Unsurprisingly, the argument that Android is the new Windows seems like it just won’t go away. So much so that it received as much attention as any topic in Tim Cook’s recent interview with Businessweek. I recently wrote about this topic in Unlocking the Value in AAPL, however it is substantial enough to deserve another look from a slightly different angle.
The basic idea is that we are experiencing another round of platform wars in the industry which will inevitably lead to a winner-take-all conclusion with Android crowned the victor as Windows was in the 90’s.1 The reason this view is so widespread is because the narrative is simple and compelling: network effects form naturally around one technology platform in a category, leading to lock-in and switching costs.
These forces are at work in the industry today, but not in the way that a simple look at history and current market share implies.
Most of the software relevant to the PC platform war in the 80’s and 90’s was shrinkwrap productivity software. This software tended to compete on feature lists resulting in bloated, complex software that was difficult and costly to develop and maintain. Development tools and APIs in use at the time were primitive by today’s standards further increasing development cost and difficulty.2
At the same time that development costs were very high, the desktop software itself was the product. Software development costs were the total product development cost. Each supported platform increased the total cost of product development significantly and was therefore necessarily scrutinized carefully. They often simply could not afford to support a minority platform. As consumers opted for the majority platform which enjoyed more software titles, a network effect was created resulting in further preference for that platform by developers.
In today’s mobile environment this is not at all what we see. The vast majority of popular mobile apps rely upon an internet service of some kind (backed by an advertising or subscription business model).3 Developing and operating the server back-end of these services make up the vast majority of product development costs. The result of this is that incremental support of minority platforms is a very small portion of total product cost compared to what it was for shrinkwrap software companies. Today’s mobile developers have a strong incentive to support minority platforms with a material number of users. This is why a large number of 3rd party services are available even on Windows Phone, which has a tiny share of the market.
Where we do see platform exclusivity and preference today, it is usually in cases where the team is small and opportunity cost is high. It simply isn’t possible for these teams to develop for multiple platforms at once (or at least at launch). In this case we see a developer preference for iOS, which has a user base with access to the latest version of the platform, and who spends more money on apps and services.
The other primary segment where we see platform exclusivity is in the high-end of the mobile gaming market. Console-quality games push device hardware to its limit. Many Android devices simply do not have hardware capable of delivering these experiences. Those that can deliver a console-quality experience are are more costly and difficult to support because of fragmentation (especially GPU-related fragmentation)4. Add to this the greater willingness of iOS users to spend money on apps and the result is that iOS has an advantage over Android in gaming.
In the desktop era most software was purchased by centralized corporate IT departments. With centralization comes a tendency towards standardization. Microsoft made early inroads with these customers when they secured a deal with IBM to offer DOS on IBM's PCs, where it was the low cost option. Over time, DOS and later Windows became the de facto standard favored by IT managers.
Many people first encountered computers at work on an IBM compatible PC running DOS or Windows. When they decided to purchase a computer for home use the choice was obvious.
The mobile market today is completely different. Purchasing decisions are usually made by individuals. Most of the usual considerations that apply to consumer markets are relevant. Smartphones are as much about media and entertainment as they are about efficiency and productivity. Standardization simply does not matter at all to consumers as long as the device enables them to do what they want to do with it.5
The result of this is that brand and user experience can, and indeed do, play a significant role in the choice of the consumer. This difference has played a major role in the phenomenal success Apple has achieved in the mobile market. As Ben Bajaran observes:
Getting inside the mass market consumer psyche is truly an art form. It is one most companies truly fail to grasp. It is also something most disruption theory fundamentalists have a hard time comprehending.
Interestingly, Apple also has an advantage in the corporate market. Security has become a major concern and Apple's vertically integrated approach allows them to avoid many security issues that afflict Android.
They have also quietly done a great job of adding features desired by IT departments to iOS. It is much easier and less expensive for IT to support iOS devices than it is for them to support a highly fragmented Android environment.
During the desktop era the cost to a user for switching from one platform to another could be quite significant. These costs came in a variety of forms.
Most directly, a user may have invested significant money into 3rd party software that only exists on that platform or is only licensed for the platform. Switching would mean giving up those licenses and finding and purchasing alternatives on the new platform.
With the tendency towards ad and subscription supported services we see today this is not the case. Either the app is free with ads or the user pays a monthly fee which includes access to a client on any platform the user chooses (Evernote and Netflix are a good examples of this).
Another source of switching cost in the 80’s and 90’s was proprietary data formats. The user may have a large amount of data stored in formats which could only be read by software running on their current platform. Sometimes it was costly or impossible to convert this data to a format which could be read on a different platform. This cost could manifest in the form of money, time, or data loss.
Increasingly data is being stored in internet services rather than directly in local files. These services are accessed via client apps on the user’s platform of choice. Switching is as easy as downloading a client app on the new platform. In the cases where data is stored locally, we are blessed with a plethora of standard, interchangeable data formats. Bringing data to a new platform can still be troublesome at times, but it is not nearly as significant a problem as it once was.
There is more behind network effects and lock-in than dominant market share. Android does not enjoy the same advantages that Windows did in the desktop era. A network effect favoring Android is not likely to take hold even if it continues to maintain a significant market share advantage.
In fact, if we squint just a bit it appears as if iOS may have some important advantages over Android, at least for now. This does indeed appear to be playing out in practice where we see a meaningful number of users switching from Android to iOS but few switching from iOS to Android. How large these advantages are and how impactful they are to the evolution of the market is something that only time will tell.6
The naivety of this view is exemplified in the fact that although there was a time (just before Steve Jobs returned to Apple) when it looked Microsoft would “win it all” this did not actually happen. Apple’s brand and culture had enough life left that Steve and the team were able to pull off the most dramatic and successful turnaround in corporate history. Apple’s PC market share has been climbing ever since and with Apple claiming nearly 50% of profits on the hardware side of the industry. This does not look remotely like a winner-take-all conclusion to me.
Tools and languages were much lower level than those commonly in use today. It was not uncommon for assembly language to be used in performance critical pieces of code. This is almost unheard of at the app level in the modern ecosystem (with most exceptions being in audio, video and gaming engines). Furthermore, modern programming techniques such as object-oriented programming had yet to achieve widespread adoption in industry. NeXTSTEP, the predecessor of OS X and iOS, was the first serious attempt I am aware of to industrialize object-oriented programming.
As Ben Thompson noted, app store economics seem to be pushing developers in this direction whether they want to go there or not.
The gaming business has a cost structure much more similar to the shrinkwrap software market than most other app businesses. The majority of product development cost is in the software that runs on the device.
Of course the development and widespread adoption internet standard protocols and data formats has played a significant role in facilitating the interoperability that allows consumers a wide range of devices to choose from. Interoperability was a huge challenge while the desktop market was forming and is not such a challenge today.
Obviously I believe things will work out well for Apple, one way or another.