Diffusion of cell phone viruses and network (dis)externalities
| Gabriel |
Slashdot linked to an interesting paper and appendix by Barabasi and some of his colleagues. The paper is one of several interesting things I’ve seen recently on the structure of social networks and the diffusion of innovation. The social network at issue here is very complex and involves subtle combinations of bipartite networks (our phones use the same OS), spatial networks (I walk by you and get your bluetooth virus), and plain old social networks built up from dyads (I call or text you). The analysis actually reminds me a lot of both some work in progress being done by two friends of mine and the 1997 Organization Science paper by Abrahamson and Rosenkopf, “Social Network Effects on the Extent of Innovation Diffusion: A Computer Simulation.”
One of the cell phone paper’s most interesting predictions is that the current diversity in cell phone operating systems helps protect us from viruses but when we achieve more conformity (as seems likely) things will get really ugly really fast. This resonates with our earlier experience in personal computers but also suggests an interesting dynamic for adoption of cell phone OS themselves.
By several orders of magnitude most of the viruses plaguing PCs target Windows. I’ve heard arguments that Windows has an inherently more insecure architecture and as a non-computer scientist I’m not really qualified to assess these arguments. However as a sociologist I can confidently say that an equally or more important reason so many viruses target Windows is the same reason legitimate software developers prefer Windows, it’s got the biggest user base. Likewise, the reason that most users/OEMs choose Windows is because it has so many legitimate developers writing for it. Now most users just buy antivirus software and hope that it works, but a relatively small number of users are so frustrated by Windows viruses that it contributes to their switching from Windows to Mac or Linux (platforms that are allegedly intrinsically more secure and definitely have smaller user bases). So density both cements and undermines Windows’ dominance.
In diffusion and economic history we’re used to seeing network externalities as something that promote conformity but we forget that there are also negative network externalities (a.k.a., “congestion costs”). The really weird thing is that you can have both network externalities and congestion costs simultaneously and this usually feels subjectively like you’re stuck between a rock and a hard place. For instance large cities have network externalities like robust labor markets and diverse services but they also have congestion costs like high rents and soul-killing traffic. When I visit my parents in the valley and it takes me over ninety minutes to drive back to UCLA or when I look at real-estate listings and see that (even after the crash) a 2 bedroom house on the westside is $700,000, I find it very hard to convince myself that the traffic and the property prices are just part of a package deal with the things that attract me to living in Los Angeles.
You could easily see the same thing happening in a few years with cell phone OS. If you’re a phone consumer (or a service provider or a handset manufacturer) and the dominant OS has more apps but also more viruses, you may very well decide to buck the trend and choose a less popular platform. Depending on how much people desire app compatibility versus fear virus attacks you could imagine an equilibrium much like that for personal computers, with the dominant OS getting 90% market share but never reaching 100%. More radically you could imagine that there is no equilibrium but rather a perpetual dialectic with a high turnover of temporarily dominant operating systems as we adopt for the network externalities but flee the congestion costs a few years later. Personally, the latter wouldn’t surprise me as (unlike software designed for Windows personal computers) most phone apps are pretty useless and I’m thinking most people would rather have a phone that’s virus free than one that works with their lightsaber app.