Engaging the Commons in Participatory Sensing
Practice, problems, and promise of participatory sensing in the context of dockless bikesharing.
How do we collect data from our surroundings? Advances in technology have enabled people to deploy auto-sensor networks to solve problems such as the habitat protection project in Malawi. However, the vast resources needed for deploying such large-scale sensing networks have hindered the popularity of auto-sensing projects.
Is there a convenient, low-cost solution for us to collect sensing data? With the rapid growth in mobile device ownership and many sensors embedded, such as cameras and GPS modules, there is an emerging practice called participatory sensing. Instead of relying on auto sensors, participatory sensing deploys human participants to use their mobile devices to generate and share data from their environment. It puts a strong emphasis on data gathering that meets the collective interest of ordinary people and provides benefits for people to interpret and learn from the sensing data.
A key factor to help sustain a participatory sensing project successfully is the engagement of human participants, specifically in three aspects — 1) the amount of data contributed, 2) the noise and errors in the data, and 3) how long they can contribute to sensing activities. Unfortunately, current sensing design can hardly manage quantity, quality, and persistency at the same time. For example, researchers added gamification elements to an ice coverage data sensing project. While it boosted the number of data submissions, the effect only lasted for a short time period. Prior research has tested several ways to engage people in the sensing practice, however, it did not fully discuss why people do, or do not, take part in sensing practices as expected.
In our study, we tried to understand the why question in the engagement of participatory sensing — what are the actual needs and motivations of human participants in participatory sensing projects? Following this question, we looked into the ongoing practice of participatory sensing in a novel context — the status monitoring of shared dockless bikes in urban China.
Bikesharing programs constitute a rising component of urban commuting worldwide. Dockless bikesharing refers to shared bikes that do not rely on docking stations to operate. Equipped with QR codes and IoT modules, dockless bikes provide people with more flexible commuting solutions compared with station-based ones. However, this flexibility creates difficult maintenance problems, such as broken or misplaced bikes in the city.
Bikesharing programs utilize participatory sensing to address the issue — users can file a bike status report when they encounter a problematic bike by identifying the problem and adding descriptions. The filed reports will be checked for quality by the system and sent to maintainers to follow up. See the figures below for an example workflow.
Being a service consumed by over 200 million users daily in China, bikesharing programs provided us with an opportunity to examine a real-world sensing practice at scale, across time and with potentially diverse motivations and behaviors. We conducted interviews with people who have experience in filing status reports to investigate (1) how do people participate in sensing practices? (2) how do people perceive or interpret incentives or motivations in sensing practices?
Models of Engagement Among the Bike Users
Four distinct models of user engagement in participation emerged from our interviews.
The Convenient Model & Normative Model
Most users file reports following the convenient model or normative model. The convenient users often skip many details when describing bike problems, leaving the sensing data less interpretable. Normative users are willing to file quality reports but have few clues to solve issues such as technical failures to submit reports.
The Proactive Model
“We call ourselves hunters. Like some people playing treasure-hunt games in a digital world, we hunt for real bikes in the physical world. There are dozens of hunters groups in different cities, and we do this just for fun.”
The third model we identified is a group of proactive participants or “hunters” in our interviewees’ vocabulary who contributed extra time and effort in the monitoring of shared bikes. Proactive “hunter” participants share their experience such as tips for the sensing practice on online forums. People will organize social events like monthly “hunting” competitions in groups around the city.
The Reflective Model
A final model is reflective users, who aim to provide meta-solutions that promote the effectiveness and sustainability of reporting activities. One example is the app to support “hunting” activities. See the figure below.
Users Shift Between Models of Engagement Over Time
Imagine if you are a user of bikesharing programs; which clan do you belong to? Our study suggests that people do not always stay within one model during their entire history of participation. Instead, people shifted models as they perceived changes in their connections with bike status-related data. Many hunters started their sensing following the normative model but later transitioned to the proactive model, which was triggered and reinforced by hands-on experience to make sense of the data.
Incentives Are a Persistent Open Challenge
Monetary incentives turned out to be less effective. Although encouraged with bonuses for filing reports, users such as urban commuters care more about time and convenience on their last mile to the workplaces.
The second form of incentives considers altruism, or filing reports for the benefit of others. However, due to the transactional and decentralized nature of bikesharing, participants can hardly develop a feeling of reward for knowing their reports benefit people because users do not own the bike nor build connections with others.
One Design Implication — Being Different, Being Complementary
Based on these findings, we argue that, instead of expecting the same group of people to provide sensing reports in both high quantity and high quality at the same time, a more effective way is to have different groups coordinate with each other. For example, people who notice the problematic bikes but in a time-sensitive situation, can file light-weight reports by only identifying the problems. Reported bikes can be marked on the map, and be verified later by people who are nearby but without immediate needs for bikesharing service. Leveraging the complementary strengths and incentives of the diverse user groups could help sustain quantity, quality, and persistence of participatory sensing.
In summary, we conducted a qualitative study on the status monitoring of dockless shared bikes in urban China. The studies identified four models of user engagement in participatory sensing, and proposed several design implications. These findings contribute to a better understanding of the quantity, quality, and temporal aspects of participant engagement in real-world sensing practice.
- Ge Gao, Yuling Sun, and Yongle Zhang. (2020). Engaging the Commons in Participatory Sensing: Practice, Problems, and Promise in the Context of Dockless Bikesharing. In Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI ’20). Association for Computing Machinery, New York, NY, USA, 1–14. DOI: https://doi.org/10.1145/3313831.3376439
