A useful thing might be to provide "good" GEDCOMs that are associated with good (documented) genealogical articles (and books). The GEDCOM would just cover the people in the article, and would have source entries for them. The sources could either just refer to the article, or perhaps to the sources used in the article but qualified, probably, by reference to the article. For instance, "U.S. Census, ... page 123, according to John Doe, The Doe and Dough Families (1999)".
Who prepares these? Ideally, the author of the article. This makes sense when the article is being published at the same time. It could also be done by the author later. It could be done by anyone willing to do it, but then the question arises as to how acurate it is with respect to the article.

Maybe there just aren't that many people who want to publish documented family history, at least not both for long-term access and for access via the net. Most are just involved in the hunt, I suppose. Leaving markings of the trail for others may not be of interest to them. So maybe the "glaring need" for long term net publishing is in the eye of very few beholders.

There seems to be a real lack of a good place to publish genealogical papers on the internet. There's nothing that I know of like NEHGR, NGSQ, TAG, etc. Yet the net has great possibilities for publishing for small audiences (as shown by blogs). Genealogy is certainly a field where the natural audience for any particular story is small.