You have to offer something different, and specialization is a great way to build mass. Friendster and MySpace were open to anyone from the start. It's harder to manage growth of a community in that way.
Facebook, by restricting its use to college students first, then to graduates with alumni email addresses, built a close system of people who were less wary about sharing personal details. By the time they opened to everyone, they already had a very strong user base that liked what they got out of it, but were moving into the "real world" and liked the idea of connecting with other people in their lives.
The lesson to be learned is that doing one thing and doing it well, or doing something for one small subset of people and doing it well, is a great way to develop a strong network effect. The network was built slowly but surely. The other networks took anyone and everyone and became somewhat of a mess.
All had strong network effects, but they did it in different ways. MySpace was strong because of the level of personalization and the music shared on MySpace, but very few people were well-positioned to maintain high quality inputs to maintain long-term networks. Friendster couldn't scale its technology and made it too difficult for friends to stay connected in any meaningful way. Facebook concentrated on scalability while providing more and more ways for people to stay connected (messages, chat, ecommerce).
If you're going to build strong network effects, there has to be a core of devoted and popular users who are willing to use and broadcast on the social network. There also has to be a residual value proposition for people to follow that core or the activity on the social network. And the network itself needs to continue to evolve so that people can continue to maintain strong and varied connections over time.