Urgent societal events demand scientific responses that are both rapid and impactful. Through an adversarial collaboration, we connected bibliometric databases to evaluate the speed and impact of over 2 million scientific publications in the three years following 48 urgent societal events. A pilot analysis of three cases -- the 2022 release of ChatGPT, the 2019 COVID-19 pandemic, and the 2001 World Trade Center attacks -- yielded unexpected patterns: larger teams were not only more impactful but also quicker to publish. More precisely, increases in team size were associated with (a) initial increases, but eventual diminishing returns in academic citations, (b) curvilinear returns in news and policy document citations, and (c) curvilinear returns in terms of how quickly papers were published. In other words, there are points where further increases in team sizes are either marginally helpful (diminishing returns) or counterproductive (curvilinear returns). To evaluate robustness, we pre-registered a broader test covering 45 additional events spanning two decades.

翻译：紧急社会事件要求科学界做出既快速又有影响力的回应。通过对抗性合作，我们连接了文献计量数据库，以评估48起紧急社会事件发生后三年内超过200万篇科学出版物的速度与影响力。对三个案例的初步分析——2022年ChatGPT发布、2019年COVID-19疫情和2001年世贸中心袭击——揭示了出人意料的模式：较大的团队不仅影响力更大，而且发表速度也更快。更精确地说，团队规模的增加与以下因素相关：（a）学术引用先上升后递减，（b）新闻和政策文件引用呈现曲线回报，以及（c）论文发表速度也呈现曲线回报。换言之，存在一个临界点，超过该点后进一步增加团队规模要么仅产生边际效益（递减回报），要么适得其反（曲线回报）。为评估结果的稳健性，我们预先注册了一项更广泛的测试，涵盖了跨越二十年的另外45起事件。