Most of the time when I hear designers complaining about Android fragmentation they are really just using "fragmentation" as a scapegoat to bemoan they fact that they can't just create a very small set (1-3) of universal "pixel perfect" mockups and ship those off as a PSD to a developer as the end-all, be-all "visual spec". They have to worry about things like how different parts of the app should scale as more real estate is made available, really understand DPI (I've met more than a few designers who shockingly don't really grasp DPI scaling), etc. Some designers are awesome at this and welcome the challenges (and benefits!) that "responsive layout" brings with it, but a lot of them (most of them, in my personal experience) just want to punt on all that shit.
People say this. It's not really actually true.
A nice little example encountered recently. Say you have a HTTP server which uses gzip or deflate when the client asks it to. Pretty standard, right? And say you want to return a HTTP 204, or other contentless success. Seems reasonable? And if you do this on Android 4.4 using the standard HTTP client, it will work. If you do it on Android 4.1 (the most commonly-deployed version of Android), it will throw a null pointer exception deep in the library code.
This is just one of many issues. Compatibility support libraries are all very well, but they don't really help with the standard libraries being buggy, and those bugs will _never be fixed_ for the vast majority of users.
It gets even worse when you take OEM crap into account. Want to use the camera? You'd better know about obscure Samsung quirks: http://stackoverflow.com/questions/13448731/does-samsung-gal...
"Mika says it spent 20% of its total development time last year on the process of porting two of its games to Android and then tweaking those titles for individual devices. It could never even get Battleheart to work on Galaxy S phones, because they can't download .apk's larger than 30MB with stock software. For all the months of "thanklessly modifying shaders and texture formats to work on different GPUs, or pushing out patches to support new devices without crashing, or walking someone through how to fix an installation that wouldn't go through," Mika claims it never made a profit off either of its Android games."
Geez, I wouldn't go into Android game development after reading this. And we haven't even breached the piracy issues.
[1] http://www.androidpolice.com/2012/03/12/mika-mobiles-decisio...
"It's too hard" or "it doesn't make enough money" are more than reasonable. A software limitation in a 5 year old device is not.
The Galaxy S was released in June 2010 (so it's not even 4 years old, let alone 5…), it was barely a year old at game release and was not only one of the popular smartphones of the times, as the Nexus S it was also the reference device until November 2011.
At the same time, developers like ShiftyJelly (www.shiftyjelly.com) show that fantastic crossplatform apps can be developed which have a great Android UI and are still very similar to their iOS counterparts.
I genuinely think it is just laziness, but while users continue to put up with it (especially since Android users appear to be happy to stick with a shitty UI on a free app compared to dropping a couple of dollars on a great one), it is unlikely to change.