I did some research on the declining editor rates for a course project. Without recreating our entire paper, I want to make this important point. Wikipedia is not (necessarily) nearing the completion of all knowledge, but rather what is feasible for its editors to contribute - indeed, there is much work to be done on "the history of sanitation and sewage in ancient Carthage" and "topic sensitive page rank", which at least has a page, let alone "trust sensitive page rank" which seems to be missing one. Our core piece of evidence was observing the _same_ trends across different languages (in both # edits and # page views). If you are interested, check out our final report here - it has its flaws, but I believe the three pieces of evidence we explored hold some merit in their own right. (http://www.stanford.edu/class/cs341/reports/09-GibbonsVetran...)