We Need to Talk About Android

I spoke at a conference near Cardiff recently and in Q&A, I got The Question. I love getting the question.

What's the question? This:

What's wrong with Android?

I realised, giving my answer, that I've never written down my objections to Android. Before we get into this, let's understand that I'm primarily talking about "what's wrong with Android from the perspective of someone planning a long-term 1:1 deployment in a school". You can argue that these points don't matter in the grand scheme of things but these are the things that I choose to care about in my deployments. I ask these questions of every platform.

As I see it, there are several things currently wrong with Android from a deployment perspective.

Fragmentation

I'm not specifically talking about device fragmentation here. I'm talking about fragmentation of the basic operating system as deployed in the field.

Recently there have been several useful visualisations and articles posted about how quickly new versions of Android and iOS are taken up by the respective installed bases of each platform.

Today, iOS 5 is deployed on the majority of iOS devices in the field. By comparison, variants of Android 2.x remain vastly dominant in the installed base of Android devices. Pxldot recently posted a fascinating comparison - with numbers - of the take-up rates of iOS and Android.

The basic problem is this: Google continues to evolve the core Android OS but they can't get that out to the majority of consumers in a timely fashion. I'm not talking here about user-facing features such as the latest design of the Calendar app or the visual tweaks to the home screen.

I'm talking about APIs. This really matters. I'll talk later about some specific places where it matters a lot, but it matters generally because the the APIs define the power available to third-party developers.

The Android platform is currently stuck in second gear because Google, their OEMs and the carriers can't, won't or simply have no incentive to get the installed base past the Android 2.x API set. There are better and more powerful APIs in Android 4, and there will be better ones again in the future, but developers can't take advantage of them because almost nobody is running the latest OS.

For example, Google recently shipped Chrome for Android which, by all accounts, is a pretty great mobile web browser. Unfortunately, it requires Android 4 and around 1% of the installed base is currently running that release.

This means that iOS apps are not only better than Android apps today, they're getting better faster than Android apps because Apple is deploying and the installed base is rapidly upgrading to much more powerful APIs on the devices in consumers' hands.

Backup and Restore

To all intents and purposes, Android has no backup system. That's not completely true, as there are APIs for backing up to the cloud. The API documentation is riddled with caveats:

The backup transport is the client-side component of Android's backup framework, which is customizable by the device manufacturer and service provider.

So it's no longer a question of "does Android support backup?". It's now a question of "does the customised version of Android installed on the Motorola Xoom support backup when I buy it through Carrier X?". You have to verify this on a per-device basis and you know how one of the strengths of Android is the wide array of devices you can buy? Enjoy!

Data backup is not guaranteed to be available on all Android-powered devices.

and:

Because the cloud storage and transport service can differ from device to device, Android makes no guarantees about the security of your data while using backup

To my mind this, alone, is a dealbreaker for Android in education. In a world in which the mobile device you deploy is going to be a serious computer that pupils will use to generate work for exam-level assessment, you had better have a way to back up and restore their data.

Lifecycle Support

The world of mobile devices is moving fast. We're on a one year upgrade cycle for devices and their operating systems. In any school situation, a technology roll-out program is a muti-year operation.

With iOS, you get updates direct from the source. Apple ships updates to any device for the period in which they're being supported. At the moment, this is usually three years or more for iOS devices: the iPhone 3GS from 2009, the 3rd generation iPod touch from 2009 and all iPads can currently run iOS 5.

The story so far on Android is quite different. Michael DeGusta's well-known visualisation of Android phone updates paints a sorry picture. That post doesn't address Android tablets but I judge platforms on their track record, not on vague promises that next time will be different.

This matters for several reasons. Firstly, if I'm going to be signing a two- or three-year lease for hundreds of devices, I need to have some idea of how well these devices will be supported over the lifetime of the lease. Imagine if we were still stuck with iOS 3.2 on our iPads today.

The second reason this matters is security. Lets talk about that.

Security

There are problems with security on Android. Roughly speaking, they fall into the categories of security exploits and malware. Every platform has security exploits - heck, the very basis of iOS jailbreaking is finding security holes to exploit - but the incidence of malware is not evenly spread.

One of the claimed strengths of the Android platform is the ability to download software from anywhere and install it on your device. No gatekeeper! No walled garden! That's a perfectly valid thing to aspire to.

I take the claimed importance of this at face value: if you want it, I assume you're planning to actually use it. If you're going to download and install apps from all over the web, you had better be sure that the base OS is bang up to date with security patches. That's simply not what you get with Android.

This is one of those places where Google's inability to move the installed base to new OS releases is actively harmful.

Applications

In 2011, I spoke at Mobile World Congress in Barcelona. That was the coming-out party for all the Android tablets that year. What I realised at that event was this: anyone can make a tablet of moderate quality. They can put a browser on it, a decent mail client and a calendar. Most tablets can get a port of those apps whose business model depends on them being everywhere: Kindle, Evernote, Netflix, Facebook and so on.

My question was then, and remains this: where are the apps to challenge iMovie, GarageBand, Keynote, OmniFocus, OmniGraffle, Soulver, Flipboard, iThoughts, Noteshelf, Collabracam, The Elements, Brushes and ArtRage?

I'm not saying those apps cannot be built on Android but I am saying they aren't being built on Android right now.

Our experience over the past 18 months of teaching with iPad is that it's about way more than just surfing the web and sending email. The web is vital, of course, but we've seen the most benefit when using these powerful content creation apps.

I used the Mac in the days when nobody used the Mac. I don't believe and I'm not arguing that absolute quantity of apps is the sole determinant of a platform's value. The Mac often only had one or two apps in a certain category but it almost always had a good app in that category.

I just don't see the same quality, range, depth or ambition in the Android marketplaces that I see in the App Store.

VAR Interference

In education, many schools work through resellers rather than retail channels. Something I'm seeing more and more of is the "educational tablet". This is usually some kind of Chinese OEM hardware running an obsolete version of Android with a custom UI on top. I have a real problem with these products.

My entire experience with software written by educational suppliers is that they rarely keep up with the pace of the underlying vendor. Then, when something breaks, there's the inevitable finger-pointing between the reseller and the manufacturer.

When a new version of Android comes out, you're going to have to wait for your reseller to port their changes the new version of Android. This is the same problem that Android phone users have with Motorola, HTC and other OEMs - except that educational resellers are often far less well-funded and capable than major handset manufacturers.

In Conclusion

So that's where my problems with Android lie. When I decided to start this program, there were no Android tablets to consider apart from the Dell Streak 5. The Streak 5 shipped with Android 1.6, got an update to 2.2 at the end of 2010 and was discontinued in August 2011. By contrast the iPad 1 shipped with iOS 3.2, got an update to iOS 4 in November 2010 and then iOS 5 in October 2011. It's still supported for security and functionality updates in the iOS 5.x line. Even if iOS 6 doesn't support the iPad 1, that won't be out until late 2012. That's a good 2 years of complete support, security updates and substantial feature enhancements.

Right now, the Android platform looks stalled in the marketplace at Android 2.x. Android 2, coupled with vague promises of future upgrades seems to 'good enough' for most carriers and OEM manufacturers. Even Sony just launched their Experia S phone with Android 2.3 and a promise to upgrade to Android 4 in Q2 2012 (we've heard that one before).

You're either buying into a platform or you're buying gadgets. The fundamental disconnect between the apprently solid Android engineering that's happening at Google and the actual packaging and deployment that's happening to end-users is turning into a real problem. To my mind, it's a dealbreaker for schools or anyone thinking beyond their next carrier subsidy.