It's really not, but at first it is. I am talking about social networking here. Initially, you joined Facebook and set up your profile and waited. It was about you and what the system could do for you. Then you became a more active contributor and collaboration happened.
Several weeks ago when I was blogging more regularly, I talked about automating affinity -> having the system tell you people you SHOULD know rather than people you may know. Ultimately, this requires you to have a profile in place with enough semantic information in order to generate a machine interpretable profile. There are a few algorithms / semantic languages forming to assist in determining affinity. FOAF, APML and SIOC all seek to provide a means for determing things or people you might be interested in.
One of the neat features we're rolling out in the release of our social networking platform at McKesson is the integration of social search results with traditional search results from our intranet. When a user performs a search, in addition to finding regular content, they will be presented with "Related People". For the searcher, we've implemented discoverability simply by hooking into their search request. For those who fill out their profiles, we have containers in place that allows them to drop information into containers (about me, certifications, customers, etc) that are then indexed appropriately and served up based matching search terms. This looks like a recipe for successful matching of expertise & interests – we'll see how it turns out over the next several months.