The latest advances in information and communication technology have made it possible for researchers to design innovative crowdsourcing systems that can harness the human intelligence of online communities. While crowdsourcing system designs have progressed substantially through engineering breakthroughs, some challenges in the crowdsourcing model however remain unanswered, such as (1)System Design: in various crowdsourcing systems, what roles can crowds play and what contributions can they make? (2) Data Analysis: how can the human inputs with varied qualities be properly cleansed, and how can trustworthy results be effectively generated from their myriad inputs? (3) Human Computation Theory: at a higher level, what is the symbiosis between human intelligence and artificial intelligence? With lessons learned and experiences gained from four projects, this dissertation aims to provide new perspectives and insights into answering these questions. When presenting research observations and results, we discuss a variety of technological and organizational considerations in crowdsourcing system designs.