Blog Post

How Google Is Managing the Android Fragmentation Issue

In what’s looking like a standard operating procedure, Google just released its free Gesture Search software for Android 1.6 devices and expanded availability outside of the U.S. The application launched two weeks ago on Android 2.x devices. With it, you can simply draw letters on your handset — Android will interpret them and return search results for contacts, applications, bookmarks and music. It works fairly well, although I personally prefer the voice search functionality on my phone. But the big story isn’t one single new feature — it’s how Google (s goog) is managing the Android fragmentation issue — and it just hit me as I noticed the pattern.

Google knows that it has a problem in that there are four different shipping versions of Android right now. There was talk of Google trying to get all the different handsets on one base version this year, but I don’t see such a de-frag happening. It’s not Google’s call because the carriers typically make the decision to push firmware updates, not Google. I wouldn’t be surprised to see Google working aggressively with any handset makers that offer an Android 1.5 device on a 1.6 upgrade, however. That move is just a baby step and the current hardware running 1.5 can surely run 1.6 without any performance degradation.

So what’s the pattern I see? Since Google can’t control the versioning issue, they can at least control core functions and apps among the operating system variances. So when Android 2.x learns a new trick, there’s a good chance Android 1.6 will learn it too. Need a few examples?

  • Google Navigation — this stock-dropping navigation tool debuted on the Motorola Droid with Android 2.0. The date? October 28. On November 23, Google Nav landed on Android 1.6 devices.
  • Google Maps Enhancements – On December 7th, the what’s nearby on Google Maps was added for Android 1.6 or better. In this case, Android 1.6 received new functionality simultaneously with Android 2.x phones.
  • Google Buzz — launched on February 9th, the web client and shortcuts works only on Android 2.x devices, but an update to the official post says “For Android users, and shortcuts are currently available only for phones with Android 2.0+ and we’re working to support other versions soon.” By “other versions” I take that to mean Android 1.6. And current 1.6 devices can use some Buzz features in the latest Google Maps client.

I see two things going on here. One is an effort to follow up as many Android 2.x features on Android 1.6 devices if possible. And where it is possible, Google releases a feature, function or application for both 1.6 and 2.x devices at the same time. If the feature isn’t ready for the older platform, Google releases it for 2.x and follows up relatively quickly with a 1.6 update. It fits the iterate early and often pattern found in Google approach to pretty much everything. The feature-parity approach diminishes the fragmentation issue by attempting to level the functionality playing field across devices.

My second observation? I’m starting to think that Google is passively trying to reduce fragmentation by steering towards two main OS versions as opposed to four: Android 1.6 and a common Android 2.x version. Again, it can’t force an Android 2.0 handset to 2.1, but we’re starting to see some phones getting upgraded — the Motorola (s mot) Droid is on tap for Android 2.1 as early as tomorrow, for example. And we have direct word that the HTC Eris will gain Android 2.x as well.

Without exercising direct control like Apple (s aapl) does, Google has little choice in how to deal with the fragmentation. But they are dealing it with in a unique and quiet manner, likely due to some lessons learned with four platform versions to support. It’s a clever move from where I stand and makes me wonder if this won’t be problem a problem with Android 3.0 — I expect Google will greatly reduced the fragmentation issue by then.

Related research on GigaOM Pro (sub req’d):

Google’s Mobile Strategy: Understanding the Nexus One

28 Responses to “How Google Is Managing the Android Fragmentation Issue”

  1. You know for so long the Apple fan boys have been using “Android fragmentation” as ammo for when ever they want to try and bring down Android.

    As you can clearly see, Android is no longer as fragmented as it once was. However there is still room for improvement. I came across an article that touched point on some of Android fragmentation issues and how they can be overcome. You can read the article here:

  2. standing on the outside looking in at all this mess. The problem seems to be the fact that EVERY android phone released uses the version number as a selling point. Nexus 2.1, Droid 2.0 and you can bet that 2.2 will not come out unless there is a feature phone behind it. This is should not be the case.

    What this causes is that it ties the hardware updates to the software updates. Android 1.6 users may NEVER get any updates. Who knows if the chances between the versions are silly “surface changes” or “kernel changes”. I think the nexus one had animated wallpapers. Was that a kernel change? I don’t see why wallpapers would require you to jump a version number.

  3. The problem with fragmentation for me as a user is that until android, my phone was a phone – if I wanted a new feature or capability I had to get a new one. When the iphone app store came online it really changed my expectations for my next phone. I’m not getting an iphone because I won’t change carriers, I also don’t want to trade phones every few months as new ones come out. When I upgrade to android I expect to live with the phone for two years at least. Soooo… fragmentation fills me with FUD. If I get today’s mainstream phone (probably a mytouch 3g) what do I not get in capability? For that matter, if I get a snap-dragon phone with android 2.1, will it really get the next new, shiny thing? and when. And, no its not like Windows. As Windows has evolved it has generally changed along with the capability and aging of the machines. When Windows 98 came along, it was time for a new computer if my current one couldn’t handle it. when Vista came out, I stuck with XP until I upgraded to a 64-bit golly fast machine and moved to Win7. And in the mean time, my older computers always did the tasks. And I got to choose when to upgrade. with phones, now, the hardware and software is moving fast, the upgrade cycle is out of my hands, the road map for future roll outs is not tranparent, and an upgrade OS isn’t just faster, its applications you can’t get….

    I’m really glad this debate is going on and that google seems to be listening.

  4. It would probably be easier to think of the Android versions as marketing driven shorthand, taking the codenames (Cupcake, Donut, Froyo, etc.) and exporting them from the company. These version numbers are connected with features, launches (like Droid or Nexus) and a myriad of other issues. Internally, Android uses a numbering system just for the API features and APK can use and the APK itself declares this compatibility. Beyond this, there are kernel compatibility issues in the kernel to userspace API of the Android (and Android device) extensions to the Linux kernel which evolve at a much faster pace than the upstream kernel equivalents. (For an example, compare the ALSA sound API with the Qualcomm ADSP system).

  5. Being that updates are carrier controlled, I’m interested to see how the carriers handle updates for the Nexus One now that it’s available on two carriers, not supported on one directly by the way, and soon to cover the remaining two carriers. Thoughts?

  6. Not that easy… If Android is to take off (and it will) apps will come from herds of developers. Who can guarantee they’ll update backwards and forward? How much time can and will google keep up with that? Probably until it takes off. Welcome j2me hell…

  7. doctor-don

    Gesture works nicely on myTouch with 1.6. As soon as I had installed it, Maps was upgraded to 4.1 and WeatherBug was also updated. WB seems to be faster.

  8. Paul Calento

    From an enterprise and volume implementation perspective, several versions of Android add complexity to deployment, management and security. Yes, organizations that choose Android will (hopefully) attempt to standardize, but according to a recent Forrester study cited in a Sybase marketing piece ( ), almost half of respondent companies indicated they were already supporting personal devices of some type. Different versions of Android doesn’t make IT’s life any easier. Not ideal. (About Me – )

  9. This is so much better than Nokia which releases anything new only for their latest handsets and usually only for some of them (Ovi Maps, Maps Racing and tons of other software)

    • I really don’t see the problem with Fragmentation as long as it’s handled properly. Google allows handset makers to use Android and they use the version of Android of their choice. They customize it how they see fit and they can decide when and if they want to upgrade it. If they do a lot of heavy customization then that takes time and upgrading it properly will take time…and they might not decide it is worth their time to do so. If I buy a device and it doesn’t have a feature I want then why did I buy it?

      The hardware, software and applications that run on it are known at the time of purchase. No one complains that they should get free upgraded hardware once better hardware specs comes out. Why should they feel entitled to free upgraded software?

  10. Fragmentation is about giving the customer, choice, as in Apple copying that move with the iPhone OS- not all apps that run on the iPhone run on the iPad.

  11. Re: Google’s inability to push updates out because of carriers.

    Palm is now on 2 domestic carriers, 4 total in North America, and on a handful of Telefonica subsidiaries in Europe. They were able to deploy their latest update across all of their carriers within a week of each other.

    If Palm can do it, why not Google?

    • Rob, it’s not a technical reason, per se. Ultimately the carriers control what firmware goes on phones that they sell. Palm doesn’t have much of an issue because it doesn’t have a wide range of handsets to support and carriers can test the firmware faster with fewer handset models. Google has to make the firmware updates available to the carrier and the carrier generally decides to push it, and if they’re going to, they typically test it. Simply put: it takes far more effort in Google’s case.

      Looking at it from a different perspective — the Verizon Palm devices could get an update that Sprint chooses not to implement. Again — not a technical issue, but one of carrier control and business practice.

      • And why is it left up to the carriers? Handset makers and the OS developers for those handsets should never agree to that kind of deal.

        You have to admit that’s one thing Apple has going for it. When they put out an update, EVERY iPhone user gets that update regardless of hardware.

        With an Android phone, who knows?

        • It’s left up to carriers in the U.S. because that’s been the business model for so long. Under the current scenarios, if handset makers or OS developers don’t agree to a carrier’s requirements, the device doesn’t go on that network so they work through carrier. Google is a little more autonomous in this regard, so they’ve got the best shot at breaking the cycle currently in place. I agree with your Apple example, but bear in mind for the purpose of this discussion, they’re only on a single U.S. carrier. If they were on multiple carriers here, they too would have to wrest control — they did it with AT&T, but there’s no guarantee they could do it with the other 3 major carriers. Probably part of the reason that they’re only with AT&T here.

          • Google is not powerful enough to break the carrier-control model. Not even Apple is that powerful. Only the federal government can do that.

            Unsubsidized phones have a distinct disadvantage because of their price. Also, because you can’t take any phone and take it to any carrier of your choice the customer has to think about carrier when thinking of device. The Nexus One users who used the T-Mobile version on AT&T are probably upset that they couldn’t get full use out of it.

  12. Interesting observation. Though in some ways expected. But how long can they keep this up ? If you can keep putting on 2.x features into 1.6, what was the difference btw 2.x and 1.6 anyway ? And then where does 3.0 fit in ?

    Also I isn’t the real issue of fragmentation the fact that vendors such as HTC are adding their own variances such as the SenseUI. While others leave out things.

  13. Android actually causes less fragmentation

    If android (or an alternate os for multiplatforms ex. winmo) was not around we have multiple manufacturers (HTC, Motorolla, SonyErricson, Dell, ASUS, etc…) either not able to have a cell phone or would have a cellphone with there own OS which is more fragmented than having one os on different versions.

  14. I liken the situation to Microsoft’s consumer desktop operating system where the company is simultaneously supporting Windows XP, Vista, and 7.

    Fragmentation is a consequence of a desirable practice: continuing and improved support of legacy devices. It’s a good sign when a company gives user’s of older phones new features and improvements, but it’s to be expected that not all features developed for the latest hardware will be portable to older equipment.

    The effects of fragmentation can be mitigated but that it exists at all is a good thing…