T ackling global challenges such as food insecurity, or advancing complex technologies like quantum computers, requires collaboration. ‘Team science’ may involve two researchers in the same depart- ment, or thousands across the globe: teams of teams, such as those at the Large Hadron Col- lider at CERN, Europe’s particle-physics labo- ratory near Geneva, Switzerland. To develop techniques for 3D printing of human tissues, say, researchers must integrate life science and material science with electrical and mechani- cal engineering; transcending such disparate disciplines complicates collaboration. As the complexity of team science increases, so does demand for sophisticated skills, strategies and resources. Yet currently, although it is relatively common to find sci- entific structures and norms suited to small, single-discipline research teams, support for more-complex teams remains inadequate. In 2006, a new cross-disciplinary field was launched: the ‘science of team science’ (SciTS). Its aim is to build an evidence base to help administrators, funders, researchers and others determine the best ways to struc- ture and support scientific teams and improve their effectiveness. The field examines the impacts of, for example, science policies, organizational structures, technological tools, team management and individual competen- cies on the success of science teams. In The Strength in Numbers, science-policy special- ists Barry Bozeman and Jan Youtie delve into one aspect of SciTS: managing teams. Drawing heavily from a survey of 641 researchers, interviews with 60 faculty members and web posts from 93 anonymous contributors, the authors focus on conflicts in relatively small teams and co-author relationships. They classify collabora- tions into four types — dream, routinely good, routinely bad and nightmare — and offer advice for addressing factors such as working style, career stage and trust. For many, The Strength in Numbers might come across as a missed opportunity. Bozeman and Youtie state that research on collaborative teams has become fragmented, or “balkanized”, yet they risk fuelling such divisions by citing literature from leading SciTS scholars in just a handful of paragraphs. The authors intermittently mischaracterize and dismiss existing SciTS research and resources such as the Team Science Toolkit, instead of considering how those might bolster their “prescriptions”. Much of their advice is either overly specific or vague. To one researcher, they recommend: “Get through the project the best you can, and then do not work with the senior colleague again.” Meanwhile, they tout their newly devel- oped “Consultative Collaboration” strategy as the primary answer to the complexities of team science. All team members, they argue, should be consulted at key points in a col- laboration to pin down values and choose the next steps. Yet fewer than a dozen pages are devoted to discussing the approach, and only a handful include explanations of how to use it. This leaves the reader to ponder what strategies such as “effective communication, not constant communication” actually mean. Consulting the decades of existing literature on the science of management, leadership or teams would have provided detail and depth. RESEARCH MANAGEMENT What makes teams tick Kara L. Hall examines a study of current research on scientific collaboration. level, there is Reykjavik’s success in using new technology to extend the reach of its municipal lawmaking insti- tutions by enabling citizens to suggest, and vote on, initiatives. Surprisingly, Mulgan devotes an optimistic chap- ter to ways of improving how we run meetings. He calls for smaller meetings that promote a shared understand- ing of their purpose through clearer agendas, allocation of defined tasks, well-stated goals, and better use of space, moderation and gadgets. Equally surprising is his ultimately dour and dispiriting assessment of the limits of collective intelligence for improving parliaments and legislatures at scale. He overestimates the success of new plat- forms for generating ideas, which, over time, have not led to much in the way of outcomes and have only increased frustra- tion with democratic institutions. At the same time, he potentially underestimates emerging models for “crowdlaw” — that is, those online processes for engaging broader publics in making decisions and evaluating their impact. Mulgan points to examples of complex and large-scale political col- laborations, such as the Paris Climate Accord and the 2015 ratification, by 193 countries, of the 17 global Sus- tainable Development Goals. And he rightly concludes that the jury is still out on the question of which processes or technologies could sustain new forms of collective public governing. Inspired by this question, the fifth annual Collective Intelligence Confer- ence, held in June in New York City, focused on democracy. Experts from computer science to the social sci- ences came together to examine what democratic institutions need to do to better tap the intelligence and exper- tise of those they govern. As Mulgan concludes, answering this question is hampered by a stark fact. Although parliaments fund and universities con- duct research, neither invest much in ways to improve how institutions actu- ally mobilize collective intelligence. Despite the advent of the Internet, these bodies look the same as they did a generation ago. The trenchant ques- tions and thoughtful discussion in Big Mind, however, will help us to reimag- ine our institutions and convince us of the urgency of doing so. ■ Beth Simone Noveck is the Jerry Hultin Global Network Professor at New York University’s Tandon School of Engineering and director of the Governance Lab. e-mail: noveck@thegovlab.org Researchers collaborate at the ATLAS experiment at CERN. CERN The Strength in Numbers: The New Science of Team Science BARRY BOZEMAN & JAN YOUTIE Princeton University Press: 2017. 562 | NATURE | VOL 551 | 30 NOVEMBER 2017 BOOKS & ARTS COMMENT ©2017MacmillanPublishersLimited,partofSpringerNature.Allrightsreserved.