I think this is a really bad idea. 64-bit in userspace buys you nothing for almost everything, and for those things it _does_ help, you can already install 64-bit versions.
You're talking about splitting the already relatively small PPC userbase into 32-bit and 64-bit parts, and doubling the amount of QA and release work we have to do for PowerPC. The 32-bit code is mature and well-tested already.
You'd do better finding any remaining multilib problems and fixing them, rather than changing wholesale to 64-bit.
The amount of work it would take to produce both a 32bit tree and a 64bit tree for ppc users far outweighs the amount of ppc user/feedback we get currently.
If it were a matter of just taking our current 'ppc' offering, which is a ppc64 kernel option, with mostly ppc32 userland, and instead make it just a ppc64 kernel, with some ppc32 userland, but mostly ppc64 that would be about the same amount of work. However it would cut off the people whom only have ppc32 hardware to use, aka much of the Apple hardware would now be useless to Fedora.
Given that the bits are free, anybody who is interested can create a pure ppc64 spin for use wherever. I just don't see this as useful to be Fedora's "ppc" offering, nor do I see the ppc* user base as large enough to warrant 2+x the work in order to have both a ppc32 and a ppc64 offering. In short, I'm very against this Feature, from a release engineering and QA perspective, as well as a user perspective given some of the above alternatives. I'd much rather see a community of interested ppc64 users/testers join together and produce a ppc64 spin for your user.
-- Jesse Keating - Fedora Release Engineer