[jsz0]

This fragmentation problem is starting to get pretty ugly. I have an HTC Hero which, IIRC, went on sale via Sprint in October 2009 and I cannot run Google Nav or Google Goggles, or use Bluetooth tethering via PDANet because I'm stuck with 1.5 My 3 year old iPhone 2G runs almost everything in the App Store while my <3 month old Android handset cannot even run Google's newest official applications anymore? This feels like a rerun of the Windows Mobile situation to me.

[jrockway]

because I'm stuck with 1.5

Well, yeah. The problem is not Android, the problem is that HTC is not keeping their Sense devices up to date. They apparently think that their proprietary theme is worth more than being able to use up-to-date apps. As a result, HTC should not call the Hero "Android", but they do. Caveat emptor.

I am sure that if you find an original iPhone and don't update the software that you won't be able to use modern App Store apps. This is the same scenario.

[snowstorm]

HTC doesn't even release a SENSE UI emulator. Neither does Motorola Motoblur.

It's just not reasonable to expect developers to have access to all real devices.

All the device manufactures need to stop customizing Android UI unless they are willing to release a emulator.

[jrockway]

Is this really important? From an apps developer standpoint, Sense and Motoblur don't really affect anything. It goes into the background while your app is displayed. If your app works in the emulator, it will work on a Sense / Motoblur device.

The problem here is that Sense is an ugly hack that deeply depends on Android 1.5 internals, and Android 1.6 and 2.0 are much better than 1.5. The result is that apps target 1.6, and Sense's dependencies prevent HTC from deploying 1.6 to Sense devics.

That's my guess anyway, it could just be that HTC is lazy about updates. I still don't have Android 2.0 on my Sapphire, after all...

(The deeper trend I notice is that hardware companies are not into incrementally improving devices. The release date comes, the product is shipped, and it's forgotten in favor of the next device. A simple firmware update that may take on developer one week to prepare is completely ignored, leaving thousands of users with a device that's not as good as it could be. Depressing.)

[sjs]

Agreed, but even then how many emulators do you want to install? I already have iPhone, webOS (Pre and Pixi), and Android.

[DenisM]

It's amazing that Google is heading down this treacherous path, I would expect them to learn from Microsoft's mistake and Apple's success.

[netcan]

How would you avoid it, sell their own line of phones?

I think the value of Android is that it might produce an iphone quality phone at a (real) $200 price at some point by allowing cheap hardware experts to go to work on making hardware cheap.

[mechanical_fish]

How would you avoid it, sell their own line of phones?

Yep, that was the way to do it, all right. Too late now, I fear, although it looks like Google might be going to give it a try.

I suppose it might even work. Although the successful precedent that I can think of right now -- Apple's move, under the newly-rehired Steve Jobs, to take back control of its hardware market and the Mac brand from the clonemakers by more-or-less nuking them from orbit [1] -- involves a level of aggression that I'm not convinced is in Google's repertoire, especially since phones aren't their core business. We shall see what they choose to do.

---

[1] Of course, that plan also involved deploying Jonathan Ive's designs and the NeXT operating system at around the same time. Google may not find it so easy to make such an enormous splash in the market, even though it's a market that Google helped to create.

[DenisM]

Lack of alternatives to a provably failed business model does not justify the embrace of that model. There is always another alternative - do something else.

At least they could try to build in some mitigation ahead of time. Like fat binaries, run-time version detection, app store that is friendly to mulitple versions of the app/os etc. Not sure if it would have worked, but not doing anything about it is plain insanity.

[netcan]

Would you really call it "provably failed," (especially considering that it's from the perspective of 2006-7) or are you just being provocative?

I agree that this seems like a problem, but surely Android has a decent chance of success under this model. Consider that people who need to run all the latest apps don't use 3 year old phones. In many places (like Australia where I live) you're not paying any extra to upgrade every two years. Google could introduce an Android X with no backwards compatibility without all that much fuss if necessary.

[DenisM]

I am absolutely serious. Hardware and software fragmentation proved near-fatal to old Palm, Windows Mobile and J2me.

The insidious part is that this business model results into hardware manfuacturers being disinterested in providing OS upgrade to old devices. Then a developer ends up supporting multiple versions of hardware/OS but given low software prices they don't have enough money to invest properly. A user then faces apps that are advertised as working on "Windows Mobile" but it turns out require touch-screen, different screen resolution or just crap out on latest OS, or demand the latest OS that user can not get without entering a new two-year contract. I personally stayed away from purchasing a $100 mobile app because I could not figure out device compatibility.

Been there, done that. It's a disaster.

Windows mobile had 100% YoY growth rate so at any point 50% of devices out there were new, and even that was not enough. Maybe if android grows 300% YoY they can make legacy redundant, but how long can this last?

But even then, as a developer I have established relationships with my customers, I can't just tell them to take a hike. If I do they will not advocate my product and I will lose the single most efficient marketing channel.

Bottom line is that upgrading OS to latest version should have been made a key design point form the start, both from the technical point and business-model point. Apple got it right, and as a developer I have 99.6% of my users using the latest major rev of the OS and 93% of my users using the latest minor rev.

[netcan]

The saving grace is that installable apps are still somewhat marginal to phones. IE, you could have a decent product out of the box.