Social production is a phenomenon that lets a large number of people work together to produce common resources. The advent of the Internet has enabled this phenomenon to scale to large proportions, with participants coming from a wide variety of places and demographics. Many such communities have become a part and parcel of our daily lives today. For example, for reference, we turn to the collaboratively-built encyclopedia Wikipedia, and for getting our questions answered, we turn to Q&A sites such as Yahoo! Answers. The last few years have seen this phenomenon merge ways with advancements in online mapping technologies, giving rise to a new class of applications that may be termed geographic social production communities. Resources such as OpenStreetMap are becoming increasingly popular and even Google Maps has added a feature called MapMaker to help crowd-source geographic data. From a human-computer interaction and social computing perspective, understanding how social production happens in such communities and findings ways to increase and improve it are both important and interesting research topics. Accordingly, this dissertation makes contributions in both these directions. While the research and findings of this dissertation are particularly applicable to geographic social production communities, they can be extended to other applications as well.
We first study how contributors in Cyclopath—a geographic social production community—participate by analyzing behavioral log data using visualization and statistical methods. Specifically, we investigate how Cyclopath contributors <italic>specialized</italic> in the tasks they choose to do. We find evidence for specialization by <italic>work type</italic>: Most users edit a single type of map feature, such as points of interest or roads and trails. We also see a user life-cycle effect: as users gain experience, they specialize in editing roads and trails. Our findings suggest more effective ways to organize social production interfaces, compose units of work, and match them to users who want to help. However, matching tasks to people is, at its core, dependent on compliance with requests to contribute on the part of the users. Therefore, as a next step, we investigate into techniques that may help us increase the chances of this happening.
Social psychology offers several theories of potential use for designing techniques to increase user contributions to online communities. Some of these techniques follow the "compliance without pressure" approach, where users are led to comply with a request without being subjected to any obvious external pressure. We evaluate two such techniques—<italic>foot-in-the-door</italic> and <italic>low-ball</italic>—in the context of Cyclopath and report that while both techniques succeeded, low-ball elicited more work than foot-in-the-door. However, we find that these effects were one-shot and contribution levels drop back to pre-request levels soon after. We also find that while these techniques have the potential to succeed in the short term, they could cause long-term harm, because users may feel manipulated and lower their sense of belonging to the community. We believe that one of the reasons for this was the inherent <italic>unnaturalness</italic> in the process: users were being coerced to contribute, i.e., do something that they typically do not. Therefore, as a next step, we explore ways of eliciting contributions by leveraging natural processes.
One such natural process is <italic>information consumption</italic>. Accordingly, we explore the feasibility of using the act of <italic>consuming</italic> information as a gateway to <italic>contributing</italic> information; specifically, we investigate semi-automated means to extract useful information from standard types of explicit <italic>user feedback</italic>. We analyze naturally occurring textual route feedback in Cyclopath, finding that the feedback was rich in information such as bikeability ratings, tags and notes that are useful to improve the system's route finding and navigational assistance capabilities. We also present a technique to extract such information by engaging users in dialogue immediately after they obtain a route.
Finally, we explore the utility of online social production through the lens of an application that has not been well explored before: how citizen-driven online social production helps the government activity of transportation planning. We described the design of a novel route analysis tool based on Cyclopath to assist transportation planners to make better planning decisions. We highlight the advantages of using a online social production system over other, similar ones through a real-life usage scenario. We believe that the results and ideas in this dissertation are applicable to a broad class of online social production systems.