Summary of Key Points
This article reveals an counterintuitive phenomenon: despite the explosive growth in global research investment, the number of published papers, and the scale of the research community, the proportion of truly groundbreaking achievements that can transform our understanding of the world (such as the discovery of the DNA double helix or the theory of general relativity) has significantly decreased, and the absolute number of such achievements has almost stagnated. The article analyzes four main reasons for this trend: an overwhelming burden of knowledge (it now takes decades to reach the forefront of research), academic aging (older scientists dominate resources and tend to be conservative), conservative academic systems (which favor safe, predictable research over innovation), and the dual-edged nature of AI (while it can increase efficiency, it may also amplify mediocrity). The article compares the situations in China and the United States: Chinese researchers, although younger, have an advantage in conducting follow-up studies but lack breakthroughs from scratch; in contrast, the U.S. has seen a decline in its own research vitality due to a disconnect between academia and industry.
Detailed Analysis
1. The slowdown of science is not an illusion: Groundbreaking achievements are overshadowed by numerous minor improvements
You might think that there are new breakthroughs in technology news every day, but the number of truly groundbreaking discoveries that can rewrite the foundations of knowledge is actually decreasing. How can we measure this? There is a metric called the “CD Disruptiveness Index.” If a paper becomes so influential that subsequent studies no longer reference previous papers (indicating that it has replaced existing consensus), it has high disruptiveness; if previous papers are still cited, it indicates low disruptiveness. Research shows that the average disruptiveness of papers in nearly all disciplines has decreased by 79% to 100% over the past half-century. While some argue that this index is affected by the increasing length of papers, there is consensus that the actual number of major breakthroughs has not changed; it’s just that they are being drowned out by a sea of minor improvements.
2. Two hidden barriers stalling scientific progress: an endless stream of knowledge and reluctant aging scientists
The first reason for the slowdown in science is the burden of knowledge: Scientific knowledge builds like a tower that grows ever taller; in the past, Newton could reach the summit at the age of 20, but today’s researchers must spend 20 years just laying the foundation to get to the forefront. For example, pioneers in quantum mechanics made groundbreaking discoveries in their twenties, while today’s scientists may not even reach the boundaries of their fields until they are in their forties, limiting their ability to pursue more transformative work.
The second reason is academic aging: Older scientists tend to rely on the literature they learned when they were younger, anchoring their research in the past. However, resources (funding, positions, review power) are often in their hands. In the U.S., young biologists must wait until the age of 42 to receive their first significant funding, and 10% of senior professors control 40% of the funds. After the abolition of mandatory retirement requirements, older professors continue to hold onto their positions, trapping the scientific community within the frameworks of their younger days.
3. Academic systems are becoming obstacles to innovation: Preference for safety over risk-taking
Current academic norms favor predictability over innovation:
- Index-driven research: The pressure to publish leads to a frenzy of publications, but new researchers often lack the time to review all the literature and rely on algorithmically recommended “authoritative” papers, preventing new ideas from being noticed.
- Conservative peer review: Reviewers focus on potential risks, rejecting groundbreaking proposals that challenge existing theories for fear of making mistakes.
- The rise of large teams: Large teams, with their resources and ability to experiment, are more likely to produce conventional research that builds upon existing findings rather than exploring new paths.
4. Is AI a savior or a trap? It can help, but it may also make science more mediocre?
AI can indeed solve complex problems—AlphaFold, for example, predicted the structures of almost all known proteins in just three years, accelerating the awarding of Nobel Prizes in chemistry. However, AI has its limitations:
- Limited creativity: AI learns from existing literature and can optimize existing knowledge but struggles to generate entirely new ideas.
- Amplification of mediocrity: AI tends to highlight mainstream viewpoints, filtering out unconventional insights that could lead to innovation.
For instance, a MIT paper claimed that AI increased the number of material discoveries, only for the data to be later proven to be fabricated. This case highlights how AI can create misleading results when it tries to fit predetermined outcomes.
5. The contrast between China and the U.S.: Young Chinese researchers have an advantage, but lack breakthroughs; the U.S. suffers from a self-imposed disconnect
China’s strength lies in its young research community, which has made rapid progress in fields that require both energy and quick adaptation (such as materials science, where its natural index outpaces that of the U.S.). However, China still lacks major breakthroughs due to an academic system that favors follow-up studies over original discoveries.
The U.S., on the other hand, has seen a decline in the citation rate of its biomedical research after cutting off cooperation with Chinese scientists, resulting in a 10.5% decrease in citations. This self-inflicted setback highlights the consequences of isolating itself from international collaboration.
Conclusion
Science does not always move at a constant pace; it needs periods of “hibernation” to consolidate knowledge and periods of “explosion” for breakthroughs. The current slowdown is not due to a decline in human intelligence but rather due to systems and rules that stifle innovation. To revitalize science, we need to support young researchers, reform retirement policies, promote open-source scientific AI, and fund smaller teams that are more willing to take risks. After all, it is often the young, with nothing to lose and a desire for disruption, who push the boundaries of knowledge forward.