Creating user interfaces that are natural, guessable, learnable, memorable and accessible is a persistent challenge.Involving end users in the design process is a well-established approach to address these challenges, but traditional participatory design has limitations, especially when it comes to scaling beyond the lab and reaching diverse participants. I build on the success of a popular participatory design method called end-user elicitation. Elicitation studies work by presenting the effect of an interaction (e.g., what happens after a user makes a gesture) and asking end-user participants to perform the action that would have caused that effect (e.g., the gesture itself). Despite their success, elicitation studies have important limitations. They typically are confined to a lab setting, limiting the number of participants, their diversity and the representativeness of the results. Also, analyzing the studies' results is a laborious process. Furthermore, elicitation studies lack a formal approach to evaluate the quality of their results. My work addresses these limitations by scaling beyond the lab and conducting distributed elicitation studies with online crowds. In this dissertation, I have created an open platform and formulated the Distributed Interaction Design (DXD) process. This work overall contributes methodological extensions to elicitation studies, an open platform for the research community, and empirical studies of user-generated interactions such as gestures, voice commands, and icons. The thesis I demonstrate in this work is:Using a custom-built platform to conduct Distributed Interaction Design (DXD) enables: creating user-elicited interactions; evaluating the guessability, learnability, and memorability of interaction designs; and the recruitment of participants through third-party services in a timely manner.