I certainly agree that a paper's influence metric should value more than just a simple citation count, and the CSpub system is designed around facilitating a change in that direction. I'm not sure about the query-based citation ideas, certainly if there's any kind of "auto-update" mechanism that changes referring papers references to "upgraded" papers.
Nevertheless, it's a good read. Something needs to change in the system. Reviewers must be overloaded these days (in my most charitable judgement), because many of the reviews I see are entirely too weak (e.g., a weak reject with no explanation or suggestion for improvement).
EDIT (2010-08-17): researchr.org seems to implement some of these ideas (publication interlinking and a "credit" metric of some kind.