Pei Gang is one of China’s top life scientists and chairman of the Chinese Academy of Sciences’ research ethics committee. Throughout his career, Pei has shed light on how cells communicate with each other. But in 2019 he was accused of fraud over an experiment he conducted more than two decades ago. His accuser, Rao Yi, president of Capital Medical University, said he could not reproduce Pei’s results. The claims and counterclaims have drawn in members of the scientific establishment and gripped not only the research community but the public as well. The case centres on an experiment Pei carried out in 1999 on the function of a cell membrane protein. Pei published his findings in the peer-reviewed journal, Proceedings of the National Academy of Sciences of the United States of America. Then last year Rao accused Pei of cheating because he could not replicate Pei’s results. Pei promptly took Rao to court for slander and a decision is pending. In the meantime, the academy rejected Rao’s accusation, with an expert panel led by Professor Zhong Nanshan, the nation’s top science adviser on pandemic control, saying Pei’s team had successfully repeated some “crucial steps” of the experiment. China to crack down on fraud in scandal-hit scientific research amid ZTE wrangle That might have been the end of the matter but Rao requested another investigation involving Western scientists, including some Nobel Prize winners. The request was immediately denied but to the authorities’ surprise, the public sided with Rao. On social media platforms such as Weibo and Zhihu, many users hailed Rao as a hero. And in a rare move, some state-controlled newspapers ran opinion pieces questioning the credibility of the academy, the nation’s largest and most prestigious research organisation. Some researchers and science policy experts saythe Rao vs Pei case reflects the “reproducibility crisis” in the country. China has led the world in research output in recent years, thanks to huge investments in research and development. But most of these “discoveries” have never been verified, and there are growing doubts in government, the research community and the general public about their scientific or practical value. That is in part because of the high hopes pinned on scientific endeavours. From government officials to members of the public, people have looked to the nation’s scientists to come up with innovations that will enable China to catch up or even overtake the West, especially the United States, in a wide range of cutting-edge technologies. In 2019, China spent US$322 billionon research and development, second only to the US. But the reputation of Chinese scientists has been blighted by a series of scandals, from a massive retraction of medical papers by Chinese doctors to illegal modification of babies’ genes. Nevertheless, the reproducibility problem is not limited to China. In an international survey by journal Nature in 2016, more than 70 per cent of the respondents said they had failed to reproduce the results of another scientist’s experiment. More than half could not even reproduce their own work. It is most apparent in fields such as psychology and medicine. California-based drug company Amgen Oncology, for instance, was only able to replicate 11 per cent of the studies that it looked into. Tim Errington, who has been leading an effort to replicate high-impact cancer papers at the Centre for Open Science in Virginia, said fraud and irreproducibility were not the same thing. Equipment and materials might vary from one laboratory to another, and some experiments required extra skills, experience or knowledge, he said. “There are a number of other reasons the findings don’t replicate, fraud is just one of them and the most troubling,” Errington said. “I’m not sure conducting a replication of the experiments will necessarily confirm or deny a claim of fraud.” China’s gene-editing ‘Frankenstein’ jailed for three years in modified baby case But, in China, there is less patience for scientists who cannot repeat their results. Duan Weiwen, director of the department of philosophy of science and technology in the Chinese Academy of Social Sciences, said Rao’s “truth-seeking spirit” should be promoted and protected. “People are not satisfied with the research quality and the credibility of some senior scientists,” Duan said in an article on China Science Daily last week. “Even people who disagree with Rao on this case have to admit we cannot achieve true independence in science or technology without fighting misconduct and fraud.” Investigating allegations can also be expensive. The review of Pei’s study, for instance, involved six central government ministries and many experts with different backgrounds. They spent almost a year talking to the researchers involved while looking for data records in dusty archives. The final report was 5,000 pages long, according to the academy. “There were many difficulties and compromises with this report ... [the panel] cannot become a legal mechanism like a grand jury. Research ethics must rely primarily on the researchers themselves doing the work,” Duan said. Retired palaeontologist Wei Qi said the public was angry in part because there was no transparency in the investigation. The public suspected the panel was controlled by an “academic aristocracy” with a small circle of senior scientists making the call. Despite spending a lot of time and resources, the academy obviously failed to deliver a conclusion that society would accept, Wei said. “We need more democracy on academic affairs,” he said. Errington said that even without misconduct or fraud, it was important to ask why mistakes occurred so often in one’s line of research. “Being fast and sloppy in conducting and reporting our research does not advance progress, it does the opposite, it slows research down,” Errington said. One of the main drivers of the rush is the priority journals, funders and institutions give to novelty and volume instead of rigour and reproducibility. That means researchers have a pressure to publish quickly and often, creating uncorrected, and in some cases unidentified, mistakes. “And sadly, it also encourages some researchers [to take] short cuts,” he said.