There is a tendency to think of science as a production system: resources are invested, experiments are conducted, and those experiments are expected to produce useful outcomes. The better the system, the more efficiently it should convert investment into useful products. Under this model, an unsuccessful experiment is a failed investment, and a scientist who cannot articulate the eventual application of their work is difficult to justify funding.
But science was never meant to function as an enterprise for producing predictable returns. At its core, science is a formalized way of exploring the unknown.
Exploration is inherently uncertain. It is costly, resource-intensive, and often futile.
Most attempts will not produce anything immediately useful. But exploration has an asymmetric payoff: a small number of discoveries can generate benefits vastly greater than the resources invested in the many attempts that fail. The possibility of those discoveries—and the knowledge they add to civilization—justifies maintaining a capacity for exploration despite its uncertainty. You cannot know in advance which questions will matter, where an unexpected observation will lead, or what value a discovery may hold in the future. This is not a flaw in science; it is the nature of exploration.
Part I — The Value of Science to Society
Science creates value in ways that extend well beyond the outcomes we can predict in advance. That value takes several distinct forms.
1. Instrumental value
Scientific discovery has enabled more efficient agriculture, reduced food waste, improved housing and infrastructure, and transformed the prevention, diagnosis, and treatment of disease. It has expanded our capacity to improve the material conditions of human life in ways that would have been unimaginable to previous generations.
This is the form of scientific value that society tends to recognize and encourage most readily because its outcomes are comparatively tangible. A new therapy can be evaluated by the number of patients it reaches. A new technology can be measured by the efficiency or productivity gains it generates. A discovery can be translated into a patent, a company, or an entirely new industry. These outcomes are visible, quantifiable, and relatively easy to translate into the language of policy and economics.
But visibility is not the same as true value.
An invention that solves an immediate problem is not necessarily beneficial in the broader or longer-term sense. Simply put, an improvement in the short term may create consequences tomorrow that we did not anticipate. Antibiotics, for example, transformed the treatment of infectious disease while simultaneously creating evolutionary pressure for antimicrobial resistance. AI illustrates the same broader principle: technologies developed to increase productivity, expand access to information, and automate difficult tasks may also introduce forms of misinformation, dependency, concentration of power, and social disruption that were difficult to anticipate at the outset.
Neither example makes the underlying scientific or technological advance a mistake. Rather, they illustrate how difficult it is to assess the ultimate value of an intervention when its consequences unfold across complex systems and over long periods of time.
The question of whether a scientific advance is ultimately good for humanity is therefore considerably more complicated than we often acknowledge.
2. Epistemic value
Science also produces something more fundamental: knowledge about the world.
We learn something that we did not previously know. That knowledge may have no obvious application when it is discovered, but that does not mean its eventual value can be assessed as zero.
Scientific knowledge is cumulative and interconnected. A seemingly obscure observation can become the missing piece that makes a later discovery possible. A biological mechanism that appears irrelevant to human disease may become important when another researcher discovers how it interacts with a different pathway. A mathematical principle developed without any practical purpose may eventually become foundational to a technology that did not yet exist.
This is one of the fundamental difficulties of evaluating scientific knowledge by its immediate utility: we rarely know which pieces of knowledge will become useful until we have accumulated enough of them to see the connections.
Science therefore serves not only to solve the problems we currently recognize, but to expand the body of knowledge from which future questions and solutions can emerge.
There is also a deeper reason to value the pursuit of knowledge. Exploration is one of the ways civilization has learned to confront the fundamental uncertainty of existence. We want to know where we came from, how life emerged, whether we are alone in the universe, and what governs the world around us. Science does not necessarily provide comforting answers to these questions, but it gives us a way to ask them rigorously and replace some measure of uncertainty with understanding.
In this sense, the pursuit of knowledge serves a purpose beyond its eventual applications. It helps us make sense of the world we inhabit. A civilization that can investigate its origins, understand its place in the universe, and continually expand the boundaries of what it knows is better equipped to confront the uncertainty inherent in existence.
Scientific knowledge has epistemic value in its own right: there is value in understanding the world more accurately, even when that understanding has no eventual practical application.
3. Aesthetic value
There is a third form of value that is harder to quantify: aesthetic value.
Scientific training teaches us to observe—to look carefully, to notice patterns and differences, and to remain attentive to phenomena that might otherwise seem ordinary. It teaches us to approach the world with curiosity and attention, and to recognize complexity and beauty where we might previously have seen only the familiar.
A microscope can transform a drop of water into an ecosystem. Evolution can transform a forest from a collection of trees into the product of billions of years of history. Astronomy can transform the night sky from a scattering of lights into a universe containing billions of galaxies.
In this sense, science is not merely a way of understanding or controlling the world. It is a way of experiencing it. And there is value in that experience. Human beings spend much of their lives moving through a world that can easily become familiar and unremarkable. Scientific observation offers a way of continually rediscovering it.
Perhaps this is one reason the pursuit and teaching of science matter even when they produce no immediate practical benefit. To teach someone how to observe the world is, in a small way, to give them more of the world to experience.
In a world that can easily become familiar and unremarkable, learning to see more closely allows us to keep finding beauty in things we might otherwise overlook.
If scientific exploration is inherently uncertain, then the central question becomes not how to predict which research will succeed, but how to identify the people best equipped to explore what cannot yet be predicted.
Part II — Who Should Perform Science?
Scientific exploration is expensive. It consumes public and private money, specialized infrastructure, equipment, biological and material resources, institutional capacity, and enormous amounts of human time. Every scientific project therefore carries an opportunity cost.
If exploration is costly, then society has a responsibility to be selective about who is entrusted with it.
What should we select for? Not simply intelligence, technical competence, nor the ability to publish.
A good scientific explorer needs the following:
Curiosity: the ability to notice interesting questions where others see routine observations.
Independence: the ability to think for oneself, resist intellectual conformity, and pursue a question even when it runs against prevailing opinion.
Judgment: the ability to distinguish an interesting result from noise, a promising question from a trivial one, and a productive failure from wasted effort.
Tolerance for uncertainty: the ability to spend years pursuing a question without knowing whether it will lead anywhere.
Imagination: the ability to recognize that the most important discovery may come from asking a different question altogether.
These qualities are difficult to measure using conventional academic metrics. Publication counts measure output. Citation counts measure influence. Grant dollars measure funding success. Patents measure commercialization. None directly measures the ability to explore.
More importantly, these qualities are rare. Many scientists are capable of conducting rigorous experiments; far fewer can identify important questions, pursue them independently, and generate genuinely unexpected insights.
This creates a problem when the number of people pursuing independent research grows faster than the number of exceptional scientific explorers. Not everyone who is capable of doing scientific work will necessarily be capable of leading genuinely exploratory research. Yet our academic system often treats these capacities as interchangeable. Researchers are evaluated through metrics such as publications, citations, grants, and demonstrated impact because these are easier to measure than curiosity, judgment, or originality. Scientists then rationally optimize for what the system rewards: questions that can generate papers, projects with predictable endpoints, and research whose potential impact can be articulated before the work has even begun.
The problem, therefore, is not that too many people participate in science. It is that we have blurred the distinction between participating in scientific research and being an independent scientist. Not everyone who can conduct good science should necessarily be expected to lead an independent research program. Independence should be reserved for those who demonstrate the qualities of a true scientific explorer.
This does not mean that independent scientists should be expected to produce predictable outcomes. They should, however, be held to a high standard of intellectual honesty, rigor, and judgment even when the outcomes of their work are uncertain. Failure is inevitable. A failed hypothesis is not necessarily a failed scientist; an experiment that produces no useful result may still eliminate an important possibility. What matters is distinguishing productive uncertainty from poor scientific judgment.
We cannot know in advance which expedition will succeed. We can, however, become better at choosing the explorers.