I think that package management systems are
1) Not perfect (at least I haven't run across one that is perfect).
2) The most important part of a distribution.
Right now, GSoC package management system proposal seem to fit into one of the following
1) Infrastructure to improve how developers maintain packages.
2) Usability improvements.
3) Clean up or introduction of a feature.
There are exceptions, such as a proposal to convert packaging formats.