Virtual communities are everywhere: national policies are influenced by debates on mailing lists like com-priv and cypherpunks. Our children flirt on AOL and our parents debate Medicare changes (and flirt) on Third Age. But what are virtual communities? How are they designed? When do they enhance the lives of their members, and when are they a big waste of time? As these technologies increasingly surround our lives, will real people have meaningful control over them, or will that remain the privilege of specialists? In this talk, I will outline what I see as some of the important outstanding research questions about virtual communities, focusing in particular on issues of end-user programming. I'll present results from two ongoing projects: MediaMOO (a MUD designed to be a professional community for media researchers) and MOOSE Crossing (a MUD designed to be a constructionist learning environment for kids), and share early progress on Net Flyer (a distributed, 3D, multi-user environment designed to encourage high-school students to learn about computation, math, and art).