One of the fallacies associated with social games is that they need to be designed for play among friends.
If you look at the games that have succeeded on Facebook, most are single-player games. If they do have interaction with friends, it's indirectly through leaderboards.
For game that have more direct interaction, such as Texas Holdem Poker (full real-time interaction) and Mob Wars (direct turn-based interaction- by attacking other players), gameplay revolves around interaction with strangers.
And yet, in order to be viral, games need players to invite their friends to play. Mob Wars is the prime example of this, a game where you complete jobs to earn points so you can battle others, but need to invite friends into your mob to advance through the game. It's a very good model: competition is focused on strangers, and cooperation is focused on friends. Find ways to emulate it.
But I Thought Facebook Was All About Interacting With Friends?
It used to be. Not any more. A long while ago, I designed a game around the assumption that people on Facebook only befriend people they know. In the game, players were challenged to find a picture of their friend in a grid of photos. With each level, the amount of stranger's photos in the grid increased so finding a friend's photo would get more difficult. Basically, Where's Waldo with your friends as Waldo.
We started playtesting the game and quickly realized it was never going to work. People couldn't recognize their friends' photos. In fact, some people didn't even recognize their friend's names!
When you have hundreds of friends, it turns out that you don't even know who many of them are.
The point I'm failing to make here is this: Facebook is a platform for interaction with friends AND STRANGERS.
If you're only building for interactions between friends, then you're not leveraging a significant chunk of the interactions on Facebook.
Finally, if you're foolishly planning to build a real-time game, than keep this in mind:
Real-time games designed to be played with friends will not have critical mass due to lack of density. -- Wade Tinney, Large Animal Games.I nominate that for the 3rd law of social game design (if only because it sounds like something out of my college physics class.)