Friendster launched in 2002 and rapidly amassed over 5 million registered accounts by January 2004. However, at that point, Friendster had already witnessed massive friend-ship inflation. Over time, members had accrued a large number of connections yet there were no metrics to indicate the weight of the connections. The connections were typi-cally binary: friends or non-friends. It is so convenient to befriend on Friendster that some users create fake profiles in order to attract other users who share similar interests and social and cultural background. These were called Fakesters. Friendster considered that fakesters devalued the social network and therefore decided to remove them from the site. The massive removal of fakesters without consulting with the users who created them annoyed a lot of users. This so-called Fakester genocide exacerbated the situation and had driven many users to MySpace, which was then a new competitor to Friendster.
It should be noted that today Friendster is still popular in the Philippines and South-east Asia. This suggests that technology was not the only major factor to be blamed for the decline of Friendster in the US. In fact, when MySpace emerged, Friendster
had just made an improvement on their system by adding more servers and chang-ing the programmchang-ing language from cumbersome Java to light-weight PHP[106]. The same argument applies to Facebook when it just arrived in the arena of social network competition. But eventually Facebook outnumbered MySpace, which had previously outnumbered Friendster. Today, people have already begun talking about Facebook fatigue, given that the number of registered users has dropped slightly. Now with the rise of Twitter, Facebook may well be another victim of friendship inflation.
One of the main attractions of social network sites is to make new friends by leveraging the existing connections which are assumed to be reliable. Friendship inflation suggests that users will have more contacts than they actually have in the real world. In a social network with nodes densely connected with each other, it is very difficult to make meaningful connections because the substantial cost of discerning the real connections.
The sites will eventually lose their original advantages. SNSs may gradually re-position themselves in competition with the new sites. For example, MySpace looks increasingly similar to a media site by providing videos and music. Facebook looks increasingly similar to a communication tool by providing web-based instant messaging and twitter-like activity updates. When they fail to reflect the evolution of users’ social network and capture the real-world network, if there are new alternatives, users may just leave the old site and switch to the new network in search of genuine connections. Here, the balance point is the effort to distinguish the genuine contacts from strangers and acquaintances in the old established social network site, versus the effort to invite friends to the new site. Because of the static link, it is always easier to establish connections on the site.
Therefore, as the old site becomes more crowded and less trustful, a new site may be more attractive.
4.4.9 Summary
Hyperfriendship social networks provide no mechanisms to verify the connections be-tween users. There is no rule to which users must adhere in order to establish new con-nections. Users can make new connections without much cost. This leads to a rampant increase of the number of friend connections. The rapid growth of the hyperfriendship social network collapses the context and social environment where users interact with each other. The increase of weak yet persistent connections brings a whole range of social implications and ramifications, complicating the issues of fakesters, privacy con-cerns, multiple personas, spamming and phishing. Social network sites seek to tackle the issues by using different technologies, human interventions and even resorting to law, but with static link as the backbone connecting method of their network, most SNSs are fighting a losing battle on the balance between publicity and privacy. The hyperfriend-ship network can be saturated but users’ real-world networks are still evolving. When an alternative social network emerges, users who are fed up with the old one simply switch
to the alternative. The critical issue for most social network sites is that they attempt to constrain users by the technologies they have developed. The technologies include static link and contact categorisations. System designers hope that users will use these technologies and use them in a way that conforms to their intention, which, according to Friendster and Facebook, is to encourage genuine identity and connections. This vision is shared by most social network sites. However, when it comes to friendship collectors, fakesters and fraudsters, system designers simply ignore the creativity of users both real and fake.
4.5 Discussion
In this chapter, we presented statistical evidence at the macro level that supports friend-ship inflation in most social network sites. Two significant statistical properties are no definite cutoffs and dissortative mixing patterns. The theory of friendship inflation is supported by our nearly three-year observation of the Facebook users in the network of the University of Southampton. We discuss the issues arising from friendship inflation.
The problems include unreliable connections, undiscernible hubs, lack of peer pressure, spamming and phishing, inaccuracy of network algorithms and information overload.
We argue that friendship inflation is one of the major reasons leading the decline of social network sites. To support the argument, we cite the case of the rise and fall of Friendster, and the battle between MySpace and Facebook. We therefore call for engineering mechanisms to alleviate the problem of friendship inflation. In the next chapter, we will present the algorithm of ActiveLink, which aims to solve the problem of friendship inflation by identifying meaningful social connections.
ActiveLink: Identifying
Meaningful Social Connections
5.1 Introduction
Most issues confronting social network sites come from the fact that they are mod-elling people’s dynamic real-world connections in a static framework. The static model adopts an implicitly stationary view of relationship formation in which connections, once formed, are permanent – thus entailing a zero maintenance cost[26]. The static model ignores the properties and topologies of real-world social network, and fails to reflect the evolution of the network. Unfortunately, there is little academic research carried out to address the fundamental issues of the static system, despite more and more commercial and experimental social network sites available. We propose the ActiveLink algorithm, a communication-based method that aims to identify the genuine connections.