Fair-warning... this is a straw man topic. I began by making the overly-haughty provocation on a web 3.0 (I guess?) enabled space, "Before we, the self-appointed 'committee of the whole,' get too unwieldy in scope and participants to actually communally-generate useful, shared tools, can we establish a representative subcommittee who can recommend initiatives back to the group at large and potentially commit the resources of our respective institutions to a central pool?" In response, I was asked how I would recommend bridging Web 2.0 and Web 3.0 communities. Implicit in this question, I think, was the assumption that Web 3.0 community was a more appropriate sphere for cultural organizations (the "authorities") than Web 2.0 (everybody else, including authorities in mufti).
So, to attack the straw man. If Web 2.0 is defined as social/wiki websites on which users participate in generating site content and Web 3.0 is the semantic web, an interlinking of reliable, bot-filterable metatags (and their respective communities the authors/participants in said content generation), I'm not sure there's really a divide. To the extent that the semantic web is reliant on reliable, consistent metadata and shared taxonomies (I dislike the term 'ontology,' which seems like a malapropism to me), it would seem that the Web 2.0 is the place where these will begin to be built and continually tested. Having read Cory Doctorow's 2001 "Metacrap" article today, I think I have a more immediate understanding of the potential pitfalls of unguided folksonomy and unmoderated wiki (or vice-versa), but I likewise think that the types of techniques Luis Von Ahn has and continues to propose are one way around them [go here for more on Luis]. Furthermore, top-down metadata development like adherence to latin naming conventions, Dublin Core, and the kind of metadata entry required by SIRIS and the Library of Congress, which are already of use to the semantic web-building of academia (and by extension, lay researchers with "serious" purpose accessing collections), are likewise promoted through shared workspaces such as this and social networking environments (such as this, for those coming to this blog via my Facebook page).
We absolutely need fora for communities to gather and spitball/play out ideas. But that still leaves action to chance. It would be nice for the community itself to empower a subgroup itself responsive to and accountable back to the community that is capable capable of action. The empowering resources required are money (whether tithed or derived from third-party funding sources), server/storage access (the workshop for development and testing), and time (for example, via commitment of relevantly skilled staff to patronized projects). I'm assuming that the online community can serve as its own extranet regarding reporting requirements, publication (of white papers and/or code), and distribution.