Technology

Tip-toes in the mobile application minefield

The mobile technology landscape is a varied and technologically heterogeneous.  As far as device capabilities have come in the last 5 years, handsets still host various run-time environments that present fragmented options for innovators.  While iPhone has been the clear market leader in smartphone applications in the past 2 years, there are a myriad of takers for the mantle of second place between Google’s Android, Blackberry, Microsoft, and the fading PalmOS.  The number of choices is good for consumers – but how do you pick a horse as a mobile application developer?

This fragmentation is certainly the least appealing facet of the mobile application market.  The investment in the construction of a mobile application is staggering.  Tomi Ahonen (author of Communities Dominate Brands) did a recent breakdown of the economics of the iTunes App Store and concluded that on average it costs over $35,000 USD to build the typical iPhone application.  But that same average application will only net $682 USD in revenues per year.  Compare that with the cost of developing a web application targeted at mobile devices – about 1/10th that cost.

Now think about developing an Android, Windows, Blackberry, and iPhone port of your mobile application.  The numbers are as prohibitive as they come.

The high stakes here are incentivizing bad behavior: Apple’s license agreement for iOS development being the prime example of defensive maneuvering in the fragmented market.  Not only are we being cornered into repetitive development for multiple platforms – but now extra steps are being taken to ensure that we have to.  Adobe’s model to metacompile flash applications for iPhone deployment became a lightning rod for Apple this past spring.  PhoneGap skirts metacompilation by providing a development toolkit more akin to a scaffold, but consistently lags behind in support for more recent device features.

Fragmentation of desktop platforms has always been a far less limiting problem. Virtualization and tools for cross-platform compilation has enabled developers to target a ubiquitous number of platforms or more often than not, simply disregard the problem.  Furthermore, the development of rich internet applications delivered by the browser has provided a natural cross-platform channel.  With the evolution of the HTML5 standard and it’s continued support by big players like Google and Apple – the ability to take a “write once, run anywhere” mentality to application development via the web browser will continue to be the disruptive force in software delivery as deployment costs continue to be pushed down.

The most ubiquitous mobile browsers – Safari and Chrome, both support much of HTML5 for their host devices.  So why haven’t developers turned to the browser to solve the problem of mobile deployment in the same way they have for desktop applications?  As big as the promise that have been delivered by RIA development has been, there have always been significant limitations – ones that mobile environments exacerbate:

• Connectivity:  Browser-based applications have long required that internet access be reliable.  Browser security prevents access to a users local storage (with Google Gears innovating around this limitation, though at the cost of a plug-in), and thus a lack of connectivity meant a lack of access.  In mobile development, this is a consistent concern.  HTML5’s native features for controlled local storage will solve this;
• Access to native environment:  Likewise with connectivity, the browser has always been severely restrictive (in fact, entirely restrictive) of access to local hardware like built-in cameras, microphones, or external peripherals.  The compact and feature rich accessories in the modern smart phone are a particularly high-value draw to mobile application developers who can only access these features through the native APIs.  Not having access to this from the web environment is certainly secure, but it completely removes some of the value of developing mobile web applications;
• Performance:  While this has improved, web applications are powered by inherently slow JavaScript.  Google’s Joel Weber compared noted in 2008 that the web application is to a native desktop app as a Commodore 64 is to today’s PC.  And while the processing power of the smartphone is leaps and bounds better today than it even was 3 years ago, it still is extremely limited;
• Commercial viability: The most important limitation is also the business problem – how do you monetize web applications?  It’s certainly something that has been done – otherwise the growth of web applications targeted at desktop users would not have exploded as it has.  But it has proven tricky in the mobile space.  The various App stores have proven effective in creating a marketplace for native applications by providing a simple deployment and distribution model.  This channel has allowed mobile devices to put apps at the user’s fingertips, something unmatched by web applications.

Until these limitations are addressed, mobile applications will continue to be loss centers or complimentary pieces of larger application ecosystems.  If mobile deployment is ever to replace or even rival the ubiquity of desktops, they will need to take the human factors learning’s of the past years that have enabled such rich mobile applications and apply them to a deployment and commercial model that enables platforms solely targeted at mobile to exist profitably on their own

Ryan Norris is a technical architect at Medullan. He has a background working with physicians and clients in the eHealth space in developing and delivering solutions that better interconnect providers and create more informed patients through extensible service architectures and highly-usable interfaces.