Of Network Effects and Fads
The brief history of the Internet is peppered with episodes of revolutionary products that eventually turned into fads. The obvious point in case is the dotcom bubble of the late nineties. According to Kunz (2008), the ultimate cause of the fatally inflated valuations of companies such as pets.com or eToys.com was blind trust in network effects and, in particular, in Metcalfe’s law (Shapiro & Varian, 1999), which in its general form postulates that the value of a network increases more than linearly (n2, nlog(n), etc.) with respect to its size. About ten years after the dotcom bubble, friendster.com, the world’s first general purpose SNS, was first adopted and then abandoned by users en mass, ending up being called “one of the biggest disappointments in Internet history” (Chafkin, 2007).
While network effects help explain the increase of SNS value as a function of network size growth, Zipf’s law and Dunbar’s number point to forces that are actually inversely proportional to network size. Zipf’s law (Briscoe, Odlyzko, & Tilly, 2006) states that each additional member in any series of items (such as an additional SNS contact or “friend”) has a predictably diminishing value. Offering an evolutionary psychology viewpoint to the SNS topic, Dunbar (2010) suggests that there is a cognitive limit on the number of people with whom one can maintain stable social relationships. The value for that number ranges from 100 to 230, while a middle-of-the-road estimate, often cited by many authors as “Dunbar’s number”, is 150 (Gladwell, 2002).
McAfee & Oliveau (2002) indeed point out that Metcalfe’s law cannot be regarded as “true ad infinitum” for there are “forces that put the brakes on network effects, i.e., saturation, cacophony, contamination, clustering and search costs”. The consequence, they argue, is that the growth of a network may eventually slow down, or even reverse itself, as it becomes too large.
However, network effects and their network-size related counterparts described by Zipf’s law and Dunbar’s number are not the only factors that affect the fate of a product. This problem runs deep in the marketing literature where two models stand out, namely, PLC and diffusion.