Fragmentation. It’s the word that clings to Google’s Android platform like the smell of grease hours after leaving a diner. There’s no question Android is succeeding in the marketplace, with a plethora of new devices appearing on a near-daily basis And Google has the massive device activation numbers to prove it: in February, that figure was 850,000 devices per day. But there’s also little question that Google faces significant challenges advancing Android. After all, it doesn’t matter what awesome new features Google adds to Android if virtually no one gets to use them — and that’s exactly what seems to be happening with Android 4.0 Ice Cream Sandwich.

How bad is Android fragmentation? How is it impacting the Android community? And, perhaps more importantly, why don’t consumers seem to care?

Five easy pieces

Google offers some fairly detailed information about the version of Android used by Android devices that have connected to Google Play (formerly Android Market) during the previous two weeks. As the data currently stands, it demonstrates how slowly Android 4.0 Ice Cream Sandwich is ramping up amongst Android users who connect to Google Play: as of April 2, all versions of ICS accounted for fewer than three percent of all devices checking in with Google’s marketplace.

According to Google, the most common version of Android out in the wild is Gingerbread, accounting for 63.2 percent of all devices checking in with Google’s marketplace.

One problem with Google’s data is that the company rolls it off the left end of the chart after six months — and once it’s gone, Google acts like it never existed. To counter that, Ron Amadeo over at Android Police has really done his homework, tapping into the Internet Archive to pull up deeper historical data on Android’s version distributions going back to February 2010. Amadeo put it together on a high-resolution chart with pixel-accurate notes for when Android SDKs were released and Android Open Source Project (ASOP) code drops landed — and even mysterious missing slivers where Google’s data doesn’t quite add up.

Amadeo’s chart shows the same data for the beginning of April as Google’s own assessment — hardly surprising, since they’re using the same data. What is intriguing about Amadeo’s effort, however, is the rate at which new Android versions show up in Google’s figures. Two years ago, Android 2.1 burst on the scene and within a month accounted for roughly a quarter of all Android devices checking in with Google. Later on, Android 2.2 took three times as long (three months) to reach a similar percentage, and Android 2.3 (as manifested in Android 2.3.3) took six months to reach a similar point.

Android 3.0 Honeycomb? Never got there. The version of Honeycomb Google originally aimed at tablets never totalled up to even five percent of Android devices. And Android Ice Cream Sandwich? As Google notes, it accounts for less than three percent of Android devices checking in. It also represents an adoption rate that’s phenomenally slower than any previous version of Android, save Gingerbread. At current rates, it may be the end of 2012 before Ice Cream Sandwich accounts for even 25 percent of devices checking in with Google — and, depending how one projects forward, that could well be a generous estimate.

It’s important to note that these charts do not represent the Android ecosystem as a whole: Just Android devices that check in with Google’s marketplace once in a two-week period. Plenty of Android devices don’t check in with Google’s marketplace at all, with high-profile examples being things like the Barnes & Noble Nook Tablets. If those particular devices were considered, the proportion of devices running Android 2.x would be even higher — and, proportionately, the share belonging to ICS would be even smaller.

So no one has Ice Cream Sandwich?

These figures might make it seem like the Ivory-Billed Woodpecker of the mobile industry: some reported sightings, but practically a cryptid. That’s a bit misrepresentative. Although the charts above show Android platform versions as a proportion of the overall population of Android devices, it’s important to note that population has been growing by leaps and bounds.

Precise figures on overall Android unit sales are hard to find — after all, the Android ecosystem is populated with dozens of device makers, each of whom report their sales in different ways (or, in some cases, not at all). But as an example, Kantar Worldpanel ComTech put Android smartphone sales in the United States essentially neck-and-neck with iPhone sales during the fourth quarter of 2011. We know Apple sold about 37 million iPhones during that time. Apple doesn’t break out smartphone sales by market, but that quarter about 58 percent of its revenue was from international sources. Applying that proportion to iPhone sales (an admittedly hazardous estimate), one can could loosely estimate Apple sold about 15 to 16 million iPhones in the U.S. during the fourth quarter of 2012. If reports like Kantar’s are accurate, that means 15 to 16 million Android smartphones were sold in the same period. Unfortunately, Android 4.0 Ice Cream Sandwich was on…well, virtually none of them. After all, the Android 4.0.3 SDK only became available in mid-December. But for grins, let’s apply the same rough estimate to the first quarter of 2012: Apple sold 35.1 million iPhones, 64 percent of revenue was international, so maybe 12 to 13 million were sold in the United States. If Android held basically even (likely a safe assumption), then approximately 300,000 to 400,000 devices with Ice Cream Sandwich were purchased in the U.S. market in the first quarter of 2012.

If you’re still doing the math, that means Apple smartphones with the latest version of iOS outsold Android smartphones with the latest version of Android by a factor of nearly 90 to 1 — at least in the United States during the first quarter of 2012. (Maybe. We are estimating.) But it’s still an idea to keep in mind.

Calm before the storm?

There are several factors that could lead to a strong uptick in Ice Cream Sandwich adoption during the remainder of 2012.

First, device makers are finally announcing plans to bring Ice Cream Sandwich to selected devices. Samsung has announced a selection of handsets on AT&T, Verizon, and Sprint will be getting updates “quickly,” and HTC is supposedly bringing ICS to some of its devices. (Motorola is largely waffling, at least for the U.S. market.) In any case, a significant number of consumers who purchased recent Android smartphones should be converting to Ice Cream Sandwich — albeit perhaps as long as a year after its debut — and that means ICS’s path on the charts above should rise more steeply, rather than creep up at a steady pace.

Second, the population of Android 2.2 devices — currently about 23 percent of the overal Android ecosystem — will soon be start to reach their second birthdays. Although metrics vary, many mobile industry analysts put the average lifespan of a mobile device at about two years — that’s when customers finally pay off the two-year service contract on their existing device and start to shop around for something newer and better. Android 2.2 hit its marketshare peak only about a year ago, but started to account for a significant portion of the Android ecosystem in August 2010. The Android ecosystem was considerably smaller in those days, true, but over time more and more of those Android device owners (most of whom have devices that won’t receive ICS upgrades) will start shopping for new devices. Those new devices — if they aren’t iPhones — will most likely run ICS. Device replacement also ought to make the curve of ICS adoption steeper than it is now during 2012 and (especially) going into 2013. Looking farther forward, Android can expect to see another upgrade boost as Android 2.3 Gingerbread devices start to reach their two-year expiration dates.

Fragmentation, differentiation… or indifference?

Despite the still-rare status of Ice Cream Sandwich, it can’t be denied that Android devices are selling: they must be doing something right, right? One argument is that fragmentation is the wrong word to apply to the Android ecosystem, and perhaps differentiation would be better.

The differentiation argument is basically that where Apple has shipped five (yes, just five) iPhones in the last five years, Android device makers have shipped hundreds of distinct devices, offering a wide range of screen sizes, input methods (like QWERTY keypads), designs (candybars, sliders, even flip-phones), and capabilities — cameras, dual cameras, HDMI output, SD card slots, USB storage, and much more. If you like hardware keypads, you simply can’t get one on an iPhone. However, in the Android ecosystem, you can choose from a range of devices from different manufacturers. Plus, Android devices can be cheap: many Android devices are available at prices well below competing products running iOS (though, when it comes to phones with standard data plans, the ultimate cost difference is minimal).

Considering tablets, differentiation takes on a whole new dimension. Although the Android tablet market has yet to pose a significant challenge to the iPad, Android has made a mark in purpose-specific devices like e-readers and media consumption. Devices like the Kindle Fire may not be full-fledged members of the Android ecosystem because they tuck Android far away behind a custom interface and perhaps block access to things like Google Play (the Kindle Fire did, initially), but they are running Android under the hood. They use Android technology, they benefit from hardware development done for the Android platform and (likely or certainly) fork over money to Microsoft for simply using Android.

So the questions become: Does it matter what version of Android a device is running? If a device meets its owner’s needs and at an acceptable price, what’s the problem? Do consumers care whether they’re getting the latest version of Android or whether they can upgrade — particularly on devices they only expect to have for a year or two?

The answer: Apparently not. Consumers don’t seem to have any qualms buying Android devices, even though only a tiny fraction of them are currently available with the current version and many won’t be upgradable. With purpose-specific devices — like e-readers and media players — that aren’t even trying to be members of the Android ecosystem, the fact Android is inside hardly seems to matter at all.

Developer impact

Where indifference — or fragmentation, or differentiation — will ultimately impact the Android platform is with developers. With a handful of exceptions like Instagram, developers are struggling to justify developing apps for Android. So far the most successful application model for Android seems to be ad-supported services (which, of course, come with their own raft of privacy issues) — in part because those apps can make money without using Google’s still-limited payment system.

But Android developers with apps ready to go are almost certainly targeting Android 2.3, not Android 4.0. That’s even more true if the app is ad-supported, because it needs to reach as many potential users as possible. And developers have to be watching the success of something like the Kindle Fire with a very wary eye, because the current version is almost certainly not going to get an upgrade to a newer version of Android.

If consumers are indifferent to what version of Android they’re using, developers will be, too — and that means the future of Android might be a series of vertically-integrated walled gardens like the Barnes & Noble Nook and Kindle Fire (perhaps joined by efforts from the likes of Sony and Samsung), rather than a broad, generally-consistent platform.