The most effective incentive mechanisms are often among the simplest technically. I am currently studying three very simple incentive mechanisms that are designed to induce users to contribute information to a social media system. My goal is to understand how each mechanism effects the user behavior in a manner that allows me to predict behavior in novel systems that use these mechanisms. In other words, I am looking for incentive mechanisms that generalize to a wide range of social media systems, and establish causal relationships between technical designs and valuable user contributions.
First, I am looking at the idea of setting a “minimum threshold” for contributions, and excluding any user that doesn’t contribute enough. This mechanism is a simple variant of a number of mechanisms in the excludable public goods subfield of economics. To study this, I developed an economic model of user behavior, and used that model to rigorously derive the expected changes in behavior when a minimum threshold is implemented. The results aggregate behavior is complex, but as an overall description, this mechanisms increases breadth of contribution by encouraging more users to make a non-trivial contribution, but sacrifices depth in that formerly high contributors voluntarily decrease their contributions. I am in the process of designing an experiment to validate the model predictions in a more realistic experimental setting.
Second, I expand on an idea I discovered studying the social bookmarking website delicious: giving users a private reason to contribute and then making those contributions public as a side effect. Not all contributions need to be done for social reasons; users can often use a system for purely personal reasons. However, their contributions can be shared with everyone else as a side effect; however when using this mechanism it is important to ensure that the type of contributions made by the users is the same as the type of contribution valued by the consumers; I call this design constraint “incentive alignment.” I discovered that this constraint was important on delicious; I am now trying to establish that this concept generalizes to other social media systems, and measure its impact.
Third, I am studying what is probably the most simple and straightforward mechanism for inducing contribution: ask the user to contribute. The “power of the ask” is a well-known mechanism that is quite effective at inducing people to contribute to charities. However, monetary contributions are significantly simpler than contributions of information because information can be of varying quality: good or bad, biased or unbiased, right or wrong. I explore first how powerful asking can be at inducing contributions, and second how different types of asks can result in different qualities of information.
To study these, I am excited to be conducting two randomized field experiments using existing social media systems. First, I am conducting an experiment on Everything2 that attempts to measure both the power of the ask, and the effect of incentive alignment on users’ contributions of 5-star ratings. The ask mechanism focuses primarily on inducing more contributions, and the incentive alignment condition focuses primarily on the quality of contributions. I suspect these two will complementary, resulting in greatly increased contributions when used in combination. Second, I am conducting an experiment on the Great Lakes Echo website to look at how timing of the ask can influence both the quantity and quality of ratings contributed.
This is joint work with Cliff Lampe.
Resources:
Rick Wash and Jeff MacKie-Mason. “Using a Minimum Threshold to Motivate Contributions to Social Computing,” Working Paper, June 2009. (PDF)