Distilling market noise into market sense

VisionMobile is the leading research company in the app economy. Our Developer Economics research program tracks developer experiences across platforms, revenues, apps, tools, APIs, segments and regions, via the largest, most global developer surveys.

A game of ecosystems: Crashing the Android party

[In a game of ecosystems, it it possible to bend the rules? Research Director Andreas Constantinou examines the few technologies that can bring the Android ecosystem on non-Android devices – compatibility layers, virtualisation or emulators – and their impact in a post-‘Googlerola’ world]

A game of ecosystems: Crashing the Android party

A game of ecosystems

The mobile industry in 2011 is an industry ruled by ecosystems. The Android and iOS platforms are jostling for the top spot in terms of number of applications and developer mindshare. Earlier this year, Nokia had to jump from a burning platform (its in-house Symbian OS) into an ocean of uncertainty (Microsoft’s Windows Phone).  MeeGo has to find foster parents, as both Nokia and Intel (its founders) seemed unable to gather enough momentum. Nokia is re-launching its effort at building the Qt ecosystem through a more ‘open’ platform governance model. Qualcomm is trying to rebuild loyalty in BREW MP by investing in online marketing. The platforms today are fighting for who can build the biggest, most active and most sustainable developer ecosystem.

VisionMobile - App Store figures 2Q11

What determines the success of a platform? We know it’s the wealth of its ecosystem (lets call it developer ‘mindshare‘) as well as its consumer reach (the ‘market share’). It’s also patents, as these legal instruments can create not just barriers to entry (see how Apple barred Samsung tablet), but also market costs (see how HTC, Acer and Viewsonic are paying patent royalties to Microsoft).

Therefore, ecosystem wars are fought on three fronts: mindshare, market share and patents.

Let’s focus on developer mindshare. Developers are extremely critical as ‘platform consumers’. Anyone who has tried to recruit developers through traditional outbound marketing has failed. Developer marketing is about evangelising developers to a higher cause, more like crafting a software-centric religion – what can be called “religion engineering” (see evidence here on Apple as a religion). More importantly, building developer mindshare is extremely costly and tricky – you need software marketing know-how to execute successfully. Which is why many ecosystem-building efforts, including Java ME, Symbian, MeeGo and Palm OS 5, have failed.

Crashing the party 

Developer marketing needs a very different toolbox and billions of investment; unless you can crash the party.

Two types of solutions exist that allow OEMs to piggy-back the Android ecosystem on non-Android devices: compatibility layers from Myriad, OpenMobile and virtualisation technologies from OK Labs, Red Bend, and VMWare. RIM has also been using an emulator approach to bring Android apps on QNX.

Myriad (www.myriadgroup.com) (SIX Symbol: MYRN). is a mobile technology company that offers browsers through to messaging infrastructure, with mobile software deployed in more than 2.2 billion mobile handsets. Based on the earlier acquisition of Esmertec (the early Google partner on the Java virtual machine) Myriad offers an Android emulator it calls Alien Dalvik. Launched in February 2011, the emulator allows “thousands of Android apps” to run on non-Android platforms like MeeGo. Myriad claims that once the APK files are repackaged for Alien Dalvik, applications can run unmodified and with no loss of performance. The company is focused on making its emulator compatible with popular apps like Dropbox, IMDB and Evernote, presumably focused at OEMs that want to bring the out of box apps experience on proprietary platforms.

OpenMobile (www.openmobileww.com) is a Massachusets-based company founded in 2010. The company is privately funded and led by Nachi Junankar, a serial entrepreneur, and Bob Angelo, a software veteran who, as COO of Phoenix Technologies, led the team that opened up the clone business and later created the PC BIOS. OpenMobile has developed an Application Compatibility Layer (ACL) that claims to bring all 250,000 Android apps to non-Android device. The ACL is available for MeeGo (webOS and Windows support are in the pipeline), and claims 100% compatibility as “all Android Apps run exactly as they do on an Android device”. Moreover, OpenMobile says that app developers don’t have to modify, recompile or repackage their Android apps to run under ACL.  OpenMobile doesn’t use virtualisation or emulation, but integrates the Android application runtime into the native OS and supports Android API Level 4 or higher, as well as NDK 6 or higher.

The alternative approach to bringing Android apps on non-Android devices is to use a virtualisation technology from Red Bend, OK Labs or VMWare (see our earlier analysis of virtualisation technologies). Virtualisation allows the Android OS (including the apps ecosystem) to run in sandbox, completely isolated and independent of any other platform (including OEM proprietary OS). Like on desktop virtualisation, the apps from both platforms can surface on the same menu, so that virtualisation remains transparent to the user.

Challenges and the new shape of Googlerola

Can an OEM crash the Android party? Yes, but only with a bodyguard. We suspect that the above vendors will be facing two primary challenges that they‘ll need to muscle their way through.

Firstly, keeping in sync with the Android upstream changes is no small task, as new Android (including NDK) APIs have to be integrated into the target platforms and Google introduces API changes every two months on average.

Secondly, OEMs who use an Android compatibility layer will also be exposed to potential patent lawsuits. Now getting access to an Android Insurance Policy (see our Googlerola article) from Google will also mean complying with Google’s tight software/hardware specs. The one region which seems opportune for Android-compatible technologies is China, where telcos and OEMs have long been trying to develop their own OS flavour, but have been stuck with antiquated OS versions as they haven’t been able to keep up with Android upstream.

So can an OEM crash the Android party? Yes, but you‘ll need some pretty good bodyguards to escort you in.

– Andreas
you should follow me on Twitter

  • http://tabulacrypticum.wordpress.com Texrat

    Andreas, this is extremely useful and timely for me, as i will be hosting a short session on Open Tech Ecosystems at Intel's AppUp Elements next week in Seattle. I will definitely point attendees to this informative article! You've done an outstanding job capturing the critical elements and helped me tremendously in summing things up. Well done.

  • Frank

    OpenMobile and Myriad are not going make any headway on Windows Phone or any of the Nokia-owned systems (MeeGo Harmattan/N9, Symbian) if they continue to target only OEMs – these two platforms will NOT welcome Android support as I'm pretty sure Microsoft and Nokia will see it as strengthening Android while at the same time weakening their own "ecosystems". However end users of these already weak platforms would no doubt take the polar opposite view.

    To that end, what if OpenMobile are able to provide a product without distributing any Android source code at all, thus avoiding the OEM/bodyguard route? Imagine a situation akin to FOSS Linux distributions that don't distribute closed or proprietary binaries, but still allow these binaries to be installed at the click of a button?

    In the case of Open Mobile, imagine they provide an application – which runs on Windows, Mac and Linux, of course – that automatically downloads the freely available Android source *direct* from Google servers, and then applies the OpenMobile proprietary "diffs" to create a "package" suitable for the target OS (MeeGo, WebOS etc.).

    What the end user purchases from OpenMobile (via the various online app stores) is the application that installs on the device, which hosts Android apps but also deploys the end-user created Android "package". If various end-user created "packages" subsequently find their way on to torrents, so much the better, as the package is of no value without the OpenMobile on-device app and would save users the bother of creating it themselves. Who would Oracle sue, tens even hundreds of thousands of end users? Not likely.

    With this scenario, there is no liability for Open Mobile. It should be entirely possible for OpenMobile to create a distribution system based on a specific build of Android source code and then through a system of diffs (differences) provide only *their* changes and none of the original (and legally questionable) Android source code.

    This is certainly possible, though technically tricky, but it would disprove your assertion that a bodyguard is required. Factor that into your argument and it radically alters your conclusion.

    • Andreas Constantinou

      Hi Frank,

      This is a very interesting deployment route indeed. While I 'm no expert in IPR your solution seems to have merit, although there is no precedent for or against such an approach. I suspect that it would take at least a year before such a solution is legally proven or disproven.

      Besides IPR, the main challenge I see with compatibility layers is the technology (and degree of compatibility) is not yet deployed in commercial devices and therefore field-tested, which would not resonate well with the typically risk-averse OEMs.

      Andreas

    • http://twitter.com/openmobileww @openmobileww

      Frank, it would be good to discuss this. Please contact me at njunankar@openmobileww.com

      regards
      Nachi Junankar
      Chairman, OpenMobile

      • Frank

        Replied via the OpenMobile ACL thread on forum.meego.com – PM me if you would like to continue the discussion elsewhere.

  • Jeff Gould

    This is all very interesting, and I thank you for going to the trouble of bringing this information to our attention. But who is actually lining up to use these Android "compatibility layers"? I saw a nice demo of the RIM Android Player. But RIM’s CEO says Android apps will run slower than native. I also saw the Alien Dalvik and OpenMobile demos of Android apps running on MeeGo. Very impressive. But who is using MeeGo? Getting these Android compatibility layers to run on something not based on Linux (like Windows Phone) might be a lot harder. You say OpenMobile has Windows support “in the pipeline”. I’ll believe it when I see it. And notice they don't say "Windows Phone", which might be more relevant than just "Windows". Lots of small unknown tech companies announce marvels that they then fail to deliver. And as you point out, the problem for all the alternative Android platforms is keeping up with the gorilla in Mountain View. It seems that consumers don’t want out-of-date versions of the OS on their phones. This looks like a major reason why OPhone hasn’t set the world on fire, despite being launched by the world’s largest mobile carrier. So my question for you is: what reason is there to believe that Android compatibility layers will ever have any real impact on the market beyond some nice demos on YouTube?

    • Andreas Constantinou

      Hi Jeff,

      IPR concerns aside, I see the drive for "compatibility layers" coming from the likes of Sony Ericsson, Samsung and LG that want to "app-ify" their proprietary platforms (e.g. SHP/Bada for Samsung) and accelerate the conversion of their "proprietary" phone projects into smartphone projects. Still, compatibility layers are high risk projects (as with any early adopter technology) which will meet a skeptic audience amidst risk-averse handset OEMs.

      Andreas

  • http://twitter.com/openmobileww @openmobileww

    Jeff, at OpenMobile, we have a long roadmap for compatibility across platforms (including windows 8), QNX, and others. Currently ACL runs beautifully on a variety of linuxes including of course Meego/Tizen. Interestingly, Windows 8 has included the wp7 runtime, which is very similar to the approach we are taking. We treat android's runtime as just another java runtime- no emulation/virtualzion therefore full performance. We will be making customer announcements soon. In terms of keeping up with Google – Google releases certain things publicly on a regular basis – without going into details, that's all we need. Our CTO Zigurd Mednieks and VP of Engineering Blake Meike wrote the book on Programming Android (for O'Reilly) and we have good understanding of the complexity and scale of the problem we have solved and will continue to solve. Do connect with us on twitter and do stay in touch.

  • Jed

    Excellent article & subsequent comments…
    I hope things become much clearer soon from an end-user perspective.
    Especially for users of the ailing Maemo/MeeGo/Tizen communities.
    We've been through the wars over the years.

VISIONMOBILE BLOG

Distilling market noise into market sense

3D Printing: The 3rd Dimension of Mobile Marketing

3d-printing

Despite the hype, 3D-printing today is far from a household technology, mostly used by non-professionals for fun, entertainment and utility.…

Continue reading ...

The kingmakers of the Internet of Things

image00

For the first time we have data to understand who IoT developers are, where to find them and how to…

Continue reading ...

The 3 key Apple Watch features that nobody talks about. Yet.

apple-watch-09

If Apple wants to create a new, large product category out of smart watches, it must empower developers to discover…

Continue reading ...

VISIONMOBILE RESEARCH

Research on the app economy and developer ecosystems

Developer Segmentation 2014

WF-Developer-Segmentation-Q3-2014

The Developer Segmentation Q3 2014 report is the most sophisticated study of developer segments to date. The report delivers a…

Continue reading ...

App Profits and Costs

AppProfits

This research report examines the critical success factors for a profitable app, and how business and technology choices, such as…

Continue reading ...

Developer Segmentation 2013

Developer Segmentation 2013

The Developer Segmentation 2013 report delivers a needs-based segmentation model that actually works, with extensive profiling of the eight principle…

Continue reading ...