The "Collaboration" section that starts on page 9 of the draft is a great start. Over at
I've listed a few more. Summarized here:
1) For each data set, provide a set of developer-generated recipes for data manipulation so that future developers can stand on the shoulders of previous users of this data set by sharing tools and techniques.
2) Provide tools for crowdsourced analysis, such as labeling points of interest on a map, interesting pages in a large document, or parts of a dataset that need more attention.
3) For each dataset, provide a list of current tools built on top of it.
4) Allow non-developers to curate and rank wishlists of analyses that they wish someone would do on a dataset.
5) Allow the community to curate and rank wishlists of datasets that are not yet available.