Unexploited Network Externalities

What’s up with the proliferation of networking websites? I joined Friendster a few months ago. Then a friend invited me to join Orkut, which I did (it turned out to be superior to Friendster in various ways). But lately, several people have asked me if I’m on MySpace, and today I got an invitation to join Hi5.

There’s something rather inefficient about having multiple networks. I don’t want to maintain profiles on four or more networks – one should be sufficient. Moreover, the whole purpose of these networking services is to multiply connections as much as possible, thereby linking you to a larger web of friends and contacts. If people choose among the services instead of joining them all, networking possibilities are limited. I have some friends who are on Friendster but not Orkut, and some who are on Orkut but not on Friendster. If they were all on the same network, they would all be connected through me – but not so with the competing networks. (At some point, I tried to recruit all my Friendster friends to Orkut, but not all took the bait.)

I would expect network externalities to lead to the emergence of a single dominant networking service (at least for any given social function, such as business or dating), because the utility of a network is increasing in the number of other people already on it. But that hasn’t happened yet. Maybe it never will, if people choose different networks to begin and then (like me) are resistant to changing over. As Leibowitz and Margolis taught us, network externalities don’t lead to instantaneous or inevitable convergence, and in the meantime there can be competition among standards. For instance, the QWERTY keyboard existed alongside other keyboard layouts for several years before attaining its dominant position. The advantage of the competition is that it forces the providers of the different standards to provide a higher quality product or service (note that QWERTY’s alleged inferiority to Dvorak is a myth – read the Leibowitz-Margolis article). But the downside is the inefficiency of forgoing the advantages of an integrated network in the short run, and possibly the long run if convergence never occurs.