He wants social networking to solve four problems, three of which begin "find people...." And he calls for some focused experimentation.
But what I really like is this paragraph:
These are complex problems, and they will require the development of processes that are suited to dealing with complexity (most software is designed to address merely complicated problems). We're not very articulate, after all, at expressing who we're looking for, or even knowing what and who it is we're looking for (though, of course, we believe we'll know it when we see it). Chemistry is often more important than logic in making lasting and meaningful and effective relationships, and in finding the 'right' people.
"Complexity" in this case is code for "I don't know what I want." Like Pollard, I believe the people problem will have to solved partly offline because the organic chemistry there is mighty hard to duplicate at the keyboard. You know -- the part where you think to yourself "she's got the most beautiful brown eyes," which is followed closely by "I think my heart's in trouble."