Rick Wash's Research Projects
My research interest is studying incentives in social software systems. It can be difficult to predict or understand user behavior in social systems because this behavior strongly depends on both the actions of others in the system and the specific design of the system. My work looks at how users use these technologies, why they use them in this way, and how different incentive structures in these systems can produce different user behavior. I am particularly interested in software systems where users all have different information, and appropriate sharing of that information is key to the success of the system. Such systems need to be carefully constructed so that users are motivated to share their information, and to share it in a way that others can use.
One of the most daunting information security problems is home computer security. Hackers are increasingly exploiting the millions of these machines to build Òbotnets,Ó consisting largely of home machines that hackers compromise in order to remotely control them.ÊÊThis mixed method project focuses on users and usage of security technologies that protect home computers. We use qualitative interviews to understand the mental models of the threats facing home computer users. We are designing a new security technology that helps users to better secure their own home computer. This design is strongly grounded in social science theory; we use social psychology theories and economic modeling to understand incentives in the system, individual responses to them, and inter-individual strategic awareness and behavior. Finally, we use a human-subjects lab experiment to validate these design principles and evaluate this new technology.
User-contributed metadata, also known as tagging, is increasingly receiving attention as a low-effort tool for digital information management. Collaborative tagging systems provide a means for users to associate personally salient keywords or labels with content items, and then publicly expose these associations, so everyone can benefit from this information. Most research in this area has focused on the outputs of these systems, such as the ÔfolksonomyÕ that emerges from many users making many tag choices. Instead, we focus on the inputs; how do users choose which words to use as tags and when to apply them to a content item? We also look at how systematic differences in this input -- tag choices -- lead to specific patterns in the output folksonomy, which after all is an aggregate of these tag choices. This mixed method project uses qualitative interviews, large-scale statistical analysis, and computer simulations to understand user behavior on a prototypical tagging system -- del.icio.us.
Spam email is fundamentally an economic problem. We characterize a number of existing technological and regulatory solutions to the spam problem in economic terms, and explain a number of their faults. We then develop and analyze a new anti-spam solution (the Attention Bond Mechanism) that uses economic incentives to discourage spammers. We take advantage not of the information contained in a message, but of the information known by the sender to screen email before it is even sent. This idea is currently being developed into a product by boxbe.com