Mechanism design is a branch of game theory and economics that is used to create methods for allocating scarce resources based on user reports, such as the value for a good being auctioned off. One of the desiderata of mechanism design is to design truthful mechanisms, in which it is optimal for participants to submit truthful reports. In his seminal work on mechanism design, Myerson describes how to design a truthful mechanism that can optimize expected revenue. This mechanism is applicable when participants are fully characterized by a single parameter, called their type, which is assumed to be private to each participant. While Myerson's mechanism relies on full distributional information about the participants' private types, this thesis is concerned with the design of single-parameter mechanisms that relax this assumption. Myerson's optimal mechanism concerns agents with quasi-linear utilities, in which utility is linear in money. We study settings in which agent behavior is affected by the Pareto principle, also known as the 80:20 rule, which makes utilities non-linear. For example, in a contest for artwork, one can easily submit a stick figure; submitting anything at the level of the Mona Lisa is exponentially harder. Our mechanisms, which operate despite missing details about the distribution of types, such as abilities in contest settings, are truthful and guarantee near-optimal performance. Next, we study mechanisms for cloud computing, in which users' types reflect their demand for compute resources, based on their intended workload. We built a classifier to label a public dataset of virtual machine traces according to workload categories, from which we extracted per-workload resource demand distributions. Then, we designed a truthful, revenue-maximizing mechanism to allocate and price cloud resources, assuming users' demands are drawn from these empirical distributions. Our mechanism is a natural one for the cloud, as users report their needed quantity given a known value, rather than their needed value given a known quantity.
