|
2:35 p.m. - 2003-09-19 however! the practical properties of this situation mandate a fairly broad study. that is: any individual's friendster growth curve is going to be affected by a number of unique factors. these include the number of friends they have who are already on the service checking it regularly, their willingness to add fakesters (who tend to be very widely connected), how much their immediate friendsters bother to get widely connected, and (most difficult to even think about modeling) the degree to which their first two layers of friendsters are connected *to each other*. to name but a few. since these distortions are mostly unavoidable and achingly complex where they can be dealt with, the best approach would be to study a great number of individual users and hope to find general "pattern" curves from a synthesis of the results... � � |