discovery? Most scientists do not commonly focus on studying and extending the very methods that enable them to do research. But if researchers and science institutions began devoting much greater attention and resources to tool innovation, we could unlock many important questions at the very core of scientific progress: How can we best incentivise methods projects, not just science projects? What research programmes and networks can we restructure or create to catalyse tool innovations—especially among earlyand mid-career researchers? How can we reorganise scientific research to tackle the hidden bottlenecks within our current tools that hinder progress? How can we best support interdisciplinary research that combines method developers, tool innovators and scientists collaborating together? How can we reallocate part of existing funds to pioneer new tools that push the frontier? Mobilising more funding, training more scientists or building larger teams—without addressing such questions—will not solve our challenges. Without identifying such strategic methodological areas for advancing science, breakthroughs risk being more driven by chance than by design—ultimately slowing the pace of progress. The central paradox of science and its incentive system emerges here: the very researchers building the tools that define the frontier of research are often overlooked by the institutions meant to advance it. Targeted research on tool-building would vastly accelerate science, so it is surprising that it is not yet valued greatly across science and society—at least not nearly as much as scientific discoveries and theories. It is astonishing that, despite centuries of scientific expansion, no field has emerged that systematically studies tool development from all angles and perspectives across science—building the very methods and tools that make science possible. Students in many fields are required to take a course in research methods or statistics, but these courses typically focus on using existing tools, not on how to innovate them and tackle their limitations. So why has a field devoted to innovations in tools across science not yet emerged? One explanation lies in a lack of general awareness about how profoundly methods and tools actually shape scientific progress. Another is a lack of incentives for researchers. Science's reward system—journals, universities, major awards and grant agencies—prioritise scientific discoveries. Science textbooks and public media also spotlight these final outputs. In short, science and society celebrate experimental findings and theories over foundational inputs—the method innovations that enable them. Like some areas of basic research, research on tool innovation can be less appealing for scientists. Since the scientific ecosystem rewards those who publish striking results quickly, researchers often have little motivation to invest time in method-making. Yet because our available tools largely determine what we are able to know and discover, what could be more important than a research field devoted to expanding our scientific tools—our very means of knowing? This makes a field like the Methodology of Science essential—one that does not just study knowledge, but invents the ways we can produce new knowledge. With no field yet existing that aims to advance tools of discovery across science, we propose such a field here and outline what it can look like: The Methodology of Science is a field dedicated to understanding and designing the scientific methods and tools that enable better ways to discover, measure and explain the world. It studies the foundations, limitations and advancement of our tools: from observational and experimental to statistical and computational tools. The field does not view methods and tools as technical supports, but places them at the centre of scientific progress—because how we investigate determines what we can discover. It identifies the constraints, assumptions and biases facing tools and develops ways to tackle them. The field continually scans across disciplines for new method combinations and maps how new tools shape knowledge production and redefine disciplines. It builds systems for interdisciplinary method-building, offers a platform for rapidly disseminating promising tools across fields and trains researchers how to adapt and invent new tools of discovery. Ultimately, the field offers a framework for how to make method innovations—not just to conduct science, but to build science and discoveries themselves. The field would represent a new domain of basic research. It would serve as a cross-disciplinary bridge, bringing together toolmakers, methodologists, experimentalists, computational scientists, engineers and field-specific researchers in a culture of innovation. Existing research to improve a method or tool—such as statistical techniques or microscopy—has been domain-specific and fragmented. But this field would rise above such disciplinary silos as a kind of meta-method discipline, integrating and recombining entirely new kinds of tools and methods from across the sciences to accelerate progress. The vision is to reframe tools not as background instruments, but as evolving systems of perception, reasoning, exploration, imagination and innovation. And the hope is that developing more sophisticated telescopes, computational methods and x-ray techniques would no longer be seen as less important than discovering new planets or decoding DNA or protein structures—the breakthroughs these very tools make possible. The new methods-driven discovery theory can provide a theoretical foundation for the Methodology of Science. For the theory explains how the foundations, limits and advancement of science are largely determined by the foundations, limits and advancement of the tools we have developed so far—that produce the data enabling our theories. What is the best test to assess a new field? We can ask, how useful is it to address unsolved problems? To open pathways and solutions to problems not yet formulated? To reveal systemic blind spots to progress? And to advance new discoveries and entire lines of research? On these measures, a new science of methodology has great potential: because the greatest accelerations in scientific progress throughout history have followed major method breakthroughs (Chapters 1–5) and because of the range of constraints and untapped opportunities of today's leading tools (Chapters 6 and 11). So what would it take to establish the field of Methodology of Science and accelerate the pace of discovery? We propose seven key reforms: developing a broad method community, embracing tool experimentation, rethinking our conception of science, introducing new incentives for tool creation, building dedicated method institutions, recalibrating citation practices, and training the next generation of tool developers. 1. We need a methodological shift in the scientific community towards a much larger share of researchers dedicated to tackling method bottlenecks and making tool innovations. About nine million scientists work across science worldwide. Yet if we ask scientists to name researchers dedicated full-time to developing or improving methods and tools, most would not be able to—or, in some fields, not even know they exist. Such researchers are rare and largely not visible. While tools power every experiment, model and dataset, few researchers are explicitly devoted to building them. Our ability to investigate the world is generally not constrained by the lack of funding or collaboration—for most discoveries; but by the scarcity of tool-builders. To accelerate progress, we urgently need a methodological shift: a structural change that moves many more researchers towards solving barriers facing our tools. What obstacles prevent us from observing the world more precisely? What measurement systems and instruments do we need to redesign to expand our scope to the world? How can we strategically foster the diverse methodological pathways—described in the previous section—to spur new research areas? Answering these questions systematically across science requires more than ad-hoc methodological curiosity—it requires a growing community of method innovators and tool architects who treat method-building as a core scientific endeavour. Imagine if just 1–3% of the global scientific workforce, a modest fraction, was devoted only to tool development, as proposed here—dedicated to the Methodology of Science. This shift alone would give rise to 90,000 to 270,000 full-time researchers building the tools of discovery. This is a small share but the impact could be enormous, since every breakthrough fundamentally relies on the strength of our tools. Such a shift would catalyse a generation of tool architects who reimagine not just what we study, but how we study. Some can for example develop new AI systems that integrate with any scientific data, helping researchers generate experimental designs and make better predictions. Others can invent hybrid tools that blend simulation, visualisation and real-time analytics, enabling advances in fields from neuroscience to climate science. By embedding toolmaking more deeply into the scientific workforce, we can build a faster, more self-correcting and more adaptive research system. Hidden bottlenecks would be flagged and outdated methods would be upgraded, allowing us for the first time to more strategically redirect slower or stagnating fields. 2. We need to make testing and expanding our best methods a regular part of what we all do—often as an end in itself. Making our most powerful methods in statistics, spectrography and chromatography even more powerful, making our best telescopes and particle accelerators even better has consistently paved the way for new discoveries. Many big advances in science are not guided by a specific hypothesis that led to the specific discovery, but are sparked by experimentally developing a new tool applied to do exploratory research and then seeing something entirely new (Chapter 2). Just as we run experiments in science, we need to run instrumental and method experiments—we need to design research not to generate findings or test a theory, but to test a tool. What if labs began running method trials—like clinical trials, but for scientific tools—where a control group uses a conventional tool, and experimental groups test new tools to compare which are the most effective in achieving advances? It is essential that we treat tool developments as a primary scientific goal even before we know what discoveries they could yield— while of course always taking our planet and human wellbeing into account. Just like with experimental research, tool innovation requires us to try new things and be open to inevitable failures, as learning opportunities that refine approaches. Once the scientific community becomes fully aware of the fact that new tools trigger new breakthroughs, we can begin to target research programmes and exploratory experimentation to catalyse tool invention across science. This shift would reframe how we understand discovery itself: from its traditional focus on just experimental and theoretical triumphs to equally focus on the tool discoveries that make them possible. In sum: method innovation needs to be a core pillar of science. 3. We need to reconceive science and science policy around methods-driven research that is exploratory—not just question-driven research. In science, articles and grant proposals are structured around a research question: 'what causes x?' or 'how does y work?' This structure reinforces the idea that all science should start with a hypothesis or puzzle that needs solving. But many of science's transformative discoveries have begun not with a question but with new tools through exploratory research (Chapter 2). We need to revise the traditional view that science must always be questionand hypothesis-driven, it must also be methodsdriven—open to the unknown and sometimes questionless. We can for example opportunistically exploit new methods created in one's own and other fields. What happens when we apply new imaging methods to biological or ecological systems to observe what they show? Or when we test a new statistical technique or new machine learning method on climate or psychological data with no hypothesis in mind? Tooldriven research has enabled spotting patterns, anomalies and structures—and even formulating many of our biggest questions in science—that we did not even know to explore. It comes with uncertainty and less clear outcomes, but the payoff for method opportunists is often discovering what no one else has thought to look for. Science agencies need to equally support exploratory research projects—not just traditional projects that already have consensus among established research communities. A methods-first mindset, where researchers invent, adapt and apply methods to new domains and shifting challenges, can be just as powerful as traditional hypothesis testing. 4. We need to fix science's reward system so journals, prizes and funding bodies equally incentivise method experimentation and innovations. When asking scientists what brings recognition in their field, the answer is generally straightforward: the main incentive in science is to chase immediate scientific discoveries. This focus is deeply embedded across science's institutions: journals and science prizes reward groundbreaking findings, universities hire and promote based on breakthroughs, and funding bodies support the greatest prospect of discovery. But what if we have been overly incentivising the final part of the discovery process? Here we provide an alternative logic: if we reform science's institutions to value not only results, but equally developing new tools and methods that make those results possible, we can accelerate discoveries. Rather than largely only prioritising the finish line, we must prioritise tool-building: the roads to get to the finish line and often the key part of the finish line itself. Foundational research in methods, at first glance, may seem to even distract our attention—deter from publishing experimental and theoretical studies: the central measure in science. But it does not delay gratification, it lays the groundwork for getting there faster and is often a multiplier: opening new paths. It can often be thought of as a difference between harvesting a tree and planting an orchard—one focuses more on today, the other more on the broader future. Today, the greatest progress in science commonly comes not from chasing prestige, but from bold new ways to explore and measure—like highly advanced microscopes, machine learning methods and the CRISPR method. These enormous breakthroughs are made possible by method-makers: researchers who often step back from the race for headlines to redesign the track. Our current research system overwhelmingly invests in science projects —experimental and theoretical research—as we see looking at projects supported by science's largest funding agencies like the European Commission and the National Science Foundation. The current system does not yet systematically invest in method projects. Yet we need to begin reallocating a large share of current funding to method projects, tool programmes and tool fellowships and also give priority to science projects that include method innovation. Funding currently disproportionately bets on established researchers and elite institutions. But we need to disproportionally redistribute resources to bold original tool innovations to unlock discoveries faster, independent of prestige and institution. For our current incentive structure is at odds with advancing science more rapidly. It is time we recognise the inventor of a breakthrough telescope—the tool of discovery—not only the researcher who discovers the planet using it. And the taxonomy of scientific methods—spotlighted in Table 6.2—highlights the need to prioritise the most scalable, transferable tools. Just as in basic research, researchers dedicated to the foundational research of designing tools deserve equal recognition within the scientific community as those generating scientific studies and theories made by leveraging them. To speed up progress, we must shift science's priorities: experimenting with new tools and method combinations needs to be seen on par with the experimental findings they produce. Method-makers need to equally be able to publish method advances in top-tier journals, win prestigious grants and prizes and secure tenure-track positions. We need professorships in methodology, in tool innovation, in integrated machine learning. The hope here is that method-inclined researchers receive the same opportunities, resources and positions as those carrying out experimental and theoretical research using their powerful innovations. These measures are crucial to incentivise young researchers to enter this basic research field and explore high-risk, high-reward toolbuilding projects—projects that do not fit into conventional disciplinary silos but can reshape their disciplines. The way we publish research does not help: many leading journals like Nature and Science follow an 'introduction–main results–discussion' format, while relegating the methods used to the appendix or a separate 'supplementary material' file. This sends a signal: they view methods as subordinate or supplementary. Journals need to correct this bias. They must also vastly increase featuring high-impact tool advances. They should begin requiring a methods section with much greater detail in all studies, outlining all techniques, protocols, study context and method limitations throughout the research process. This would empower other researchers to improve and build on those methods, understand and reduce method constraints and biases and push the field forward faster. It would enhance method awareness. So this can help us rethink questions about science: should we prioritise funding discoveries, or toolmaking that sparks them? Should we chase new findings, or expand the foundations that enable them? And is it better to fund research or development—to foster new knowledge or expand existing knowledge? For each question, the answer—of course—is both, but we are currently biased much too far towards the former. Supporting method-makers directly gives science its infrastructure of discovery. It empowers researchers to pursue opportunities as they emerge, to shift and re-shift their focus as new problems arise and to take risks that conventional structures do not yet reward. 5. We need to develop new Methods Labs, Methods Hubs and Departments of Methodology of Science—as incubators of innovation. Just as scientists work in science labs, we need to establish methods labs and hubs—spaces designed to invent new methods and tools of discovery. These labs would provide, for the first time, a systematic infrastructure for tool innovation across fields—and house the growing share of method-makers among the scientific community. These labs can connect tool and method innovators, statisticians, experimentalists, AI researchers and domain-specific researchers to work side by side on breakthrough instruments. These labs would streamline methodchains and toolchains that span the full method design—from the early concept design to real-world applications. No such systematic methods powerhouses exist today. But the Cavendish Lab and Bell Labs—the two most productive methods incubators in history—make clear the enormous effect of such labs on driving progress. And unlike those physics-centred labs of the past, the future of a global network of methods labs lies not in one field but spans across science and bridges diverse toolmakers across disciplines. It unites synthetic biologists with data scientists, neuroscientists with roboticists, ecologists with machine learning experts and statisticians. We could measure the success of the Methodology of Science by establishing methods labs, hubs, institutes, centres and departments that train early-career scientists to become toolmakers. It would include creating a society, journals and conferences across fields focused on improving how (not what) we discover. Communities of method-makers could then pool resources through open-ended experimentation with innovative and adaptive tools that tackle our methodological barriers to progress. In a landscape where science grows increasingly fragmented into narrower subfields, methods labs could reconnect disciplines by giving them a shared language of innovation. This is important because current trends of dividing science into ever more narrow units of knowledge can slow different paths of discovery (Chapter 4). But methods labs worldwide would not only build bridges for using new general-purpose and field-specific methods across disconnected fields. They would also export groundbreaking method innovations—through shared method and tool repositories that open access worldwide. The greatest accelerator of science may not be any one discovery, but a strategic shift: where institutions and scientists begin to treat toolmaking itself as a key priority. This is not just a reform—it is a redesign of how science builds its future more efficiently. What could Departments of Methodology of Science look like? Imagine a department where the logic, mechanics and innovation of science itself are studied, taught and advanced. It is an academic space of scientific self-reflection and invention. Researchers here can explore questions like: what features make a research method more likely to generalise across fields? How can we design early-warning systems to detect flawed or outdated methods in a field? What constraints are introduced by statistical, computational and experimental methods? What are the best ways to measure a method's impact—on clinical trials, climate models or public policy? How can AI and automation be best coupled with research methods? Can we map the genealogy of science's major research methods—in a periodic table of methods—to better identify methodological blind spots and missing combinations quicker? Such departments can also collaborate with all fields to detect and design new method and tool combinations. They can also work specifically with medical schools to co-develop trial designs, with engineering departments to prototype sensors or robotic tools, with social science institutes to rethink survey and causal inference techniques and with AI labs to explore automated tools. Courses in Departments of Methodology of Science could also range widely. A course on Tool Innovation: Designing, Building and Testingcan cover how to devise, prototype and evaluate method innovations—from experimental designs and new types of sensors, to AI systems, measurement instruments and software platforms. A course on The Architecture of Scientific Discovery can introduce how methods emerge, spread and evolve, and what the hidden barriers are behind measurement, modelling and experimentation. A course on Method Transfer and Translation Across Fields can explore how methods from one discipline (like advanced statistical and machine learning methods) can be adapted for another (like ecology or epidemiology). A course on Bias, Inference and Error: A Statistical Toolkit can cover statistical reasoning, alternative inference frameworks (like Bayesian methods and causal inference) and teach how to detect and correct methodological errors. A course on Meta-Science and the Practice of Science can explore how the scientific system works and cover topics like incentives, research, publication systems, funding systems, replication crisis and how to reform scientific institutions. These courses would prepare a new generation of tool builders—those fluent in the scaffolding of tools and discovery themselves. Creating such departments would make the scientific enterprise more self-correcting and more innovative. 6. We need to shift our main measure of scientific success from just outputs (especially article citations) to focus equally on inputs—method innovations. In today's scientific landscape, getting hired, promoted or funded is overwhelmingly tied to how many citations and publications scientists accumulate. This metric is easy to measure but overlooks something crucial: scientists do not yet cite the full range of methods, tools, techniques and protocols that underpin their work. Instead, they primarily just cite other research papers—not the methodological scaffolding. This creates a distorted picture of what matters in science and discourages developing new instruments and techniques. A rethinking is needed away from using common ex-post indicators like citation counts towards ex-ante indicators—the process of science—that recognise tool innovations. Today, many view articles as high-impact if they, depending on the field, receive one or a few hundred citations. But what if we begin to systematically reference the methods and tools applied in each study? The foundational role and impact of methods would become powerfully clear. The most important method-making articles in history, such as Ruska's article on the invention of the electron microscope, Tiselius' electrophoresis method, Martin and Synge's partition chromatography method or Bloch and Purcell's NMR spectroscopy have only received a few hundred to a few thousand citations. But each is mentioned in millions of publications on Google Scholar, making them the real giants in science. If citation practices accounted well for tools, then inferential statistics, x-ray crystallography, computational methods, randomised controlled experimentation and lasers for example would each have received millions of citations. For they are mentioned millions of times on Google Scholar but are not properly cited. This far surpasses the most-cited scientific studies of all time that rarely give (citation) credit to the discovery of the tools they use. The paradox is clear: our best scientific tools are so foundational to science that they are largely taken as given, without the need to even cite them across most—but not all—fields. By relying on citations as our primary measure of success, we foster an outputoriented, retrospective view of knowledge. Many new breakthroughs would go unseen when just focusing on citations, and new discoveries are important precisely because they have never been seen and cited before. Citations not only just capture past success—in a crude way—but, more importantly, cannot explain why or how breakthroughs emerge. For that, we need to shift our attention to methodological and scientific innovations themselves. When our incentive system recognises the indispensable power of methods in driving discovery, researchers will also recognise the enormous value of making innovations in tools. 7. We need to train the next generation of tool-innovators—and integrate deep methodological training into university degrees across science. In fields from biology and medicine to earth sciences, graduate students are typically required to take at most only one course in research methods or statistics during their degree. And yet, these are the very tools they later rely on to do and publish their research. It is not difficult to see how this gap can constrain research. With the purpose of graduate degrees to train students how to do science, we need to place much greater focus not only on training how to apply but also improve methods. Limited method training—both in breadth (the range of tools) and depth (their limitations, assumptions and biases)—narrows a scientist's ability to recognise the flaws facing the tools used to develop our evidence and theories and to fix them. Yet this ability is critical in sparking discoveries (Chapter 1). Without first learning how to detect our methodological blind spots, we cannot tackle them—with the pathways to do so laid out earlier. Most education systems instead focus on the outputs of science—facts, theories and discoveries—rather than the process in which we generate them. We need to reform this imbalance: shifting much greater weight on studying the complexities of how we implement and fine-tune our tools to gain that knowledge. Scientific training must be recentred around the how of discovery, not just the what. This means immersing students in the complexities of measurement, experimental design, inference, calibration and innovation—not as side topics, but as the core of scientific literacy. A critical barrier here is hyper-specialisation in education, a barrier that leads to long time lags in developing and adopting tools, especially across fields. This leads to path dependency, a kind of method inertia: a field-wide anchoring effect on the specific methods that researchers already use, while powerful new techniques from other fields often go unnoticed (Chapter 1). Path dependency, disciplinary silos and a lack of cross-pollination stifle innovation. To break this inertia, education systems need to systematically expose scientists to the broader universe of tools and techniques. Just as modern biology has been transformed by techniques from computer science and physics, many of tomorrow's breakthroughs will come from such cross-disciplinary fertilisation—from researchers fluent in multiple toolkits. To make this shift sustainable, we must go one step further: we need to institutionalise the training of method-makers. To establish the field of Methodology of Science, tool and method innovators would need to be formally trained, just as chemists and psychologists are. We cannot assume that some researchers, who happen to develop an inclination for method and tool questions, will find the time to tackle them. Without structured support and training, many potential innovators will never get the chance to build the tools science needs. Historically, only a few exceptional scientists have strayed away from their original academic training as the top ten method outsiders did—science's greatest architects of discovery. But our most transformative innovations in science should no longer depend on chance. They must be systematically built into the system. Establishing dedicated training in this field will empower early-career researchers with the necessary methodological skills to push the boundaries of what we can investigate in the world. And method workshops and training will help equip established researchers across fields with these skills. Only when tool innovation is valued and taught can science truly accelerate its capacity to discover the unknown. Imagine a generation of scientists trained not just to apply existing tools, but to question and reinvent them. This would be nothing short of a shift in how knowledge is created.