it's nice to see more and more people joining this group, so let me try to stimulate debate a bit by asking what reuse scenarios you are aware of, which ones you would like to see further explored, or which ones you deem problematic.
I think one basic reuse scenario is text mining, that certainly isn't as easy as it should be, yet. Are you aware of any projects that have text-mined the PMC OA subset, and published the results in a linked-open-data way? I think that would be a worthwhile project.