Definitely twitter?

So, I'm thinking... twitter is maybe not the best corpus, really. (See pourquoi twitter?)


Main disadvantages: restraints (character limit) are likely to make the corpus more homogeneous; friendship/social connection data is uncertain.




Other possibilities: 

• facebook - I felt that the main problem with this would be data rights issues. That being said, I'm not sure that the twitter data is free (to anyone, to do what they want with) either. Of course, there's the justification that it would be for research rather than commercial purposes. On the other hand, the difference between facebook and twitter is that (most) people on facebook protect their data to some extent - making it only available to their friends, for example. If I were to use data accessible from my own facebook account, I would definitely need to ask permission from the (potential) participants.
  The amount of data I could get from facebook would be a lot smaller than that from twitter, but this isn't necessarily an issue. I'm pretty sure that I could still get enough data do some interesting research.
  
  Main advantages of facebook data: longer individual posts; better data on social connections.

• blogs - blogs would provide a lot more data per participant, as individual posts are generally longer than on facebook or twitter. However, the type and amount of interaction on blogs is completely different. I feel like the type of social connection data from here would be "less good", but maybe I need to think about it for a while.
  
  
• forums / newspaper comment sections - this is a totally different set-up.
  For one thing, less social connection data. I guess I would have to take all the discourse from one place and say that that's a social group, and take all the discourse from another place and say that that's another. But I could take data from a hacker forum and from a wine lovers' forum and show that there are linguistic differences... what would be the point?

...trying to think of other ideas.