It depends a lot on (a) whether friendship is symmetric (Facebook) or asymmetric (Twitter or Facebook subscriptions); (b) whether friendship is IRL (more often the case with Facebook) or knowledge-based (Twitter).
The best scenario is an asymmetric network where I follow X and X follows Y. There's a good chance I'm interested in Y. The worst scenario is an asymmetric network where the opposite is true. X follows me and Y follows X. Good chance Y isn't even human.
So when I'm shown a Twitter FoaF, there's a 2x2 matrix of possibilities. Jelly seems to use at least 3 of them, since one label says "From X, follower of Y" while another says "From X, followed by Y" and another says "From X, friend of Y". I'd guess their stats will shown "From X, followed by Y" to be the most useful.
Also networks based on social/URL/family connections are probably less likely to be useful when it comes to FoaFs. Just too much variation and a lot of content being shared like random photos is only intended for immediate connections.
The best scenario is an asymmetric network where I follow X and X follows Y. There's a good chance I'm interested in Y. The worst scenario is an asymmetric network where the opposite is true. X follows me and Y follows X. Good chance Y isn't even human.
So when I'm shown a Twitter FoaF, there's a 2x2 matrix of possibilities. Jelly seems to use at least 3 of them, since one label says "From X, follower of Y" while another says "From X, followed by Y" and another says "From X, friend of Y". I'd guess their stats will shown "From X, followed by Y" to be the most useful.
Also networks based on social/URL/family connections are probably less likely to be useful when it comes to FoaFs. Just too much variation and a lot of content being shared like random photos is only intended for immediate connections.