Well, doing better than Perl5 shouldn't be that hard. I mean, come on, it can't even collect reference cycles. In 2013. It doesn't have bloody subroutine signatures. And let's not mention WTFs like "Variable "$foo" will not stay shared".
That said, the field of "dynamically typed" (read: unityped), object-oriented programming languages is crowded enough already; I don't think we need Perl 6 either. Parrot would have been interesting if it had actually delivered as the "one dynamic VM to rule them all", so that code can be shared between different languages. But the article seems to indicate otherwise. Perhaps PyPy will be more successful, given the recent development of Topaz.