Entelligence is a column by technology strategist and author Michael Gartenberg, a man whose desire for a delicious cup of coffee and a quality New York bagel is dwarfed only by his passion for tech. In these articles, he'll explore where our industry is and where it's going -- on both micro and macro levels -- with the unique wit and insight only he can provide.

A few weeks ago I sat down with the father of Android, Andy Rubin. Andy's a super smart person, having done stints at Apple, General Magic, WebTV and Danger before starting the Android project. We talked about a lot of things, and we particularly spent time discussing Android fragmentation. I've written in the past about my concern that the Android platform is fragmenting much like desktop Linux has over the years, and the potential for the platform to turn into a patchwork of devices and vendor specific modifications that bear little relationship with each other. I've spent a lot of time thinking about my conversation with Andy, and I've rewritten this column more than a few times as a result.

Today, there are at least five different versions of Android on the market. Many of them are highly customized to allow for new features and device differentiation, but that same customization also makes it harder for vendors to update them to the latest versions. New releases and versions of Android are often outdated by newer versions in the span of just a few weeks. For example, the Nexus One when released was capable of running apps like Google Earth that devices such as the Droid could not, because it ran Android 2.0, not 2.1.Tablet vendors complain their Android offerings lack features such as Android Market because Google forbids them to install the marketplace app, forcing them to create proprietary alternatives. It would appear Android is indeed fragmenting -- but perhaps there are other forces at work.

When I spoke with Andy, he pointed out there are several classical symptoms of platform fragmentation. First, older APIs no longer work and break in new releases. Second, multiple application marketplaces offer different applications that lack uniformity across platforms. Both of these are true when you look at desktop Linux. Neither are true of Android.

Andy's point was simple. Older Android devices that can't be upgraded to newer versions of the OS or run newer apps are no different than an iPhone from 2007 not being updated to OS 4. It's not fragmentation -- it's legacy. If so, legacy systems are now aging faster than ever before, due to a rate of innovation that has never been seen before in history. A rate that Google shows no signs of slowing anytime soon.

Does this rate affect Google's partners who need to produce the devices? Absolutely. The difference is Google believes the partner model is now fundamentally different. No one pays for Android, and consequently, the partner relationship isn't the same as when vendors pay to license an OS. One of the issues Microsoft had with Windows Mobile was that as all partners were paying to be part of the ecosystem and license the platform and all were treated equally -- which wasn't necessarily a good thing. Some WinMo vendors produced great devices, while many others produced mediocre devices that made the platform look bad. Some couldn't keep pace with others, forcing Microsoft to sometimes delay key releases so partners could absorb them and create the hardware needed for them.

Android isn't summer camp for handset vendors and not everyone gets a trophy for showing up.


Google has a different view. Android isn't summer camp for handset vendors and not everyone gets get a trophy for showing up. Google is treating partners equally, but will not slow the rate of innovation so weaker players can keep up. By constantly raising the bar, both in terms of reference devices and software, Google aims to keep innovating and drive that innovation as a differentiator. Google wasn't looking for volume sales with the Nexus One, it was looking to raise the hardware bar -- and arguably the best way to do that is to do it yourself.

The net result? A pace of innovation that shows no sign of slowing combined with even more reference hardware that keeps raising the stakes of what's possible. Keep up or don't -- what appears to be a fragmented market is merely the result of shortened cycles of innovation. Older devices seem obsolete faster than before, but this pace also brings speed new innovations to market faster. Google argues that the reason it doesn't permit tablet vendors access to Android Market just yet is to prevent devices that fail comparability tests from actually fragmenting the platform. The message? Keep up -- but don't try to jump ahead.

I'd argue perhaps Android isn't fragmented, at least according to the classical definition, but that the practical result is the same. Devices going obsolete in months and new operating systems released on weekly cycles make it difficult for even Google's best partners to keep pace. Worse, users are confronted with a dizzying array of devices, many of which are out of sync from a software perspective at the time of purchase -- causing some to delay purchases in fear of buyer's remorse or purchase a competing platform.

The open nature of Android is what allows this to happen. Google won't control what partners do with Android, but by constantly raising the bar it controls the platform's pace as well as the vision. As with most things, in the end it will be Google's partners and their customers that will determine whether that pace is acceptable, and that will either become a competitive advantage or provide an opportunity for other platforms to succeed.


Michael Gartenberg is a partner at Altimeter Group. His weblog can be found at gartenblog.net. Contact him at gartenberg AT gmail DOT com. Views expressed here are his own.