According to some testing done by Curator Randy Schoenberg (IIRC) a very LARGE percent (30-40% ??) of Projects are empty shells, for some restrictive definition of "empty", such as having no project descriptions, few if any profiles and a single Project Collaborator (or 2-3 at most). If we expand the definition of "empty", we will probably include more than half of all projects.
We presently have 10533 projects, so if some 30-50% of them are of no real "value" then all they do is prevent users from finding the ACTUAL projects that people put a LOT of effort into. To use my favorite project as an example, the Biblical Tree, if you search for "Biblical" you will find 63 projects, "Bible" will return 95. This is crazy! NOT all of these project are empty, but if we get rid of the garbage projects, people would be less likely to duplicate efforts (this being the whole point of Geni's model).