The goal of this project was to enhance physical crowdsourcing--the presentation of physical tasks and activities to people who make small contributions toward a collective problem--by designing interactions, algorithms and architectures that produce globally effective behaviors from opportunistic contributions. Our research aims to enable physical crowdsourcing systems to tap into the rich daily physical routines of on-the-go Americans to better transport goods, map the world in exquisite new detail, and accomplish a broad range of tasks at scale in their communities. A core challenge is in bringing together the benefits of opportunistic approaches--which give users the freedom to contribute where they wish but make it difficult to guarantee globally effective solutions--and directed approaches--that prompt users for input based on high-level system goals but can require contributions outside people's existing routines or otherwise require strong incentives. To overcome this challenge, the PIs and team advanced a hybrid approach through building theory, algorithms and a framework to indirectly coordinate individuals who when induced to make small contributions through their existing routines, will jointly achieve complex communal and societal goals, such as surveying the natural world at unprecedented resolutions, performing efficient search over large regions, and producing on-time transport of goods.

Specifically, the project led to four key technological and theoretical developments:

(1) We advanced principles for flexible coordination, or ways to engage people to contribute when it is convenient for them to do so, but that nevertheless are indirectly coordinated to achieve globally effective outcomes.

(2) We developed a new data-collection framework we call 4X---eXplore, eXpand, eXploit and eXterminate---that uses flexible coordination principles to dynamically switch across data collection strategies to collect low-effort contributions by reasoning about volunteers' dynamic changing state of interests and current knowledge about the world.

(3) We developed two decision-theoretic mechanisms for flexible coordination: (A) Opportunistic Hit-or-Wait, which makes dynamic decisions about when to engage a volunteer to make contributions en-route (e.g., searching for a lost item) when it is convenient for them and valuable for the system (e.g., where the item is likely lost); and (B) Opportunistic Supply Management, which makes dynamic decisions about how to engage a community of volunteers in a way that can optimize the desired balance between the experience of volunteers (e.g., avoiding disruption) and the goals of the system (e.g., transporting goods for community members).

(4) Using our understanding of flexible coordination from physical crowdsourcing, we also developed technologies for helping friends and families connect at distance. Unlike existing social technologies that largely promote engaging with posted social content (e.g., via feeds and commenting on posts), our early design and technical work on Opportunistic Collective Experiences (OCEs) promoted active engagement in shared activities with friends and family at a distance during opportune moments in their lives.

Our evaluation work demonstrated the effectiveness of the flexible coordination approach. Specifically, we conducted user studies on 4X, Opportunistic Hit-or-Wait, Opportunistic Supply Management, and OCEs that collective provide evidence for how flexibly engaging individuals to contribute to collective problems or connect with friends and family can support completing communal tasks, gathering useful information, and enhance social connections -- all while minimally disrupting users by weaving tasks and interactions into people's routines, and inline with their interests. For example, the 4X framework created 34% more opportunities for contributing data without increasing disruption; moreover, participating in OCEs led to higher feelings of being a part of a group than viewing posts and replies on social media and helped people connect with friends who they normally wouldn't reach out to or direct message.

These findings suggest that communities and organizations can better achieve desired goals while attending to the needs and interests of volunteers, workers, and citizens by using the flexible coordination type of mechanisms that we pioneered. More broadly, our findings show the promise of using flexible coordination principles for creating entirely new way for intelligent systems to help people manage the complexities of modern life, which often requires us to intelligently address multiple competing goals, in which systems do not prescribe actions but instead surface them within the flow of people's changing activities, availabilities and interests. In other words, they allow individuals and communities to achieve the benefits of coordinating with one another, while at the same time preserving the benefits of flexibility and individual autonomy.

Last Modified: 11/04/2021
Modified by: Haoqi Zhang