Collaborative Research: Understanding Team Success and Failure

现代科学技术中最根本、最普遍的变化之一是，随着孤独的研究人员的消失，科学、学术、发明和创业等各个领域的团队都在蓬勃发展。团队构成了推动新发展的社会引擎，在科学和技术方面的主导地位日益增强。然而，由于绝大多数的研究仅仅是基于对成功团队的观察和分析，我们对团队成功和失败的过程知之甚少。例如，大多数团队研究被限制在最初成功组建的团队中，结果是联合发表或专利。在现实中，大多数团队都失败了，有时会以一种惊人的方式失败。这一事实表明，我们目前对团队的理解受到系统性选择偏见的影响，失败的团队在很大程度上被忽视了，因为追踪它们的数据要少得多。先前的调查记录了团队给其成员带来的职业优势，但没有记录他们如何影响科学发现和技术发明。在这里，通过分析在许多领域和结果度量中与成功的团队一起失败的团队，该项目将揭示关于团队失败的原因、方式和时间的经验基础见解。如果不分析导致团队失败的诸多原因，研究人员仍然无法确定与成功相关的有力因素。该项目考察了广泛的科学和技术相关背景下的团队成功和失败，从生物、社会和自然科学、学术到技术、软件和创业。团队可以是大的或小的，或多或少的结构集成，并且涉及不同成员角色的组合或先前经验的混合。该项目包括两个阶段的研究计划，旨在了解不同规模和形状的成功团队如何“以不同的方式思考”，以及如何设计以加速科学和技术发展。首先，该项目从团队规模、网络结构、角色构成和经验等方面评估了100多年来超过1亿个研发团队的成功和失败结果。其次，从这项调查中获得的见解将使大规模的在线团队实验能够启动，以分离驱动这些影响的因果机制。这些实验将为关键的团队机制带来确定性，并促进政策建议，这些建议可用于设计团队，以优化推进科学和技术的特定目的。总的来说，这项研究有望显著提高我们追踪、评估、预测、培养和设计高影响力和高破坏性团队的能力。具体来说，我们的项目首先涉及(1)大量数据清理和来自科学、发明和创业等各个领域的团队数据之间的链接。然后(2)团队的成功和失败是在许多阶段衡量的，包括未能获得资金，发表论文和申请专利，为前沿注入新颖性，吸引科学和技术关注，保持对复制的稳健，以及实现持续的影响力。接下来，研究(3)通过考察团队规模、复杂性、角色结构和团队经验多样性，分析了团队规模和复杂性对成功和失败的影响。该项目利用该调查的见解(4)部署在线团队实验，以因果关系确定团队特征对成功和失败结果的影响。最后，根据期望的科技成果，推荐最优团队，以及最优团队变更或调整。这项工作的结果可以影响全球科学和技术政策，提高人们对不同类型的团队——小型和大型、简单和复杂、多样化和相似——相对于它们所支持的科学和技术成果（包括破坏和集体进步）所带来的好处的认识。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

One of the most fundamental and universal shifts in modern science and technology is the flourishing of teams in all areas of science, scholarship, invention and entrepreneurship as solitary researchers vanish. Teams constitute the social engines that drive new developments with an increasing dominance in science and technology. Nevertheless, little is known about the process through which teams succeed and fail as the vast majority of studies are based on observation and analysis of successful teams alone. For example, most team research is restricted to teams that successfully formed in the first place, resulting in a joint publication or patent. In reality, most teams fail, sometimes in a spectacular manner. This fact suggests that our current understanding of teams suffers from systematic selection bias where failed teams have largely been ignored because the data that trace them are much less abundant. Prior investigations have documented the career advantages teams confer on their members, but not how they influence scientific discovery and technological invention. Here by analyzing teams that fail alongside their successful counterparts across many domains and outcome metrics, this project will uncover empirically-grounded insights regarding why, how, and when teams fail. Without analyzing the many ways in which teams fail, researchers remain unable to identify robust factors associated with success. This project examines team success and failure across a broad array of science and technology-related contexts, ranging from biological, social and natural science and scholarship to technology, software, and entrepreneurship. Teams can be large or small, more or less structurally integrated, and involve distinct combinations of member roles or mixtures of prior experience. The project involves a two-stage research program to understand how successful teams of different sizes and shapes "think differently" and can be designed to accelerate scientific and technological development. First, the project evaluates success and failure outcomes for than 100 million R&D teams over 100 years in terms of team size, network structure, role composition and experience. Second, insights developed from this investigation will enable the launch of large-scale online team experiments to isolate the causal mechanisms driving these effects. These experiments will bring certainty about critical team mechanisms and facilitate recommendations for policy that can be used to design teams optimized for specific purposes in advancing science and technology. Overall, this research promises to dramatically improve our ability to trace, assess, predict, nurture and design high-impact and highly disruptive teams. Specifically, our project first involves (1) massive data cleaning and linkage between data on teams from a variety of domains in science, invention, and entrepreneurship. Then (2) team success and failure is measured at many stages, including the failure to secure funding, publish papers and prosecute patents, inject the frontier with novelty, attract scientific and technical attention, remain robust to replication, and achieve persistent influence. Next, the research (3) analyzes the impact of team size and complexity on success and failure by examining the size, complexity, role structure and diversity of experiences within teams. The project uses insights from this investigation to (4) deploy online team experiments to causally identify the influence of team characteristics on success and failure outcomes. Finally, (5) optimal teams are recommended, as also optimal team alterations or adjustments based on desired science and technology outcomes. Results from this work could influence global science and technology policy by increasing appreciation for the benefits of distinct types of teams -- small and large, simple and complex, diverse and similar -- relative to the science and technology outcomes they support, including disruption and collective advance.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.