Big Teams Deliver Fast Science in Crises

When society faces urgent crises—from pandemics to terrorist attacks—scientists must respond quickly with impactful research. A new study analyzing over 2 million scientific publications reveals that larger research teams not only produce more influential work but also publish faster during these critical moments, overturning the common belief that big teams are slower due to coordination s. This finding, based on an adversarial collaboration among researchers with differing views on team science, suggests that mass collaboration can be a powerful tool for addressing pressing societal needs, with for how science is organized and funded in future emergencies.

The researchers examined the speed and impact of scientific responses to 48 urgent societal events over two decades, including the COVID-19 pandemic, the release of ChatGPT, and the 9/11 attacks. They measured speed as the number of days between an event's onset and a paper's publication, and impact through citations in scholarly articles, news outlets, and policy documents. Contrary to expectations, larger teams were associated with faster publication times and greater attention across all domains. For example, in a pilot analysis of three high-profile events, each additional co-author was linked to a slight decrease in publication delay—equivalent to about two hours—and increases in citations, with 1.08 more scholarly citations per co-author.

Ology involved connecting bibliometric databases like OpenAlex and Altmetric to track over 250 million publications, 2.6 billion scholarly citations, and millions of mentions in news and policy documents. The team used a large language model, Claude 3.7 Sonnet, to identify urgent events and associated keywords, ensuring each event had at least a two-fold increase in related papers in the three years following its onset. They analyzed data from papers published within three years of each event, applying mixed-effect regression models to account for variations across different crises. To avoid overfitting, they weighted analyses by the inverse frequency of team sizes, giving equal priority to rare large teams and common small ones.

Showed nuanced patterns: while larger teams consistently led to faster and more impactful papers, the benefits followed different curves. For scholarly citations, increases in team size yielded diminishing returns, meaning additional authors provided smaller gains after a point. In contrast, for news citations, policy citations, and speed, the relationship was curvilinear—performance improved up to a threshold before declining. The estimated optimal team sizes were 75 co-authors for news citations, 85 for policy citations, and 49 for speed, though these sizes are rarely reached in practice, as 97% of papers had 15 or fewer authors. Supplemental figures illustrated these trends, with quadratic models fitting best for non-scholarly impact and speed, indicating that while bigger is generally better, there are limits beyond which collaboration becomes less effective.

These have significant for how science responds to future crises, such as AI advancements or disease outbreaks. They suggest that fostering larger collaborations could accelerate valuable insights without sacrificing speed, potentially guiding funders and institutions to support big-team initiatives. However, the study acknowledges limitations, including its focus on bibliometric data rather than underlying mechanisms, and the need for further research to confirm causality. Secondary analyses explored potential confounds like field differences or institutional prestige, but the core patterns held, reinforcing the robustness of across diverse events from the Zika virus epidemic to the Russia-Ukraine war.