Writing code in a language which doesn't fully exist yet seems rather ill-advised, the same way that writing to a library that's not functional or basing a project on software or hardware that doesn't yet exist is ill-advised. Doesn't matter if it's perl 6, Duke Nukem Forever, or whatever. (The same advice, several years ago, would've been applicable to betting on Apache 2 or Mozilla or Boston's Big Dig)
This isn't to say that perl 6 won't exist. It may, it may not, and when it does you may or may not care. Betting a project on another project that is significantly incomplete and whose completion date has a significant amount of uncertainty is just unwise.
This isn't to say you shouldn't use some of perl 6's features, not at all. The perl 6 modules on CPAN all provide interesting and useful functionality, and if you think that perl 6 may be the way you want to go in the end then it'd be sensible to use them where you can, to make transition easier.