Wow! Charles Nutter is indefatigable. Since the implementers summit, he's joined the YARV mailing list and been asking questions and making comments there, he's gotten the RubySpec wiki fired back up, been plugging the rubytests project, and now he's announced that he's working on implementing a stack based machine based on YARV bytcodes on the JVM. This experimental work should help clear the way to a fast, stackless bytecode interpreter for JRuby.
I've also seen Kevin Tew cropping up in discussions about non-cardinal implementations. I'm looking forward to seeing more examples of this as the various implementers get more used to one another and as the communications channels get a bit more established. The future looks bright!