Arrrrgh! This topic is too big!
I wrote an essay on interplayer emotion. It was extremely long. Here's the condensed, condensed, condensed version:
Interplayer emotion is based on shared resources. Depending on the nature of the resources, you get different player emotions. The survival resources most MMORPGs share lead to the more primitive emotions, whereas long-running LARPs frequently share storylines and politics between characters, producing a very different set of emotions.
More factors are how difficult resources are to obtain and keep. The more independent a character can be, the less emotion is likely to rise out of a situation, and the smaller the groups will tend to be. In SecondLife, the basic group size is two people. In Eve Online, it tends to be two digits. That's because in SecondLife, life is easy, and in Eve Online, life is hard. MMORPGs generally let you choose how difficult you want life to be, the only real reason I like treadmills.
Now, if you want to design a game - computer, tabletop, LARP, whatever - keep these things in mind. If you want to lock players together, give their characters some level of shared resources of whatever type is likely to cause the emotions you're looking for. Give players a jointly owned child (another player, ha!) and suddenly you have a remarkably powerful relationship popping into existence out of nothing - even if the players don't like RPing.
Think of it like this: by forcing players to share resources (either cooperatively or competitively) you can produce incentive to RP. You may have to "echo" the resources into other resources - produce a tangible "end result" of working together - but that's no problem.
Think about a MMORPG with this kind of shared resources: you can only have a home if you can find other player(s) to cohabitate with you. You can only have a child if you can find a player to be the other parent, or if a player agrees to play the child. Etc, etc, etc.
Of course, these resources need to be ongoing situations: resources need to be difficult to keep. If they have a "solution", they aren't going to be very interesting. Children always cause trouble. Politics always get you in hot water. Colonies grow and have issues. Romances aren't fire and forget.
How would you design a multiplayer game using these ideas?