I've been interested for several years in the ability of "player-generated content" to create niche or small multiplayer online games, at a company I've invested in called Skotos . At this point its oldest game Castle Marrach is completely run by the players, and 7 more games are being developed by their customers. The company is at break-even, but is not yet a 'commercial success'. There are some learnings from the world of massively multiplayer online games that apply to other kinds of social networks.
Some of these learnings have to do with size of groups -- different sizes of teams and groups in games can have very different effects, both positive and negative. I've never seen any Social Software that innately understands that a team of 7 behaves very different then a clan of 50 or a megaclan/tribe of 350. These behaviors have parallels to size of mailing lists, online communities, and even to the size of corporations. If Social Software can integrate an understanding of this then it might be significantly more successful
I'm also interested in what I call "social games" for online games. I don't believe that current MMPORGs serve well what the industry calls the 'socializer', which consists of 15% of current gamer population, but may in fact be much higher if you include women who don't currently play as many online games. Two recent articles that I've been collaborating with Shannon Appelcline on this topic: