In my opinion one of the main reasons for large numbers of "raw" gedcom profiles is the learning curve associated with wikitree. You find the site, look around, like the idea, start importing gedcoms, find all these "orphaned" profiles and think, well I have an ancestor there, I'm not there yet importing that line of my family, but, hey, they are my ancestors, so I'll adopt them, and then it takes only one large branch of the family (in my case 3,000+), and you are over the 5,000 limit. By that time you may have discovered that you can contribute good information to some G2G questions and discussions, you may want to continue with some research, and even in my case, wanting to hang in there and complete these profiles, I'm currently sitting on a huge backlog of profiles needing clean-up because clean-up is slow. Not to forget new projects and/or categories just developed recently so that you have to go back and re-edit your profiles. And I assume some people will have given up going back to these profiles all together.
It seems to me that there are two questions that should be addressed: One: how can we gently, but still up-front, make it clear to newcomers that importing a gedcom is not the end of things but only the beginning - without scaring them away. And two: how can we address Margaret's issue, the technical side of making gedcom results more userfriendly.