Raph Koster has [as usual, certainly] summed up a fair bit about social gaming in his GDC presentation. I wasn’t there, and so I didn’t get to hear it, only read it. I like that the main additions to his Austin version were to point out where the greatest potential lies/the greatest current failures [of vision, primarily.]
Various of his concepts are fairly ripe for exploration – and I wholeheartedly agree that there is little to no consideration of psychology, of gestalt – but one stands out at the moment; that of clusters (see page 121 and onwards of his PDF presentation.)
Of course social networks are comprised of multiple, smaller clusters, but so little is done to exploit or explore that (other than the nigh-constant barrage from the game players in one’s circle to join or play this or that.) Why not develop a mechanism by which existing clusters (that is to say, a clump of friends all playing the same game) of players can work together, cooperatively, on something? That could work easily into a business-sim game, wherein the cluster of friends all formed a “company,” as just one example. Assignment of game-defined roles would also be able to play a part here, as well as, potentially, elections (these are also all concepts from Koster’s presentation.) Allowing real-world friend clusters to form guilds in strategy/RPG-style games, developing communication mechanisms to piggyback on the social network’s existing cluster-oriented messaging and forums…all of this has great potential.