When people connect with others on a given social network, they are conscientious about whom they will connect with, because an exchange of information, both immediate and ongoing, will result from the connection.
Just as in offline life, people don't like to send and receive information to and from random people; their relationship with those people is crucial. The things you say to those you encounter on the street will differ from the things you say to familiar people in your own home. Conversely, your interest in what strangers have to say will differ from your interest in what your friends can tell you.
The types of relationships that people experience aren't simply divided between friends and strangers; they are manifold and impossible to label with complete precision. Strictly speaking, no particular relationship gets formed between two pairs of people because nuances invariably come into play. You may be office mates with both Tim and Joe, but you're a bit fonder of Joe because he invites you to lunch.
Relationships also aren't perfectly symmetrical. While you think warmly of Joe, he might think you're kind of a jerk and only asks you to join him because he's interested in your sister. Consequently, any label and assumption of symmetry you assign to a given relationship will constitute an approximation at best.
However, approximations are useful when trying to identify the type of relationships a given social network should or does facilitate, because individuals themselves map their relationships to approximate groups. And despite the efforts of designers to diversify the types of relationships that thrive on their networks, consumers tend to view each social network as primarily suitable for only one of their groups.
Understanding a group to be simply a set of people who share the same approximate relationship to each other, we can identify an array of such groups that might be facilitated by social networks. On a high level, there are expansive groups of people you've met and people with whom you've simply communicated. There are also people you admire and people you want to impress.
More specifically, there are acquaintances from colleges, companies and organizations. There are peers in your industry and collaborators on your specific projects. There are close friends whom you see weekly as well as old friends from high school you see once a year. There are family members and teammates. And there are folks you may or may never have met but who share the same interests as you.
Whatever the group and however specific, it needs to have enough members who both find the group important and desire better ways to share information with each other to warrant a dedicated network. And its importance is often tied to the group's size and its predominance in members' lives. Facebook initially took off among college (and then high school) students because it intensified the already intense relationships that existed within academic communities. Likewise, Twitter and LinkedIn initially thrived by bolstering professionally important relationships within the Silicon-Valley-centric tech scene.
Furthermore, when someone encounters a new network, it's important that they can actually identify which of their relationships it will facilitate and how they will benefit as a result. Otherwise, they are presented with the communications equivalent of a hammer without a nail; they won't know what to do with the social network and it will seem pointless. Similarly, if you signal that the network is meant for a particular type of relationship they don't have, want or care for - or if they feel as though they don't have an unaddressed communication need for that relationship - they won't feel compelled to participate.