My luck being what it is, I'd write some program based on the specifications and Pugs will have not implemented some key feature of that program. I'm not saying that I'm sort of über coder, just saying that I'm unlucky. ;)

What is wrong with that? This type of thing has been driving the Pugs project from the very begining. I started writing (failing) tests based off the Synopsis in the early days of the project, and within days autrijus would have turned them into passing tests. Both chromatic and Darren Duncan and their work on the p6 versions of Test::Builder and Locale::KeyedText respectively, drove much of the early OO functionality in Pugs. The current work on the Javascript and Perl 5 backends have helped fuel autrijus's work on PIL (Pugs Intermediate Language) and my work on the Object MetaModel.

My point is really that Pugs is a different sort of project than your usual compiler/interpreter project. It is (in autrijus's words) optimized for fun. And autrijus has always said he does not intend Pugs to be "the" Perl 6, and one of the original goals of the project was to help iron out the design of Perl 6 itself by providing a reference implementation.

Another thing about Pugs is that all you need to do to get a commiter bits is ask, and all you need do to get your name in the AUTHORS file is contribute something (even commiting it yourself is not a requirement). No one is expected to contribute any more then they want to or can. Because after all, it's optimized for fun :)