This is a guest contribution from Tim Closs, chief technology officer at Ideaworks3D (pictured right).

App Stores: Do we really need any more musings? Fragmentation: Surely this is so well understood that it’s almost an industry cliché?

In the stampede of recent OEM, MNO and platform-owner app stores announcements - and the anticipation of a more widespread replication of the 'Apple model' - little attention is being paid to the serious development challenges that will have to be solved if such ecosystems are to flourish.

Let's examine some of the pieces Apple has put in place to establish a healthy App Store proposition for developers, and consider how a competing OEM or MNO might be able to measure up to it.

Piece 1: Simple process for developers to register, upload apps, and track app statistics

This is simple web programming. Any competing OEM or MNO can and must tick this box.

Piece 2: Workable business model for developers

The 70/30 split is becoming an iconic ROI benchmark for developers. Any competitor being more greedy risks being sidelined by developers.

Piece 3: Great developer SDK, making it as simple as possible to create great apps

This is where it starts getting interesting. Apple has done a good job with the iPhone developer SDK and tools.

There are some downsides (no support for PC development, usage of Objective-C, difficulty in integration existing C++ codebases) but in general it’s a powerful toolset that gives developers a great leg-up.

Developers can leverage the power of native code and HW graphics acceleration through OpenGL ES 1.1, to deliver smooth and compelling user experiences.

Where does that leave competing OEMs and MNOs? They are currently relying on developer SDKs and runtimes from the 'platform providers'.

For native code, that means (primarily) Symbian OS, Windows Mobile, BREW and emerging mobile Linux platforms. For interpreted code, that means Java ME and Flash Lite.

Let’s not beat around the bush: Symbian, Windows Mobile and BREW are not providing development platforms that stack up against the iPhone.

Rather than enabling developers, the tools and API choices often become things to be worked around - additional problems to be solved. Neither of these platforms provides slick and easy-to-use UI frameworks. Neither of them provides a great desktop emulator, or easy on-device debugging.

What about Java ME? If we’re talking purely about the development tools, then Java ME ain’t that bad (although nobody could seriously argue that the debugging experience matches that of Xcode or Visual Studio).

But for creating great apps? There are too many unknowns for the developer in terms of available APIs on each device; and of course the performance is shocking compared to native code.

What about Flash Lite? Well, someone please show me a great Flash Lite mobile app and then we can discuss.

Whether it’s down to performance issues or versioning issues, or the dubious appeal of ActionScript, I don’t know, but the current reality is that desktop Flash is great for user interfaces, but Flash Lite is not a realistic choice for developers looking to distribute compelling apps across devices.

Piece 4: No need for developers to worry about porting between platforms or devices

Now it gets even more interesting. Arguably Apple and its developer community are in the 'honeymoon period' here. The handful of minor variations on a single device is manageable even for the smallest of developers.

But other OEMs, and especially MNOs are all hosting myriad devices with widely varying form factors, hardware, operating systems and application environments. A developer looking to support all devices hosted by the OEM / MNO is faced with the familiar and depressing economics of device fragmentation.

How much will it cost me to port my app between platforms (for example, from iPhone to Symbian)? How much will it cost me to port my app between all devices within that platform (for example, to support touchscreen and keyboard, software rendering and OpenGL ES)? And critically, how much will it cost me to test across all devices?

For small developers, even acquiring the devices for testing can become a logistical and financial headache.

Over time, it’s likely that this will become an increasing problem even for Apple’s developers. Today it’s primarily a problem for Apple's competitors.

Summing up

The spectre of device fragmentation, and to a lesser extent the inferior quality of non-Apple mobile platform SDKs, puts OEMs and MNOs are at a huge disadvantage when trying to establish a healthy app store ecosystem.

This begs the question of whether any of them have the vision and capability to build a new developer SDK that solves these problems?

To some extent, this is what Nokia has attempted with the N-Gage SDK - to eliminate the fragmentation headache for games developers targeting a number of S60 devices.

A worthy goal, certainly, but if an OEM or MNO really wants to transform its revenues from apps retailing, it needs to transform the economics of apps development for its 'channel' in a more radical way.

At Ideaworks3D, we’ve been developing our Airplay solution over many years to solve exactly these challenges across wide device portfolios. Market awareness of the 'app store phenomenon' has accelerated interest and uptake of Airplay within multiple industry segments.

Who will provide the 'hero SDK' that crosses platform and device boundaries whilst providing great development tools and compelling apps performance?

Whoever does so would surely find themselves in an enviable position in this exploding market.
Tim Closs is chief technology officer at Ideaworks3D Ltd, a leading developer of mobile games and mobile applications technology. Tim had several 8-bit games published whilst still at school. He gained a Maths degree at Cambridge University before returning to the games industry. At I3D, Tim has overseen the creation of Airplay, a binary-portable solution for native applications on mobile devices.