Scientific Research: Investment Can Be Planned, But Breakthroughs Cannot Be Scheduled — A Discussion Inspired by Two Fields Medals

In July 2026, Wang Hong and Deng Yu were awarded the Fields Medal. In its citation, the International Mathematical Union commended Wang Hong primarily for her significant advancements in harmonic analysis, geometric measure theory, and the three-dimensional Kakeya problem; Deng Yu was recognized for his work in partial differential equations, the rigorous derivation of the Boltzmann equation, and probabilistic methods in nonlinear dynamics [1].

After the news reached China, people naturally began to trace the connections of these two medals to China: where they grew up, what education they received, and at which institutions they completed their key research. Some regarded this as proof of the success of China’s basic education, while others interpreted it as evidence of brain drain and issues within the scientific research system.

A podcast episode from 《记者下班》 discussed the popularity of a feature article from Guangxi Daily titled 《王的猜想》, dividing the public discussion surrounding Wang Hong’s award into multiple dimensions, including media narrative, education system, and academic evaluation. The divergence in the comment section is also noteworthy: some listeners believed that while an individual’s award is worth celebrating, there is no need for it to immediately escalate into a collective frenzy; others argued that people’s embrace of the feature article also signifies support for a reporting style that more respects scientists’ professional identity, expressing an unwillingness to frame a top female mathematician with templates like “small-town girl’s underdog story.”

These comments cannot prove which research system is more effective, but they expose three problems often confused in public discourse: how we celebrate a scientist’s personal achievement, how we evaluate the role of different educational and research institutions, and how we use their experiences to confirm a collective identity. Only by separating these three will personal biographies not be simply rewritten as material to defend or convict a certain system.

However, if the discussion remains on “whose medal it should be attributed to,” we will soon miss a more important question.

China already possesses a large research workforce, continuously growing R&D investment, and globally leading paper output. China can also build large-scale research facilities, organize major engineering projects, and rapidly narrow the gap with the world’s frontier in some fields. Since talent, funding, and organizational capacity are no longer scarce, why does the growth in research scale still not automatically translate into more high-risk, long-cycle, difficult-to-pre-evaluate original research?

This question cannot be answered solely by counting medals. We also need to observe how scientific evaluation shapes researchers’ choices: Is the system willing to support long-term explorations without clear immediate results, or does it mainly reward results that can be quickly published, applied for, and accepted?

Two Medals Cannot Simply Answer “Who Nurtured Them” #

Wang Hong completed her undergraduate studies at Peking University, then pursued studies at École Polytechnique in France, and earned her Ph.D. at MIT. She subsequently worked at various research institutions in the United States and France, collaborating with Joshua Zahl on research into problems such as the three-dimensional Kakeya conjecture [1][2].

Deng Yu was admitted to Peking University but later transferred to MIT, where he received his Bachelor’s degree in Mathematics in 2011, followed by a Ph.D. from Princeton University. Most of the key work for which he was awarded the Fields Medal was also completed at universities and research institutions in the United States [2][3].

The experiences of both individuals neither prove that Chinese education is ineffective, nor do they prove that leaving China automatically makes one a top scientist. China’s education system played a role in early talent identification and foundational training, while universities in France and the United States provided subsequent courses, mentors, peer networks, and career opportunities, and the final research achievements stemmed from their own and their collaborators’ years of exploration.

The podcast episode and its comment section repeatedly featured the judgment that “education excels at selection but not at nurturing.” This touches on the real feelings of many graduate students and young scholars, but still needs more precise articulation. Any education system simultaneously includes selection and nurturing; Chinese schools clearly also laid the foundation for the growth of these two mathematicians’ abilities. The difference is that selection primarily identifies abilities an individual has already demonstrated at a given point, while nurturing requires continuously providing courses, mentors, peers, and space for trial and error when results are not yet apparent.

A system can identify students with outstanding problem-solving abilities through exams and competitions, and can send them to the best universities. But once they enter the research phase, scientific research no longer has standard answers, and researchers even need to question whether existing problems are worth continued investigation. At this point, the abilities identified by early selection are merely a starting point. The system must also decide who can autonomously choose problems, who has time to go through a long period of dormancy without results, and whether one wrong choice will cause a researcher to lose the chance to continue trying.

This is precisely why research achievements cannot be simply defined by nationality and universities. Modern basic research is rarely completed by one person suddenly in isolation. Researchers need to encounter suitable mentors, problems, collaborators, and institutions at different stages, and also need sufficient time to gradually transform immature ideas into results that can be peer-reviewed.

Therefore, the question these two Fields Medals prompt is not “Did China produce geniuses?”, but another question: Can a system, after identifying promising young people, continue to provide them with ten or even more years of growth space?

The experiences of both remind us that an individual’s talent is inextricably linked to the research environment. Here, ’environment’ is not a simple national label, but rather the mentors, peers, funding, career security, evaluation methods, and collaboration networks surrounding the researcher. Cross-border mobility does not directly change one’s intellect, but it may change the problems one can access, the space for error, and how people around respond to an immature idea.

Chinese Research Has Reached Scale, But Scale Is Not Synonymous with Originality #

If we still summarize Chinese research with “China lacks talent, funding, and good papers,” it can no longer explain reality.

Data released by the National Bureau of Statistics show that in 2025, China’s research and experimental development (R&D) expenditure reached 3.9262 trillion yuan, accounting for 2.80% of GDP; basic research expenditure was 277.8 billion yuan, an increase of 11.1% year-on-year, accounting for 7.08% of total R&D expenditure [4]. Neither in absolute scale nor growth rate is this a system lacking in research investment.

Paper quality can no longer be dismissed with “only quantity.” The Nature Index tracks the contribution of countries and institutions to papers in a select group of high-quality journals. In the 2025 ranking, based on 2024 papers, China has already secured the top position in the overall country and territory list [5]. While this metric only covers a portion of journals and cannot represent the entirety of a country’s research quality, it at least indicates that the growth of Chinese research is no longer merely an increase in the number of low-quality papers.

Therefore, a more accurate question today is not “Why does China lack original science?”, but rather: Why does the already established scale advantage still not reliably translate into commensurate high-risk exploration and long-term original achievements?

Research investment first expands system capacity. More funding can build labs, buy equipment, hire researchers, and support more projects simultaneously. Paper count reflects how much publishable research the system produces. These indicators are all important, but they do not directly measure whether researchers have posed new problems, nor do they tell us whether a direction that currently yields no results is worth continued investment.

Scale can be rapidly expanded through budgets and projects, but originality relies on slower accumulation. It depends not only on how many intelligent people the system has, but also on whether these people dare to deviate from popular directions, have opportunities to collaborate with peers from different backgrounds, and whether one failure will directly end their academic careers.

Original Research Relies Not on Isolated Geniuses, But on a Career Path That Can Tolerate Failure #

When people discuss scientific breakthroughs, stories are often compressed into a genius and a decisive moment. But the real scientific research process often involves numerous steps hidden behind award announcements: researchers learning existing work, repeatedly trying immature methods, exchanging unpublished ideas with peers, abandoning dead ends, and redefining problems from unexpected results.

This type of work requires talent, resources, and a supportive collaborative network. Talent determines whether researchers can understand and advance difficult problems; resources determine whether they have time, equipment, and basic living security; collaborative relationships determine whether they can enter truly active knowledge networks and receive timely criticism, methods, and new problems.

The most easily overlooked aspect is who bears the cost of failure.

If young researchers must publish a sufficient number of papers, secure the next project, and pass tenure reviews within two or three years, it will be difficult for them to dedicate their primary time to a problem that may yield no results for years. This does not mean researchers lack ideals. On the contrary, they are simply making rational choices under career constraints: first complete work that can be quantitatively assessed, then consider uncertain directions they truly wish to pursue.

A classic study on research funding compared the “people-centric” funding program of the Howard Hughes Medical Institute (HHMI) with the regular “project-centric” funding of the U.S. National Institutes of Health (NIH). After controlling for variables such as researchers’ prior academic achievements, HHMI-funded scientists exhibited an illuminating “thick-tailed output distribution”: they not only contributed more highly cited breakthrough achievements (among the top 1% in their field), but also published significantly more low-citation or aborted research compared to their NIH-funded counterparts [6].

This result reveals a fundamental cost of original research: if a system only accepts success, it is difficult to truly support exploration. High-risk research is likely to lead to greater breakthroughs precisely because a significant portion of these attempts will ultimately fail.

When Metrics Determine Career Security, Researchers Rationally Reduce Risk #

《庄子·胠箧》 posited a radical judgment: “If the sages are not dead, the great robbers will not cease.” Zhuangzi’s target was not a specific wise person, but rather the paradox that arises when authority, standards, and interests combine. People establish measures, balances, and official seals to maintain order, yet great robbers of states are able to seize these very tools of measurement and trust for themselves.

Scientific evaluation faces a similar risk: papers, projects, and talent titles were originally used to identify academic contributions; when they simultaneously determine funding, positions, team size, and decision-making power, competitors are no longer just vying for honor, but also for the power to define academic value and allocate research resources.

This does not mean the research system should abolish honors and evaluations. Peers still need to identify important achievements, and universities and funding agencies must make choices among limited funding and positions. The problem is not whether to use metrics, but how much judgment they undertake that they cannot actually fulfill. When paper counts, journal rankings, project levels, and talent titles simultaneously determine income, positions, laboratory resources, and next-round application opportunities, they cease to be mere approximate signals of academic quality and become gateways to researchers’ career security.

Researchers then adjust their behavior. They are more willing to choose problems that easily yield phased results, split complete findings into multiple papers (i.e., “salami slicing”), follow popular and well-funded directions, and use language familiar to reviewers and policy documents in their proposals. For individuals, these choices are not necessarily unreasonable; but when most people face the same incentives, the entire system will reduce its diversity of directions.

Project applications naturally require objectives, budgets, and timelines, but exploratory research can usually only describe directions, methods, and possibilities at the application stage, not accurately promise final results. Administrators demand certainty to allocate resources, while researchers must promise scientific discoveries that have not yet occurred in their project proposals. The greater the tension between the two, the more likely researchers are to choose topics with largely known answers that only require completion of work.

What plays a role here is a common organizational mechanism (often called Goodhart’s Law): when a measure becomes a target, it ceases to be a good measure. Paper counts were originally used as an approximate observation of academic output; once positions and resources are directly tied to quantity, researchers will desperately increase the output of “easy-to-publish results,” not necessarily the importance of the research problems.

Talent titles have a similar problem. Titles were originally used to identify researchers who had already achieved results, but if titles further determine team composition, project opportunities, administrative influence, and local resources, they transform from honors into gateways to research resources. Researchers dedicating time to applying for titles does not necessarily mean they are vain, but may mean they cannot continue to organize research without one.

Therefore, the most debatable part of research politicization is not to mock individuals by saying “scientists are like officials,” but to analyze how resources flow along administrative hierarchies and identity labels. When identity is more easily trusted than specific problems, when project levels are more easily recognized by the administrative system than research content, researchers will naturally shift some of their energy from exploring problems to cultivating the visible signals that can be read by the institutionalized evaluation system.

The State Can Plan Research Capacity, But Cannot Schedule Scientific Discoveries by Quarter #

When criticizing “planned research,” it’s easy to go to another extreme: as if true science cannot happen as long as the government sets directions and organizes projects. Neither history nor reality supports such a simplistic conclusion.

Large scientific facilities, long-term data platforms, space engineering, particle physics experiments, and public health research all require sustained government funding and cross-institutional coordination. Many major scientific discoveries also rely on these public research platforms. Markets are usually unwilling to bear basic research with extremely long return cycles and difficult-to-monopolize results, so government investment is not the opposite of original research, but an important condition for its existence.

The state can also define long-term research directions based on energy, health, climate, or security needs. The question is whether managers can distinguish between “building capacity” and “pre-ordering discoveries.” Laboratories can be built on schedule, instruments can be accepted, and data platforms can be inspected, but whether a basic scientific research project can achieve a breakthrough often cannot be written into quarterly goals in advance.

Different types of research also require different organizational approaches. For engineering tasks with clear goals and relatively clear technical routes, the challenge is “execution risk”; centralizing resources and setting milestones through administrative hierarchies can accelerate progress. For exploratory research where answers are unknown, or even how to pose the problems is unclear, the challenge is “cognitive and directional risk”; prematurely unifying routes may exclude valuable anomalous directions. The difference between the two is not whether engineering or science is superior, but that the former’s goals and paths are relatively clear, while the latter’s primary task is precisely to redefine problems amid uncertainty.

A study published in Nature in 2019 analyzed over 65 million papers, patents, and software projects between 1954 and 2014. The researchers found that small teams more often proposed new ideas that disrupted existing paths, while large teams were better at developing, verifying, and expanding existing directions [7]. This is not to say that small teams are necessarily better, but rather that a healthy research system needs to accommodate research by teams of different sizes, with different risk appetites, and diverse organizational methods.

From this perspective, research management does not need to accurately predict every potential direction. More importantly, the system cannot allow one judgment to lead all researchers down the same path. Diversified funding channels, teams of different sizes, and independent peer communities, while inevitably leading to duplicated efforts and failures, can preserve more competitive academic hypotheses.

Major projects with concentrated resources can solve specific tasks, but the research system cannot retain only a few centrally supported routes. Broadly involved small teams, temporarily niche directions, and researcher-initiated problems are the source of vitality for major projects, supplying them with talent, methods, and inspiration for new problems. The two organizational methods need to coexist, rather than one replacing the other.

The Problem Is Not That China Lacks Peer Review, But Whether Formal Rules Can Change Actual Incentives #

Generalizing Chinese and foreign research systems as “China relies on administrative orders, while Europe and America rely on peer autonomy” is also overly simplistic and deviates from reality.

This episode proposed that China, France, and the United States can all reflect on their respective education and research systems through Wang Hong’s experience. This perspective is more valuable than arguing “whose medal belongs to whom.” Wang Hong’s early education, field transition, mathematical training, doctoral research, and subsequent collaborations occurred in different countries, and no single institution can claim sole credit for nurturing her. Cross-national experience also does not directly prove that a particular country possesses the optimal system; it can only help us observe at which stages different environments provided opportunities and at which stages they imposed limitations.

Research systems in Europe, America, and Japan also face publication pressure, short-term contracts, star labs, network connections, and concentrated funding. Peer review is not inherently objective; it may favor mainstream problems and be influenced by prestige and institutional background. In recent years, the global academic community has also faced common crises of precarious positions for young scholars, high application costs, and convergence of research directions.

China is not without peer review and institutional reform. The “Guiding Opinions on Improving the Evaluation Mechanism for Scientific and Technological Achievements” released in 2021 has already proposed that basic research achievements should primarily rely on peer review, encourage international “small peer” review, implement a representative works system, and combine quantitative and qualitative evaluation [8]. Subsequent policies continued to emphasize “breaking the four-onlys” (唯论文、唯职称、唯学历、唯奖项 – ‘only papers, only titles, only degrees, only awards’), classified evaluation, and long-term support. In 2025, the Ministry of Science and Technology also proposed exploring five- to ten-year long-cycle assessments for young scientific and technological talents engaged in basic research, and reducing non-research burdens such as form-filling and reimbursement [9].

These policies indicate that administrators have identified the problems caused by short-term quantitative evaluation. Therefore, the question this article needs to pursue is not “Are there reform documents?”, but whether formal rules can change actual behaviors within universities and research institutes.

If assessment departments still rely on external short-term rankings to demonstrate political achievements, or if universities are still engaged in a funding and resource game with metrics as bargaining chips, university administrators will face a “prisoner’s dilemma”: they must decompose the ranking pressure layer by layer, delegating it to every young researcher. In this situation, the actual internal allocation rules of universities will not truly shift, and a number canceled in an evaluation document will soon be replaced at the grassroots level by another set of numbers that are merely a change of superficiality. If stable positions and research resources remain highly scarce allocations, young researchers will absolutely not be able to actively undertake the career risk of zero output in five years, merely based on a guideline document encouraging long cycles.

Institutional reform will only change research choices if it impacts hiring, reappointment, promotion, salary, and resource allocation. Otherwise, evaluation documents can eliminate one number, but grassroots organizations may still replace it with another, easier-to-calculate number.

The Scientific Research System Will Ultimately Produce the Behaviors It Rewards #

The task facing Chinese research in its next phase is no longer just to continue increasing papers, laboratories, and major projects. The more difficult work is to ensure different types of research receive support commensurate with their intrinsic patterns.

Basic research needs more stable funding not tied to individual projects, so that researchers do not have to expend energy on continuous project applications; young scholars need predictable career cycles to be able to undertake the risk of temporary non-publication; beyond major projects, a large amount of small-scale funding needs to be preserved to allow non-mainstream problems a chance to survive; peer review needs to reduce the influence of titles, institutions, and administrative identity, and also needs to allow reviewers to be tolerant of exploratory directions whose specific outputs cannot currently be predicted.

This also means that basic research cannot simply measure its value by short-term industrial returns. Basic research can certainly drive technology and industry, but from posing a problem to forming general knowledge, and then finding practical applications, often requires crossing multiple research stages and organizations. The system can demand honesty in research processes and transparency in funding use, but it cannot demand that all scientific problems generate economic returns within the same cycle.

Reform cannot simply change all evaluations to “counting representative works.” If positions remain scarce, and resources remain highly concentrated, representative works may quickly become new competitive metrics. What is truly important is to reduce the decisive power of a single evaluation over a researcher’s entire career, so that temporary failures, changes in direction, and late-blooming scientists are no longer institutionally excluded permanently.

As a peer who has experienced research and long participated in reviews, I know that peer review is inevitably flawed, and researchers themselves cannot completely escape reputation, resources, and professional interests. A healthy research ecosystem does not require everyone to be selfless, accurate, and visionary. What it needs to do is allow different institutions, different funding channels, and different peer judgments to correct each other, preventing any single review, title, or policy direction from determining the academic lives of all researchers.

Wang Hong and Deng Yu’s experiences ultimately remind us not that China lacks clever and diligent people. On the contrary, China is already capable of continuously identifying world-class talent and already possesses the resources to support large-scale research activities.

What is truly scarce is a more patient institutional capacity: one that is willing to grant researchers long-term trust and support when research results are not yet apparent, allows multiple directions to compete simultaneously, and accepts that most of these attempts may not succeed.

The state can plan funding, laboratories, and talent programs, and can propose scientific directions worthy of long-term investment. But scientific discoveries rarely happen according to project timelines. What the system can do is not to pre-designate the next breakthrough and its timing, but to allow those questions that temporarily have no answers to remain within the academic community for a sufficiently long time for fermentation and exploration.

References #

[1] International Mathematical Union. Fields Medals 2026. 2026. https://www.mathunion.org/imu-awards/fields-medal/fields-medals-2026

[2] Massachusetts Institute of Technology. Yu Deng ’11 and Hong Wang PhD ’19 awarded Fields Medal. 2026. https://news.mit.edu/2026/mit-alumni-awarded-fields-medal-0728

[3] University of Chicago. UChicago Prof. Yu Deng receives Fields Medal, highest honor in mathematics. 2026. https://news.uchicago.edu/story/uchicago-prof-yu-deng-receives-fields-medal-highest-honor-mathematics

[4] National Bureau of Statistics of China. Statistical Communiqué of the People’s Republic of China on the 2025 National Economic and Social Development. 2026. https://www.stats.gov.cn/english/PressRelease/202602/t20260228_1962661.html

[5] Nature Index. 2025 Research Leaders: Leading countries/territories. 2025. https://www.nature.com/nature-index/research-leaders/2025/country/all/global

[6] Azoulay, P., Graff Zivin, J. S., & Manso, G. Incentives and Creativity: Evidence from the Academic Life Sciences. The RAND Journal of Economics, 42(3), 527–554, 2011. https://doi.org/10.1111/j.1756-2171.2011.00140.x

[7] Wu, L., Wang, D., & Evans, J. A. Large Teams Develop and Small Teams Disrupt Science and Technology. Nature, 566, 378–382, 2019. https://doi.org/10.1038/s41586-019-0941-9

[8] 国务院办公厅. 《关于完善科技成果评价机制的指导意见》. 2021. https://www.most.gov.cn/xxgk/xinxifenlei/fdzdgknr/fgzc/gfxwj/gfxwj2021/202108/t20210804_176223.html

[9] 科学技术部. 《国新办举行“高质量完成‘十四五’规划”系列主题新闻发布会》文字实录. 2025. https://www.most.gov.cn/xwzx/twzb/fbh2025091801/twzbwzsl/202509/t20250918_194727.html