The U.S. Census Bureau is one of the largest survey data collection organizations in the world, in addition to its role in the collection of the U.S. Decennial Census data. The two major statistical tools used by the Census Bureau in designing its surveys are stratification and multi-stage sampling. These tools have been successfully implemented starting in the 1940s and have continually been adapted and refined since then. While this general sampling approach has been very successful, there are increasing concerns about rising survey costs, decreasing response rates and new frame coverage issues (especially related to telephones). At the same time, advances in data collection methods, new data sources and computational tool offer opportunities for considering survey design approaches that would have been unfeasible before. In conjunction with the 2010 Redesign Program currently on-going at the Census Bureau, input was therefore sought from leading academic researchers in innovative sampling methods, as a way to initiate the exploration of possible new approaches to design surveys conducted by the Census Bureau. As a result, Profs. Steve Thompson (Simon Fraser University), Sharon Lohr (Arizona State University) and Yves Tillé (Université de Neufchâtel) were invited to give overview lectures on some of the designs they developed. I was invited to contribute a discussion to each of these lectures. In the three sections that follow, I will summarize my comments to each of these lectures. My goals in those comments were to highlight the most important aspects of the sampling methods that were presented, to discuss some of the main opportunities for using these designs in the household sampling context, and to identify possible challenges in implementation.