During the pandemic crisis of 2020–2023, it is likely that the successes of biomedical science were more than offset by its failures. These failures continue to undermine once-powerful ideals and hopes for science's role in societal betterment. Here, we dissect the underpinnings of these failures and argue that restoring such ideals first requires systemic reform of science itself. The considerable successes of medicine during the COVID-19 pandemic pertained to speed, for example, real-time information sharing1, record-time development of vaccines2, and unprecedented rapidity in getting results from some large randomized trials of interventions3, solidly documenting or refuting claims of effectiveness4. On this basis, the visible role of science and scientists in the pandemic response seems worth celebrating. Basic research, translational research and evidence-based research tools all had some bright days during the crisis. But success was accompanied by major failures. Some of these failures reflected efforts by influential scientists and their political allies to demonize dissenting scientific views and evidence. From the earliest days of the pandemic, despite weak, absent or contradictory evidence, leading national decision makers, facing the urgent need for action5, assured the public that they were adopting COVID-19 policies by ‘following the science’6. A new scientific elite of media-savvy experts from fields like virology and modelling (with or without relevant credentials) gained visibility and influence in political circles. The narrow, reductionist disciplinary and sometimes partisan perspectives7, 8 of these scientists and ‘influencers’ helped justify the COVID-19 policies that allocated sacrifice, privation and suffering across all walks of society and hundreds of millions of lives. The consequences for education, healthcare, small businesses, social life and democratic politics itself were often disastrous. Consequences were most acutely felt by vulnerable populations, from factory workers to school children, the old and the poor9. Radically different responses to the disease from nation to nation—from draconian lockdowns across all sectors, to relatively permissive and flexible pandemic regimes—made obvious to all that the value of scientific evidence was to support what was politically desirable and possible in different contexts. Rather than politics following the science, science was enlisted to follow the politics. Supposedly science-based responses to COVID-19 in turn built upon and amplified deeper contradictions at the intersections of science and public health. Some affluent nations with generously funded science, such as the United States and United Kingdom, had been experiencing slow-downs and even reversals of gains in life expectancy that predate COVID-1910, 11. These trends, unthinkable 25 years ago, both reflect and exacerbate pre-existing and entrenched health inequities. For example, life expectancy for men in London's affluent neighbourhoods is 18 years longer than for those in poor areas11. Such disparities have multiple causes, but the political economy of medical science is a critical part of the story12. Investments in health sciences focus overwhelmingly on reductionist, molecular-genetic approaches to cures, which can generate status for scientists and profits for companies, but have minimal benefits for population health and may even widen inequalities. Medical science delivered to society through market mechanisms continually inflates healthcare costs, another driver of inequities. Additional tensions at the science-health interface include misuse of direct products of science itself, such as opioids, and iatrogenic disorders. Indeed, during the lockdown, iatrogenic disease may have acquired new, unprecedented dimensions13. Yet study and communication of these disorders during COVID was too often demonized by high-level decision-makers and their scientific advisors, which in turn clouded our ability to openly debate the balance of impacts of the massive public health interventions imposed across large segments of the society. Now add the uncertain but plausible possibility that COVID-19 was itself the product of biomedical science14, 15, and an apparently outrageous consequence becomes inescapable: science itself may have become a threat to overall population health. Under these ambiguous and contradictory circumstances, the COVID-19 response could feed into a rising tide of public scepticism about the authority of experts in public affairs, and the role of science in societal betterment. Indeed, given the uneasy possibilities and accumulating tensions we have mentioned, it is hardly irrational that an increasing proportion of the dissatisfied public is wondering whether truth and the path to a healthier world must lie somewhere else than within prestigious journals and celebrities at the science–policy–communication interface16. At a time when the monopoly of science on truth has never been more contested17, the pretence that ‘the’ scientific evidence can be deployed through political processes to justify political decisions in the public interest can only make this contestation worse. This is not a ‘relativist’ perspective, but a hard lesson from COVID-19. Contradictions in the public health enterprise, uncertainties in our knowledge, limits to what science can accomplish, and disagreement about what actions to take, are real and unavoidable. In the absence of strong evidence and, even worse, with inhibition of efforts to get better evidence (with the justification that ‘the science is settled’ or that accurate but unfavourable evidence ‘can harm the public health response’), coupled with unwillingness to openly acknowledge obvious uncertainties, even nonscientists can clearly sense that there is something fundamentally wrong with both science and the public health response. The public, through its behaviour, collectively responds to this reality. More experts and less politics are not an answer. Entanglement of politics and of science happened in various degrees in most, if not all, countries. It resulted in various extreme approaches to COVID-19, such as China's prolonged zero COVID strategy that eventually had to be abandoned as unsustainable. Lack of transparency makes it even difficult to fathom the exact depth of these failures, for example, in China there is enormous uncertainty on the eventual level of COVID-19 and other excess deaths18. Moreover, even in more openly democratic states, it is unclear how far-reaching into the future the negative impact of extreme measures may last on frontiers such as education, mental health and chronic disease outcomes19-21. The COVID-19 experience makes clear that today's medical science enterprise is not an autonomous, self-correcting global public good, but a component and product of global, national and regional political economies within which science and societal needs may be poorly aligned22-27. Commodification of science, predicted half a century ago28, 29, has reached unprecedented levels30. Systemic problems within this industrial-scale enterprise are widely acknowledged and include insularity of science disciplines with lack of system thinking31 and integration32; the proliferation of biased, conflicted and low-quality medical research divorced from societal needs33; the large research waste in an enterprise driven by measures of productivity rather than accountability34; and the lack of truly open discussions about how society funds the science it needs35. Can the systemic science–policy–public rift that became so acutely visible during COVID-19 be healed? To pursue this goal, democracies will need to embrace, over the long term, a new, different political economy of science, one that moves away from the current commodification of incentives and objectives29, 36, 37, acknowledges tensions between market incentives and public values, and accepts the limits of specialized expertise—and especially the danger of new hordes of influencers made possible by the new media38—for guiding society through challenges such as COVID-19. A necessary first step in this healing process would be for scientists and science advocates to recognize our own complicity in having supported the emergence of these problems, and in so doing to embrace the need for system redesign. Absent such commitments, we should expect a continued balkanization of science, expertise and truth, and will be left to hope that societies can discover sources of cohesion and coherence outside of the remit of science. The authors declare no conflicts of interest. John P. A. Ioannidis and Joachim P. Sturmberg are members of the editorial board of the journal but had no role in the evaluation of the paper. There are no data attached to this contribution.