Facebook's founding premise wasn't easy onboarding. It was getting laid at Harvard.
And then getting laid by someone who knew someone at Harvard, or another selective-admissions university.
The initial founding cohort was one that engendered strong aspiration of others to join that group, though of course, in others joining the group, the specific appeal waned. A loss in per node value however was offset largely by the network value of the group as a whole.
(I've argued that this dynamic of balancing network value with network costs is key to how social networks grow, are limited, and ultimately decline. In particular, each new member adds a variable value (not all nodes are created equal) but a fairly uniform cost, in the form of additional friction, information overload, and the like. A large-scale network attacks those costs ruthlessly through hygiene functions, including content moderation and the like.)