hm. so the most important part of this project is this (as I see it): an ordering / structuring alrorithm. So tagging is pretty easy (in theory, not practice), so that's not the issue. And metadata can be anything, so that really isn't the crux either. The problem to me seems that metadata, unless you define it as otherwise, is highly unstructured and variable, uncontrollable. Think about the system we designed. The content_type and placenum were pieces of metadata that were specifically defined -- and were therefore easily structured. para_1, you go with header_1, etc. how will your system sort, parse, etc. the metadata affixed to objects into communities if there is no definition, no (xml-like) spec? If I annotate a texas cow with the tag "certified agnus beef", and then annotate a texas steakhouse with "best little steakhouse in texas," what does the system do. Does it match "texas" and put both objects in a texas community? Maybe that's a start, but how does it recognize that "steakhouse" is related to "certified angus beef?" I have some suggestions here, but I just wanted to pose that as a direct question.