RI: Small: Theory and Application of Mechanism Design for Team Formation

在一个任务复杂的世界里，团队合作是常态，而不是例外。因此，在几乎任何组织以及广泛的领域，如大型软件项目、课程项目、军事行动、建筑项目和搜救任务，组建有效的团队都是至关重要的。然而，成功的团队组建取决于了解团队中的每个人如何在服务于一项任务时相互补充，他们的合作能力，以及对不同任务可能实现的目标的潜在错位偏好。该项目的目标是开发一种通用的团队队形交换服务，利用新的理论基础来设计当未来队友的能力和偏好不确定时的团队组建机制。这种交流将显著提高各组织动态组建灵活高效的项目团队的能力，减少冲突，更好地协调个人和组织的激励。本项目在为一系列目标服务的团队组建的背景下，考虑了一个极具挑战性的机制设计的概念性和技术性问题。核心的概念挑战在于对问题进行足够详细的建模，以捕获诸如任务实现子目标、代理能力、任务要求以及代理对目标和团队成员的偏好等考虑因素，但要避免不必要的细节以保持易管理性。建议的研究提供了一个新的建模框架，旨在平衡这两个方面的考虑。技术贡献既包括对问题的新的理论处理，如描述来自相关领域的机制的新扩展，包括组合分配，针对手头的问题量身定做的全新机制，以及利用特定问题结构的最优可扩展自动化机制设计的算法贡献。此外，拟议的研究将把实验经济学研究与理论和计算经济学的贡献结合起来，将人体主体实验作为评估所开发机制的手段之一。

In a world where complexity of tasks abounds, teamwork is the norm rather than the exception. Formation of effective teams is therefore vital in nearly any organization, and across a broad array of domains, such as large-scale software projects, course projects, military operations, construction projects, and search-and-rescue tasks. Successful team formation, however, hinges on having knowledge about how individuals working on a team complement each other in service of a task, their ability to work together, as well as potentially misaligned preferences about goals that the different tasks may achieve. The goal of this project is to develop a general-purpose team formation exchange service, leveraging new theoretical foundations for designing team formation mechanisms when the abilities and preferences of prospective teammates are uncertain. Such an exchange would significantly improve the ability of organizations to dynamically form agile and efficient project teams, reducing conflict and better aligning individual and organizational incentives. This project considers an exceptionally challenging conceptual and technical problem of mechanism design in the context of team formation in service of a set of goals. The core conceptual challenge lies in modeling the problem in sufficient detail to capture considerations such as achievement of sub-goals by tasks, agent capabilities, task requirements, and agent preferences about goals and teammates, but avoiding unnecessary detail to maintain tractability. The proposed research offers a novel modeling framework aimed at balancing these two considerations. The technical contributions involve both novel theoretical treatment of the problem, such as characterizations of novel extensions of mechanisms from related domains, including combinatorial assignment, entirely new mechanisms tailored to the problem at hand, and algorithmic contributions to optimal scalable automated mechanism design leveraging specific problem structure. Additionally, the proposed research will blend experimental economics research with contributions in theoretical and computational economics by using human subject experiments as one of the means to evaluate the developed mechanisms.