Some thoughts to start the conversation:

In a recent paper on UGC, I argued that to evaluate the value of UGC in the museum, it is useful to begin by defining it as distinct from other interactive experiences such as feedback, souvenirs and aide-mémoires. 

Examples:

1. Feedback: voting, polling, e.g. Tate Modern's multimedia tour, e.g. Duchamp's Fountain stop where visitors vote on "Is this art?" Yes/No
2. Souvenir: e.g. photos of self in exhibition (see Visite+) 
3. Aide-mémoire: bookmarking, e.g. Boston Museum of Science, Star Wars exhibition multimedia tour
4. User-generated Content: visitor makes a creative response to the museum visit in audio, text or image, e.g. leaving voice commentary in the Pompidou Centre's experimental CineLab cellphone tours, or Tate Modern's Educational Tour pilot which allows visitors to write, draw, and record their voice comments through PDAs.

New technologies for capturing visitors’ spoken, written, and sketched responses to what they see turn visitors into teachers, active interpreters and explorers of their world rather than passive recipients of the narrowcast voice of the Museum Author-ity. 

The interactive capabilities of mobile platforms in particular remain largely under-exploited in museums, and important opportunities to change the visitor's relationship to the museum experience are untapped. Recent experiments with these tools, like the ones mentioned above, have demonstrated their potential to empower the visitor as a storyteller. By equipping visitors with the means to capture their own impressions and learnings from the museum to share with others, we turn visitors into a global teaching force that will act as advocates for the museum for generations to come.

Process, not product

Yet wary of turning the museum into a ‘reality TV’-quality experience, many museum professionals have questioned the value of UGC, which is likely to be less well informed in its interpretations than museum-authored content, and less polished in its presentation. While it is true that visitors will not necessarily produce ‘better quality’ interpretation than the museum’s own, the process of authoring that content has an educational value and impact that outstrips what visitors will retain from passively consuming the museum’s message. User-generated content is about the process, not the product.

Between User-Generated and Museum Speak: Peer-generated content

I offer for consideration the role of Peer Generated Content. This is content for targeted audiences developed by persons sharing key characteristics of those audiences.  Sliced and diced anyway one chooses, targeted groups (and their content developer "surrogates") could be characterized by special interests, age, cultural/ethnic attributes, gender, religion, sexual orientation or any way which would treat museum audiences in a niche or "long tail"  approach which would sum to larger, more engaged audiences.  Unlike simple, random, inconsistent (in quality) open user-generated content --- which does have its own important role and, as Nancy noted, may be more about process that product -- mildly directed, focused and supported peer generated content may bridge the gap between institution-talk and production values and overwhelming, unjuried random content which may not serve the institution all that well.  Artmobs and their unauthorized MoMA audiotours may be one example of peer generated content. However, I am suggesting that museums prod community based groups to produce content, with whatever technical or content assistance are appropriate (equipment use, access to curators, related images/digital assets, etc). I suggest that peer generated and targeted interpretation has the greatest likelihood of effecting meaningfulness and acceptance; fortunately the delivery tools we now have can store and deliver a wide range of personally directed content without extensive and expensive capital investment.  A very early experiment in peer developed content is described in the document "sharing_passion_ls.pdf"