Clojure seems to be the hot candidate in the "new lisp" space. I think that is partly that Rich Hickey seems a lot more committed to Clojure than Graham does to Arc. (Most of Graham's writing seems to be about ycombinator and startups.) The "100-year language" thing may also play into it; Clojure seems much more likely to be useful in the near term than Arc does, because it is interoperable with Java. Good or bad, there is a lot of Java out there.

Graham's a great essayist, but I don't think he can split his attention between yc and Arc and have both be successful.