Event hashtags work well if the event is large enough to sustain them, e.g. #altc2010, with >700 contributors. I'm not a fan of salami slicing hashtags, e.g. separate tags for multiple sessions at a single conference. This balkanises the conversation and fails more often than it succeeds. An example of this was #reps10 - with only 18 contributors and <100 tweets, the conversation fizzled out quickly after the event. I can understand why the Centre for Bioscience did not want to use #heabio (which was used for the meeting in 2009) after rebranding, but in chopping and changing, they have missed the chance to establish an ongoing conversation between distributed representatives with little else in common. By adopting a hashtag as an ongoing social object, we extend rather than limit the power of social media.
Nielsen's rule of participation inequality suggests a community of ~100 may be necessary to maintain a conversation, although this figure was derived for a more passive, asynchronous medium than Twitter and it is not clear how to adapt this for realtime channels. The fact that hashtags and RSS provide an asynchronous element to Twitter suggests that 100 is a good rule of thumb for a viable community. Having said that, the #solo10 community, with 750 contributors, limps on a month after the event. I suspect that this is because the tag was event-centred rather than formed as an ongoing social object. This suggests that purpose and expectations matter in addition to numbers. Presumably, a smaller tight-knit group such as a department or course could maintain an ongoing conversation if they are imbued with sufficient common purpose.