Fred believes there are benefits of fragmentation that are a "big plus". I completely disagree. I believe the negatives far outweigh any positives and I think the market will prove this out over the next couple of years.

I think we'll see a few key Android devices rise while the others linger. And the OEMs will start cutting their lines down to focus. Just like another company that makes another mobile OS.

A lot of this is about consumers, but more important are developers. While being on as many devices as possible sounds great in theory, it's a huge pain in the ass to try to develop and maintain a great product across them all. In fact, for many smaller development teams, it's impossible.

BTW, here's my previously blogged response to Fred, which he commented on via Tumblr. Hence why I'm creating this Branch. Comment fragmentation!

parislemon.com

I'm on the front lines - building both iPhone & Android apps. I won't comment on the revenue disparity between the two - that is for a different branch. But I will say we release lower quality, less tested apps on Android because we can never test all devices - we test live with our customers. It sucks. Luckily the Android customers know to submit their android device & OS whenever they send us bug reports.

Fragmentation in my experience has produced less, and lower quality apps.

The real question is whether or not fragmentation will be an issue in a few years. When Samsung copied Apple they were able to push themselves past other OEMs in terms of software and hardware design. While HTC, Motorola, and others were trying to innovate on their own (after some minor copying on HTC and Verizon's part) and create their own software and hardware design language, Samsung was copying Apple to get ahead of everyone else. Because of this, I'm having a hard time seeing fragmentation as much of an issue in the next few years. Does anyone really expect HTC or Motorola to make a comeback?

I sit mostly in the middle on this one, save for my utter hatred of OEM skins/themes/custom UIs. But that too is for another branch. You've nailed the main issue with fragmentation, the affect fragmentation has on application development. Often, developers don't have enough money to purchase multiple devices to quality-test their applications, and what you're left with is something that doesn't work well on some devices.

Where I disagree is I think we'll still see multiple OEMs releasing Android handsets so long as Android amasses 50+% of the smartphone market. OEMs definitely want a piece of that pie, and switching to Windows or BB10 isn't going to do it, not right now, anyway.

From the perspective of a designer, creating resources for both Android and iOS apps, I must say I have found Android is less demanding. (I am sure developers could disagree.)

Android divides graphics into categories intelligently, which leaves iOS needing more: splash screens - six of them; numerous extra app icons for Search and Settings; and more. This makes working to accommodate these devices much more tasking - whether the outcome is worth it or not.

In this sense, it it easier to make an Android app, supporting around four thousand different types of devices, including phones and tablets, than eight running iOS. (Of course, the numbers of devices doesn't match that ratio.)

To conclude: perspective matters immensely in this issue.

This is a good point as well. I asked these questions to a few developers a while back, and they had some interesting perspectives on the matter. I've posted the emails on this blog (which yes, I created just for the purposes of this branch):

anthonydomanico.wordpress.com

Food for thought...

"Fragmentation," as it's usually discussed, usually devolves into "Forking." Fred's talking about forking. And forking, done well, can be very interesting. The problem is we've only seen a handful of companies (outside of Asia, I guess, about which I know nothing) do it worth a shit. Amazon. Barnes & Noble. Anyone else? Don't say Ouya.

I define fragmentation is multiple versions of a base OS spread across multiple current devices. (Say, the generation being sold, and generation before that. But, obviously, that gets muddy in Android, as one generation blends into another.)

Amazon and B&N forked. They said to hell with Google's CTS and did their own thing. And that's exactly what Android was designed to do, for better or worse.

On the cause of fragmentation:

I should say that I have a feeling that a large number of people believe Android fragmentation is a bigger problem than it actually is because of carriers (or 'networks' here in the UK). They fail to support customers, leaving them in the cold. Many then take this a step further and assume that it is the number of devices that causes the problem. I think it may just be the carriers' attitudes towards updates that cause the issue - frankly.

On forks:

Precisely as Phil said, Android was designed to be forked. The only problem is that Google is involved enough to be held responsible, but not enough to force manufacturers (and carriers) to update their devices. That's the worst place to be.

@Graham Exactly. But is that on the carrier? Or is that on Android (and Google by very close extension.)

Should Google be more responsible for how carriers and manufacturers use Android? In one sense, it is, with the CTS. But what about all the FOSS? You can't force that.

I can see why it would be Google's responsibility, but they just don't have enough control.

To put it plainly, I think whatever Android, or Google, do now will be easy to criticise. If they do nothing, fragmentation will remain as-is and will not improve, continuing to detract from the value of the platform for developers and users alike. If they reduce involvement they will be seen as abandoning the platform, and if they become more involved people will blame them for the disruption carriers will no doubt cause upon hearing the news.

Can you see of a way to improve the situation without a backlash?

Goes back to the original question. Who, exactly, is most hurt by there being multiple versions of Android running on multiple devices. I'd argue the devs more than the consumers, the vast majority of which don't read Android Central (shame on them) and have no idea what version they're on. That said, I think they're getting smarter.

Devs have to worry about DPIs and APIs and lord knows what other PIs. OTOH, when I buy their apps, or put up with their ads, that's what I'm paying for.

Could (and should) Google work to make that as easy as possible for developers? Yes. Is Google doing that? It's certainly not doing nothing.

In my mind, there are two groups affected: developers and consumers. Developers because of the reasons you provided. And consumers because they can't get the quality or quantity of apps they are looking for - because it is a difficult ecosystem to develop apps for.

You say that's what 'you' are paying for, but I fear you may be in the minority. The lack of paying customers discourages developers. This leaves them, not only unable to produce good apps for all Android devices, but unable to buy testing devices. (I'm sure you know this well.)

It's a cycle -- a cycle of underpaying customers and low quality apps that I can see continuing for some time. There's no doubt Google is trying to break this cycle, but it will not be easy.

Is fragmentation an issue in the present and now? Definitely. But where I see it becoming less of an issue and more of annoyance is with the improvement of the mobile web. The next generation mobile browser will be "the" app. As mobile websites become more complex, I think fragmentation will bear its ugly head less and less. Why develop across platforms if you just create a killer mobile website?

(developer hat) Part of the challenge of building great android application over the last 2 years is that there has been some significant changes in the rendering model/hardware optimization between 1.x, 2.x, 4.0 and 4.1. That combined with different hardware memory and GPU configs made it really really hard to deliver a really smooth app - specially if you are targeting multiple versions. But there is light at the end of the tunnel: JellyBean rendering is a lot better. As more phone with good GPUs embrasse it, Android will start to look a lot more uniform from the developer perspective. The next battleground for mobile development is going to be more about cloud integration and inter-app interoperability and less about graphics performance

Fancy seeing you here, Edwin!

Feedly may in fact be a very good case study. Are there any specific limitations of lower versions of Android that you have had to cater to whilst developing the app besides rendering issues, despite moving forward to a heavily HTML5-based solution, that you feel may have reduced the quality of the experience for users of higher versions, or do you feel compatibility with lower versions is aided by Google's method of updating to the extent that it is not a major problem?

And at what point do you believe a developer should leave an older version behind in order to improve the experience on higher versions?

"And at what point do you believe a developer should leave an older version behind in order to improve the experience on higher versions?" The definition of a great platform is one that never forces a developer to do so. Feedly still runs well on an iPhone 3GS (and runs transparently better on iPhone 4 and iPhone 4S). The reason behind this is that Apple had nailed their rendering model from the get go. Android had a rocky start but they are now on part with iOS (see developers.google.com). This is why I believe that rendering and fragmentation are going to slowly become a non-issue.

Fragmentation isn't going anywhere in part because the world is inherently fragmented: Apple isn't making a $50 dual SIM phone anytime soon. Nor a refrigerator with a touch screen. Nor licensing iOS to the Ouyas of the world.

As a developer, would you rather those things ran some version of Android or would you rather they each ran something completely different? As a consumer buying one of those, would you rather they ran a ton of apps in some form, or would you rather they each have to beg Facebook, game makers, etc to develop a brand new app just for them?

It may not provide an ideal experience but might it be better that all the random gadgets share 1 OS in a fragmented way than for each to have a different OS? There is no 3rd choice.

Diversity has always been an issue. Diversity of opinion or culture often leads to friction. However, let us strive for harmony, as I'm sure it will come one day.

Fred knows all about investment diversification, so from that perspective, he may be right. However, as a designer, my primary concern is understanding the device I’m creating a profitable — or otherwise — experience for. With almost 300 Android devices available, designing becomes an increasingly inefficient pursuit. Understanding form factor, hardware and the expectations of users go a long way to informing the quality of a product. Without this understanding, fragmentation forces upon us a very average or lowest common denominator experience.

I've expanded on this on my blog.

MG - i prefer open platforms and open conversations. i dislike the idea that only people invited into this conversation can post.

Fred – There is a fallacy here. Android has accidentally created an awkward and problematic form of closed openness.

Pick an Android device at random and try do something you've previously done on the Galaxy S III. Try make a phone call, or send an SMS. You'll soon become frustrated, it will feel broken, even scrolling will feel broken. People are using apps that may never have even been tested on the device they just spent their hard earned money on. If they want the real experience they better go out and buy an S III. The idea of openness has compromised itself, huge chunks of Android users are sold a lie.

Whereas with Apple, they don't lie about this. They have an offering, take it or leave it. With Android, it's all smoke and mirrors.

I feel you may be thinking of Android too much as a user-facing product. I'd argue that 'Android' isn't what Samsung, or any other OEM, is selling: they're selling their own interface that is based on Android. They don't understand - or choose not to - that 'Android' is a brand in itself for which consumers have certain expectations. I can see why you say consumers are sold a lie, but it is a lie they and OEMs have created, not Google or Android.

Android's marketing is a huge problem, beside those already mentioned, as it blurs the lines of its positioning. Is it an open-sourced 'platform' for any developer to use, or is it a separate 'product' for consumers?

Graham, I blame Google. They're the policy makers, they allowed this to happen.

I can accept that. I blame everyone - previously mentioned - partially.