Party-nomics, Facebook and Positive Network Externalities

Have you ever been party-hopping with friends, found a party or a bar that wasn't highly occupied, and then (maybe due to a unanimous decision, or maybe due to the whining of a few) immediately left to seek a more popular venue? If so, you have helped fulfill a self fulfilling prophesy. In this kind of situation, if more people are willing to stick around at the unpopular venue, it can have had a chance to become popular. But with everyone leaving 5 minutes after getting there, the place will remain, in the parlance of our times, "dead" (as in "dude, this party is dead, let's split"). This phenomenon is very well described by what economists call "positive network externalities". Unlike a good without positive-network externalities such as a banana, where one person's benefit from eating a banana does not depend at all on how many other people eat bananas, a person's benefit from going to a party greatly depends on how many other people go to that particular party. This is because as another person joins the party, benefits (positive externalities) go to all the other partygoers. The more people there are at the party, the more chances there are for exciting social encounters. Therefore there is more benefit both for people already at the party as well as for prospective partygoers. This self-fulfilling prophesy will attract people to the party up to the point when things get overcrowded. After this point, each person's benefit will actually decrease as each additional partygoer arrives. With the number of partygoers on the x axis, and the individual enjoyment of the party on the y axis, a graph of the "total fun" of a party (if we could quantify it) would look something like this:

Note the steep drop-off at a certain point, when the number of partygoers reaches the maximum capacity of the party's physical location. Clearly people like a good party, but no one wants to get squished. As you can see, the appeal of a party is a function of the number of people already there, and this determines how many people will want to join that party in the future. The self fulfilling prophesy is in full effect. Because of this, the success or failure of a party depends largely on the number of people attracted in the initial stages, and this can depend largely on chance. When a party has reached a "tipping point" of popularity through word of mouth, it can fill up to the point of reaching the maximum capacity of the party's physical location. Then there will be spillover to the less popular parties, and a new equilibrium will be reached.
But what about a party with no physical location, other than tiny ones and zeroes occupying a server? Yes I am talking about the internet, and specifically the huge social gatherings that occur 24/7 on social networking sites like Facebook and Twitter. Just like a parties, these sites bring positive network externalities to their users. Meaning, the more people there are on Facebook, the more beneficial it is for a new person to join Facebook. But unlike parties, huge websites do not face the constraints of physical space. So, roughly graphing the benefit of being on a social networking website in response to number of users would look something like this:
Rather than a drop in total fun occurring at the point of some number of users as we saw with the physical party, there is simply a levelling off. There are diminishing marginal returns to fun here, because, for example with Facebook, you can only have so many Facebook friends (5000 as of today I believe, and even fewer you really interact with on a regular basis). But there is no drop-off in benefit, because there is no restriction on capacity. Unlike a real physical party, this party on the internet shows no sign of slowing down. Facebook has become a fixture, and it would be very difficult for another company to come in and compete with it on its own turf. A new site would need to provide amazing new benefits to compete with the massive positive network externalities Facebook has built up through its user base. And why did Facebook build up such a large base of users? Partially because it's a very well designed site, but also because of the initial luck of the draw. Mark Zuckerberg's party got the initial rush of partygoers it needed to sustain itself and grow. Without any capacity restrictions, it seems to be here to stay. Even if there were a better party next door, people would probably ignore it.