3. Why pattern matching is way cooler than car/cdr; why arc isn't a 100-year language; comparative study of macro systems; comparative study of module systems; implementation of a bootstrapping arc compiler; implementation of my pet lisp, chao, which is bootstrapped from arc and compiles to a (hopefully) portable IL... I have lots of ideas.
4. Summer break, since that's when I'm in Silicon Valley.
5. If someone wanted to destroy Arc as a language, this would be the time for them to do it. Everyone interested, gathered in one small space. Be on your guard, comrades! :P