Inertia and momentum - at least currently, Pugs-on-Haskell has all the momentum, and Perl 6 development has shown a lot of inertia. Of course, maybe some time somebody starts a topaz-like effort to implement Perl6 in itself; and this time, it's much more likely to succeed, as Perl6 has enough power to become self-sustaining (unlike Perl5), and can maybe even have a straight first machine-generated translation from Haskell. But I don't see that this will inevitably succeed, while I see the success of Pugs/Haskell immediately.