Click on the images in this post to view a larger, more detailed version.
A few days ago, when Google released the Android N Developer Preview, I was reading some articles about the famous Android fragmentation problem. By now, we all know why it matters and why it doesn’t.
In my opinion, it’s quite overrated. Developers are used to programming against the support libraries instead of the true Android API. It would be nice if Google could fix this annoyance, but it’s clearly not limiting developers.
So why does it still matter? It matters to the end users. If you buy an expensive flagship phone, you expect it to receive new features through updates just like your PC does. It doesn’t make sense to provide updates for devices with outdated hardware, but most recent phones can last a few years before their hardware becomes outdated. However, their software can’t. Most phones are updated for a year or two and are then ignored as its manufacturers want you to buy its newest product.
The Android version with slowest adoption ever is still Android Honeycomb. That’s quite normal. Android Honeycomb only ever rolled out to the tablet market which was very young at the time. However, the version with the second to worst adoption rate is Android Marshmallow, Google’s latest release. The versions before it didn’t do much better. The situation seems to be getting worse and worse.
This brings the Android version landscape today to this:
One thing that I personally find interesting is the highest adoption rate an Android version ever reached. Both Android Froyo and Android Gingerbread each once were on more than 60% of the active Android devices, but recent versions never reached such a high penetration into the ecosystem. Okay, there were less Android devices at the time so it was easier to get them updated. But regardless of the highest adoption a version ever reached, it’s still important to look at time it took the version to reach that adoption rate. Likewise, it has its ups and downs, but it seems to be getting worse overall.