Why the mobile operating system market won't tip

Fear of tipping dominates information technology. The most famous example of this was PC and Mac where an orientation towards business led to a greater variety of software titles being available for Windows than Mac which wasn’t dislodged until the use of the Internet became the primary use case for a computer. But today, where the primary use case remains on-machine software, Microsoft dominates (i.e., in Enterprise). (I should add that it didn’t harm matters that Microsoft was integrated in both the operating system and the key app — Office — that itself had network effects).

Many have expected the same to be the case for mobile operating systems for some time. This led to predictions of dominance for Symbian, Palm and Blackberry in their day and to continual, somewhat amazing claims, as to Apple’s monopoly. It was also the rationale behind Window’s failure to get traction in mobile.

The theory that leads to a dominant operating system is simple. Operating systems can be developed with different attributes. If it turns out that a particular attribute that is available on one platform but not another turns out to have value to consumers and, moreover, there are network effects associated with that, then the market may tip. For instance, if consumers love Mario and Mario games are only on Nintendo, then they buy Nintendo, developers focus their attention there, and the market tips. But notice that, at some level, there has to be a draw to a platform that coordinates consumers and then developers there. Absent the draw and the market won’t tip.

Standing above it all, it has been hard for we observers of the industry to notice the draw that might tip the market between iOS and Android. In 2007, many believed that BBM would prevent the market for tipping from Blackberry but that turned out to be incorrect. iOS was, of course, first to the party and could leverage the iPod so there was some reason to believe that ‘locked-in’ consumers would be the draw. But the app market was in its early stages and apps were so cheap relative to the devices they were on that this didn’t happen. Android had plenty of opportunity to bring new customers into the market. In 2010, this was repeated with the iPad that again had people wondering whether the ability to have apps across iOS devices would be the draw. But again, the iPad was a sufficiently new market and use concept that that turned out not to matter.

At its heart, tipping relies on their being a compelling and somewhat broad-based use case for developers to choose one platform over another. User coordination could be a reason as could openness (Apple’s famous walled garden) or fragmentation (Android’s lack of a single version for developers to write for).  But while individual developers might have preferences, there were apps being developed on both platforms at a healthy rate. There would be no ads claiming that you can’t do something on a particular platform to sway consumers. Even where that was the case (such as the use of NFC), the lack of multiple platform adoption was seen as a barrier and not an opportunity for developers.

At an NBER conference yesterday, I saw a paper that examined the data rather than just our broad observations. Tim Bresnahan, Joe Orsini and Pai-Ling Yin gathered an impressively detailed dataset of apps on iOS and Android to allow them to see what sort of apps ended up on both platforms. If the most popular apps on iOS and Android were distinct, then one could say that the market had the potential to tip as the consumers preference for what a platform gave likely outweighed what the apps were. On the other hand, if the most popular apps were common, then tipping would be unlikely as consumers would be less concerned about the platform and more about what apps they could access. As it turns out, and this will surprise no one, it was the latter that was the case. The most popular apps are on both iOS and Android. Thus, app availability would not be a reason for consumers to choose one over another taking away the mechanism that could give rise to tipping.

The authors also provided a theory as to why this was the case. The argued that it was costly to port an app between platforms and so you either (i) had to have access to good marketing opportunities as you moved to a new platform or (ii) already had to be well-known on one platform . They were able to get an insight on this by comparing apps developed by established firms with those by independent entrepreneurial developers. The latter were less likely to have apps on both platforms than the former.

This actually has implications for developers. If there was just one platform out there, then established and independent developers would be able to think of developing for the same market and be on a more equal playing field. However, when there are two platforms, this gives established firms an advantage (twice the market) compared to independent ones. This also suggests that the platform that can most effectively match apps to consumers will be advantageous for entrepreneurial developers.

Finally, this analysis also suggests that Apple’s surprisingly low share price valuation being based on the idea that if it doesn’t innovate it will suddenly lose it all, is incorrect. That fragility only makes sense in a winner-take-all mobile environment and it doesn’t look like that is the environment we have here.