Grand Challenges, Pasteur’s Quadrant, and the Knowledge Problem

A somewhat long read on the rise of the grand challenges agenda

Few ideas have acquired greater prestige in contemporary science policy than the notion of “grand challenges.” Across funding agencies, universities, philanthropic foundations, and international organizations, researchers are increasingly encouraged to orient their activities toward addressing humanity’s most pressing problems. These are almost invariably climate change, sustainability, inequality, pandemics, preserving democracy, and social resilience. They seem to more seldom be, say, immigration challenges or the problems of sluggish economic growth (arguably, the main challenge in Europe). In any case, such concerns has become the organizing principles of research strategies, funding programs, and institutional missions. They have also insinuated themselves into management research and the mission statements of business schools (e.g., here).

The appeal of this development is obvious. Indeed, why should society, here proxied by the tax payers and their representatives, not ask scientists to devote attention to problems of immense practical significance? Why should scarce resources be invested in questions whose relevance appears uncertain when humanity faces challenges of unprecedented scale and urgency? In an era characterized by growing demands for accountability and demonstrable social impact, the expectation that publicly funded research should contribute directly to solving societal problems appears entirely reasonable.

No contemporary scholar has done more to popularize this vision than Mariana Mazzucato Through books such as The Entrepreneurial State and Mission Economy, Mazzucato has argued that governments should move beyond “merely” correcting market failures and instead actively shape and direct innovation toward societally desirable objectives. Her influence has extended far beyond academia, informing European Commission policy, national innovation strategies, and the strategic planning exercises of universities around the world. Mazzucato has become the principal intellectual architect of the contemporary mission-oriented turn in science and innovation policy.

However, the remarkable rise of the grand-challenges agenda has been accompanied by relatively little scrutiny of its underlying assumptions (and, perhaps, little public knowledge of the scrutiny that does exist). The language of missions and societal challenges is often presented as self-evidently desirable, as though directing research toward recognized social goals were simply common sense. Critics are easily portrayed as reactionary defenders of an antiquated vision of academic inquiry detached from practical concerns.

This framing is misleading. The question is not whether science should be useful. Nor is it whether practical problems can stimulate important discoveries. Few serious observers would deny either proposition. The more interesting question concerns governance rather than usefulness. Does it follow from the practical relevance of scientific research that research systems themselves should be organized around societally defined missions? Can scientific discovery be directed toward predefined objectives with sufficient reliability to justify the growing emphasis on grand challenges?

I shall argue that the answer is far less obvious than much contemporary science policy discourse suggests. The strongest arguments commonly invoked in support of mission-oriented research fail to establish the conclusions drawn from them. More importantly, the grand-challenges agenda rests upon a series of assumptions about knowledge, discovery, and institutional incentives that are highly questionable.

Pasteur’s Quadrant and the governance fallacy

One intellectual resource frequently invoked in discussions of mission-oriented research is Donald Stokes’ concept of Pasteur’s Quadrant.

Stokes (rightly) challenged the traditional distinction between basic and applied research and argued that some of the most important scientific advances emerge from attempts to understand practical problems. Louis Pasteur’s investigations of fermentation and infectious disease transformed both agriculture and biology. His work was motivated simultaneously by a desire to solve practical problems and by a desire to understand the underlying phenomena.

Stokes’ contribution was important and warranted because it undermined a simplistic view of scientific inquiry. Research does not always proceed from curiosity-driven discovery to subsequent application. Practical problems can stimulate fundamental advances in understanding. The pursuit of usefulness and the pursuit of knowledge are often complements rather than substitutes.

However, contemporary discussions of grand challenges frequently attach implications to Stokes’ argument that it cannot bear.

Pasteur’s Quadrant is a theory of scientific motivation. It describes the relationship between practical concerns and scientific discovery. It is not a theory of research governance. Nothing in Stokes’ argument, I submit, establishes that governments, funding agencies, or university administrators can successfully identify the practical problems most likely to generate important discoveries. Nothing in it suggests that research systems perform better when organized around centrally designated missions. Nor does it imply that scientific priorities should be determined through strategic planning exercises focused on societally defined objectives.

Indeed, the historical example of Pasteur points in the opposite direction. Pasteur’s achievements emerged from a process of exploration whose trajectory could not have been predicted in advance. The practical relevance of his investigations helped stimulate discovery, but the discoveries themselves emerged through a sequence of evolving questions, unexpected findings, and intellectual detours. Pasteur succeeded, not because he was executing a carefully specified mission but because he possessed the freedom to follow inquiry where it led.

The fact that practical concerns can stimulate scientific discovery therefore does not imply that scientific discovery can be effectively organized around predefined societal missions. Conflating these propositions represents what might be called the governance fallacy of the grand-challenges movement.

This distinction is particularly important because much contemporary advocacy of mission-oriented innovation implicitly moves from observations about the character of scientific discovery to conclusions about how scientific systems should be governed. Mazzucato’s work frequently invokes examples in which practical problems stimulated important scientific and technological advances. But, the existence of such examples does not establish that governments can identify ex ante which practical problems are most likely to generate future breakthroughs. To be a bit repetitive, Pasteur’s Quadrant demonstrates that useful problems can stimulate discovery. It does not demonstrate that discovery can be successfully organized around politically designated missions.

The moonshot fallacy

However, this kind of confusion characterizes the frequent use of historical “moonshots” as exemplars of mission-oriented innovation. (See this excellent recent open-access takedown of this line of thinking as well as its predecessor, to which Peter G Klein, Samuel Murtinu and yours truly contributed a chapter).

The most influential contemporary version of this argument appears in Mazzucato’s Mission Economy, which treats the Apollo program as a model for addressing contemporary societal challenges. The implicit claim is that the institutional capabilities that enabled a lunar landing can be redeployed to solve problems such as climate change, inequality, and sustainability. But, this inference depends upon overlooking precisely those features that made Apollo unusual.

Apollo was not a grand challenge in the contemporary sense. It was a large-scale engineering project with a clearly specified objective, a single customer, well-defined lines of authority, and an unambiguous performance criterion. Success could be directly observed: either astronauts reached the moon or they did not.

Most contemporary grand challenges possess none of these characteristics; climate change, inequality, sustainability, healthy aging, or making economic growth return to Europe are not engineering problems. These challenges involve complex adaptive systems with interactions among technologies, institutions, incentives, political processes, cultural norms, and economic behavior. Their objectives are often contested, they are shot through with causal ambiguity, and their solutions depend upon knowledge that has not yet been discovered.

The appeal of the moonshot metaphor arguably derives precisely from its ability to obscure these differences. It encourages the belief that social challenges can be approached as though they were engineering projects awaiting sufficient political will and organizational commitment. However, the defining feature of most grand challenges is that they involve forms of uncertainty and unknowledge fundamentally different from those confronted by Apollo.

In a nutshell, the challenge is not merely to implement known solutions; it is to discover knowledge that does not yet exist.

The knowledge problem in science policy

This observation brings us to what is perhaps the central weakness of the grand-challenges framework: its implicit theory of knowledge.

Mission-oriented science assumes that society can identify important problems, formulate appropriate missions, mobilize resources, and thereby accelerate the production of socially valuable knowledge. Scientific progress appears as a process that can be directed toward desirable outcomes through sufficiently intelligent coordination and maximize a well specified social welfare function.

The difficulty is not merely that policymakers possess incomplete information; the more fundamental difficulty is that the relevant knowledge often does not yet exist.

This observation places scientific discovery squarely within the tradition of thought associated with Adam Smith, Friedrich Hayek, and Michael Polanyi. Smith’s analysis of social order emphasized the capacity of decentralized processes to generate outcomes that no individual actor intended or foresaw. Hayek extended this argument by stressing the dispersed and often tacit character of knowledge. Information relevant to social coordination is distributed across individuals and contexts, making centralized direction inherently difficult. Polanyi subsequently applied similar reasoning to science itself. In his famous essay, “The Republic of Science,” he argued that scientific progress depends upon a decentralized community of investigators pursuing problems according to their own judgments, expertise, and interests. The advancement of knowledge emerges not because someone successfully plans it, but because many researchers simultaneously explore a wide range of possibilities.

The history of science has repeatedly confirmed this view. Quantum mechanics was not developed to create modern computing; molecular genetics was not originally motivated by aspirations to transform medicine; information theory did not emerge from a coordinated effort to build the internet. In each case, scientific advances generated applications whose significance could not have been anticipated at the time the discoveries were made.

Scientific discovery is therefore characterized by a form of uncertainty and ignorance that extends beyond ordinary risk. The most important future discoveries are often valuable precisely because nobody knows what they will be. This creates a profound challenge for research governance. How does one direct inquiry toward outcomes that cannot yet be identified?

The grand-challenges agenda rarely confronts this question directly. Mission-oriented innovation policy often proceeds as though identifying a social problem simultaneously identifies the knowledge required to solve it. However, this assumption collapses two very different activities. Society may know that climate change is important, but it doesn’t follow that society knows which scientific discoveries will prove decisive in addressing it. The former is a political judgment. The latter is precisely what scientific inquiry is supposed to discover.

Recognizing that climate change is important does not imply knowing which scientific discoveries will ultimately prove most valuable in addressing it. Similarly, recognizing that aging populations create social challenges does not imply knowing which research trajectories will generate the most important future advances. The relevant knowledge emerges through discovery itself.

This is why science is best understood not primarily as a problem-solving machine but as a discovery system. Its greatest contribution to society often lies not in generating solutions to known problems but in creating possibilities that nobody previously imagined.

The visibility bias of mission-oriented science

In addition to the above knowledge problem, there are also incentive problems associated with grand challenges. These incentive problems are reinforced by a deeper methodological problem that receives surprisingly little attention in discussions of science policy. Put simply, mission-oriented research is often evaluated in a manner that systematically exaggerates its apparent success.

The problem is familiar to anyone with even a passing acquaintance with empirical research: One cannot evaluate a process by examining only its successes. However, this is precisely how mission-oriented science is frequently discussed.

Advocates of grand challenges can indeed point to an impressive catalogue of achievements. They can identify important technological advances, successful public programs, breakthroughs in medicine, advances in agriculture, military technologies that subsequently found civilian applications, and countless other examples in which public investment contributed to socially valuable outcomes. However, the question is what such examples demonstrate.

The difficulty is that successful outcomes are highly visible, while the relevant alternatives are not. Every successful research program exists alongside a much larger number of unsuccessful programs, abandoned initiatives, technological dead ends, and opportunities foregone. More importantly, every allocation of scientific resources necessarily displaces alternative allocations. Researchers who pursue one line of inquiry cannot simultaneously pursue another. Funding devoted to one priority is unavailable for others. Scientific attention is itself a scarce resource.

A proper evaluation of mission-oriented governance would therefore require comparison not merely with failure, but with plausible alternatives. What discoveries might have emerged had resources been allocated differently? Which opportunities for exploration were foreclosed by concentration on officially designated priorities? Which breakthroughs never occurred because researchers were encouraged to pursue approved missions rather than unconventional questions? Which scientific fields failed to emerge because resources flowed elsewhere?

These questions are extraordinarily difficult to answer because the relevant outcomes are largely invisible. Successful missions leave observable traces, but foregone discoveries don’t.

This asymmetry creates a systematic bias in policy evaluation. Mission-oriented programs are judged by their visible successes, while the opportunity costs associated with alternative uses of scientific resources remain largely unknowable. The resulting narratives naturally exaggerate the apparent efficacy of strategic direction.

The problem closely resembles a recurring error in discussions of innovation policy more generally. Histories of successful technologies frequently trace their origins to earlier public investments and then infer that those investments caused the success. Such reasoning often evaluates institutions by examining only the winners. The relevant denominator—the full portfolio of investments, including failures and missed opportunities—disappears from view. Once attention is restricted to successful cases, almost any innovation system can be made to appear extraordinarily effective.

A similar logic operates in the grand-challenges literature. Once a discovery has occurred, it becomes possible to construct a persuasive narrative linking scientific investment, institutional priorities, and societal outcomes. The success appears to validate the mission. The resulting account often possesses considerable rhetorical force because it is built around observable achievements. What disappears from view are the countless alternatives that never materialized.

In a recent text on X, Pedro Santa Clara, a finance prof at the Nova School of Business and Economics in Portugal, neatly documents such fairly basic blunders in Mazzucato’s thinking.

Pedro Santa Clara@psantacl

https://t.co/jJTpp3JjDN

1:33 PM · Jun 10, 2026 · 50.1K Views

---

16 Replies · 72 Reposts · 270 Likes

Thus, The Entrepreneurial State’s central argumentative strategy consists of tracing successful technologies back to earlier public investments and inferring from those successes a central role for strategic state action. Critics have pointed out that this mode of reasoning suffers from a fundamental methodological weakness. It focuses attention on successful outcomes while largely neglecting the relevant denominator. Public investments that contributed to successful technologies are visible. Failed investments are less visible. More importantly, the innovations that might have emerged from alternative allocations of resources are entirely unobservable. The result is a form of selection on the dependent variable: institutions are evaluated through their successes rather than through the full portfolio of outcomes they generate.

This is not merely a statistical problem. It is also a cognitive one. Human beings are naturally inclined to interpret successful outcomes as evidence of foresight and intentional design. Once a breakthrough has occurred, the sequence of events leading to it appears more orderly and predictable than it actually was. Histories of innovation are frequently reconstructed from the perspective of known outcomes. Uncertainty, contingency, and failed alternatives recede into the background.

Scientific discovery is particularly susceptible to this form of retrospective illusion. Consider once again the case of Pasteur. From the perspective of the present, Pasteur appears to exemplify the successful pursuit of socially relevant science. His investigations addressed practical problems and generated transformative discoveries. Yet this interpretation benefits from knowledge unavailable to contemporaries. What they observed was not the execution of a successful mission but a scientist navigating a series of uncertain questions whose significance emerged only gradually through the process of inquiry itself.

The path from problem to discovery often appears far more coherent in retrospect than it did at the time.

The danger is that successful discoveries become evidence for the efficacy of direction itself. Because visible successes can be linked to identifiable priorities, observers infer a degree of foresight and control that never actually existed. Scientific progress comes to be understood as the product of strategic coordination rather than exploration. The contribution of decentralized discovery becomes progressively harder to see precisely because it is so difficult to reconstruct the opportunities that were never explored.

The visibility bias therefore reinforces the knowledge problem discussed earlier. If the future value of scientific discoveries cannot be known in advance, then evaluations based on visible successes will systematically overstate the effectiveness of ex ante direction. The apparent success of mission-oriented research may tell us considerably less about the quality of strategic planning than its proponents assume.

Indeed, one might plausibly argue that the most valuable characteristic of successful scientific systems is not their ability to pursue recognized priorities but their ability to generate outcomes that nobody anticipated. Yet these are precisely the outcomes least compatible with narratives of deliberate direction.

The result is a profound asymmetry: Successful missions are celebrated and remembered, while foregone discoveries remain invisible. The historical record therefore creates a persistent tendency to overestimate our capacity to direct scientific progress and underestimate the contribution of decentralized exploration.

A case for epistemic humility

The preceding argument should not be interpreted as a defense of an ivory-tower conception of science. Nothing in the discussion above implies that practical problems are unimportant, that scientists should ignore societal needs, or that public investment in research lacks justification. Nor does it deny Stokes’ central insight that practical concerns can stimulate fundamental advances in understanding.

The issue is not whether useful science exists. The issue is whether the existence of useful science justifies organizing research systems around societally defined missions.

The answer is far from obvious. The intellectual case for mission-oriented science often rests on a series of conflations. Pasteur’s Quadrant is treated as a theory of governance rather than a theory of scientific motivation. Apollo is treated as a model for discovery rather than an engineering project. Successful innovations are treated as evidence for strategic direction while invisible opportunity costs disappear from view. Much contemporary mission-oriented thinking, from Mazzucato’s work onward, relies upon these moves. Yet none of them resolves the fundamental problem identified by Hayek and Polanyi: the knowledge required to direct scientific discovery does not exist until discovery itself has occurred.

Pasteur’s Quadrant demonstrates that practical problems can inspire scientific discovery. It does not demonstrate that administrators, funding agencies, or governments can successfully identify the practical problems most likely to generate future breakthroughs. Historical moonshots demonstrate the possibility of accomplishing specific engineering objectives under particular institutional conditions. They do not demonstrate the feasibility of directing scientific discovery within complex adaptive systems characterized by profound uncertainty.

More fundamentally, the grand-challenges movement underestimates the extent to which scientific progress depends upon decentralized exploration. It assumes a degree of knowledge concerning the origins of future discovery that neither policymakers nor researchers possess. It creates incentives that encourage rhetorical alignment, administrative expansion, intellectual convergence, and systematic overpromising. And it evaluates its own success through methods that naturally exaggerate the apparent efficacy of direction while obscuring the opportunity costs associated with foregone exploration.

The deepest lesson of Smith, Hayek, and Polanyi is not merely that decentralized systems are efficient. It is that they are indispensable when the relevant knowledge cannot be known in advance. Scientific discovery belongs squarely within this category. The most important advances are often those that nobody predicted, nobody planned, and nobody could have justified beforehand.

Again, this is not an argument against practical relevance; it is an argument for epistemic humility. A healthy research system should contain many forms of inquiry. It should contain work inspired by practical problems, work motivated by intellectual curiosity, and work whose eventual significance remains entirely obscure. It should permit researchers to pursue recognized challenges while simultaneously preserving the diversity and autonomy upon which unexpected discovery depends. Most importantly, it should resist the temptation to mistake retrospective narratives of success for evidence that scientific progress can be systematically directed from above.

The central task of science policy is therefore not to identify the correct grand challenges. It is to establish and preserve the institutional conditions under which scientific communities remain capable of generating knowledge that nobody yet knows they need.

The history of science suggests that this is the most socially valuable mission of all!