Fragmentation sucks

[czerdrill]

Comparing it to PCs doesn't make sense because you always have the option to upgrade your OS

The flaw in that logic is the user has to BUY the new version of windows, and most don't (until it's pre-installed on a new PC). There are at least 3-4 common versions of windows still running out there, all the way back to Win2000.

it's not a flaw in the logic at all. if the user had the choice to buy an android upgrade there wouldn't be talk of fragmentation and devs would be able to create apps specifically for the OS version they wanted to and not have to support older versions (you bought an OS upgrade at a premium so you get to use my app). the fact that these os revisions are offered for free by google, but then never see the light of day on the 60+ android devices proves that fragmentation is a real problem. and thats because google has no standard. they just release a crapload of OS versions, begin to ignore the old ones, and fail to realize that carriers and manufacturers are not passing the OS's along to the end user. So the dev who's developing for Gingerbread ends up supporting phones running Android 2.2 because it'll be available in their Market and then force close twenty times.

if MS OS updates were free and were dependent on the manufacturer we'd have the same problem. If HP releases updates and all their computers are running Windows 7, whereas Dell is still on Windows Millenium edition, you think people wouldn't complain about fragmentation? Especially if there was a "Windows Store" where everyone can download software? The reason you don't hear about PC fragmentation (except when you're defragging your hard drive :icon_ banana is because the user has the option to upgrade. Whether you pay for it or not is irrelevant.

 

[kodiak799]

it's not a flaw in the logic at all. if the user had the choice to buy an android upgrade there wouldn't be talk of fragmentation

Fragmentation is a description, a condition. Whether the user has an option to upgrade or not is irrelevant, the fact that they DON'T upgrade is where fragmentation comes into play - an update not pushed vs. one not bought is a distinction without a difference. The developer still has to support that old version and the new version if they want new sales. Android developers can choose not to support a version or handset as well - when the user upgrades they get the latest and get the app. You're contradicting yourself - very rarely do you see new PC software out that only supports the current and latest OS. You make it backwards compatible because the ROI justifies the R&D.

Fragmentation is nothing more than a boogeyman term. This is not a Linux "if you build it they will come" issue. The installed base is real and large and the ROI for developers is real. Like I said, fragmentation certainly hasn't hurt the growth of the market. Fragmentation is every bit as much of an issue in PC yet we don't here about it because the ROI makes it irrelevant.

 

[czerdrill]

it's not a flaw in the logic at all. if the user had the choice to buy an android upgrade there wouldn't be talk of fragmentation

Fragmentation is a description, a condition. Whether the user has an option to upgrade or not is irrelevant, the fact that they DON'T upgrade is where fragmentation comes into play. The developer still has to support that old version and the new version if they want new sales. Android developers can choose not to support a version or handset as well - when the user upgrades they get the latest and get the app. You're contradicting yourself.

Fragmentation is nothing more than a boogeyman term. This is not a Linux "if you build it they will come" issue. The installed base is real and large and the ROI for developers is real. Like I said, fragmentation certainly hasn't hurt the growth of the market. Fragmentation is every bit as much of an issue in PC yet we don't here about it because the ROI makes it irrelevant.

Fragmentation hasn't hurt the growth of the market because everyone and their mother can access and install every app that a dev puts out even if it's incompatible with their OS version. Growth is not a direct correlation to whether or not fragmentation exists, it is a direct correlation to how fragmentation will also grow...you think the devs are not annoyed by having to release maintenance updates for their apps every other day? (exaggeration, yes, but I've received more updates to my Handcent then I can count and they're almost always "adding support for Galaxy S", or "fix force closes on phones running Gingerbread roms") If you're not a dev, you really can't say that with certainty could you? I created one app for the Android market, that got like 200 downloads, an utter failure, haha, but I could imagine the nightmare I'd face if it became popular and I had to release updates for each new android device or os version. The ROI for android devs is not anywhere near it is for iOS devs and that's because of fragmentation. Even from a time perspective, the time spent on maintaining an android app is ridiculous compared to an iOS app.

I've never once heard anyone complaining about PC fragmentation (until now) simply because it's a non-issue. I just tried to Google and find one article about PC fragmentation and couldn't find even one. If you can find one, please provide it, as I've never even heard of such an issue before. In Windows, software is often compatible across OS's by the work of the developers themselves ensuring compatibility...Or you simply cannot install that software unless you meet software requirements. With android, someone running android 1.5 can install an app that was developed for GB, run it somewhat and then watch it force close and then bother the dev about it. there is no standard with the way apps are created, run or installable in android. Couple that with manufacturers including their own "tweaks" to the android base and you get a mess. That's why fragmentation is real and not some boogeyman concept...

 

[kodiak799]

The ROI for android devs is not anywhere near it is for iOS devs and that's because of fragmentation. Even from a time perspective, the time spent on maintaining an android app is ridiculous compared to an iOS app.

I've never once heard anyone complaining about PC fragmentation (until now) simply because it's a non-issue.

Fragmentation on PC's is a non-issue because there is big money on it.

Again, you can't discuss fragmentation without talking about the ROI. ROI can be thought about in terms of two opposing forces - fragmentation and installed base (or market size). So long as installed base increases faster, or at least proportionately, with fragmentation the ROI is at least stable.

And that's precisely why you don't here about fragmentation in PC even though it has virtually the identical issue - multiple screen sizes, literally hundreds of hardware configurations, 4 or 5 or more versions of Windows (and then there's Apple, too, and Linux)...not to mention the million updates Windows sends out that sometimes requires a developer to update their app. That is the "definition" of fragmentation being used here, but you don't hear it because the money involved makes it a non-issue.

Is the ROI higher for IOS? Sure, but that's irrelevant, all that matters is if the ROI for Android is high enough to make the investment worthwhile, and clearly it continues to be.

Just to make an extreme example to illustrate the point - what's worse, 3 OS's on 4 D1's, or 5 OS's across 4000 DX's? Fragmentation, by definition, is greater in the latter, but fragmentation will be a much bigger issue with the former because you cannot ignore the installed base (it's an opposing force)

 

[czerdrill]

The ROI for android devs is not anywhere near it is for iOS devs and that's because of fragmentation. Even from a time perspective, the time spent on maintaining an android app is ridiculous compared to an iOS app.

I've never once heard anyone complaining about PC fragmentation (until now) simply because it's a non-issue.

Fragmentation on PC's is a non-issue because there is big money on it.

Again, you can't discuss fragmentation without talking about the ROI. ROI can be thought about in terms of two opposing forces - fragmentation and installed base (or market size). So long as installed base increases faster, or at least proportionately, with fragmentation the ROI is at least stable.

And that's precisely why you don't here about fragmentation in PC even though it has virtually the identical issue - multiple screen sizes, literally hundreds of hardware configurations, 4 or 5 or more versions of Windows (and then there's Apple, too, and Linux)...not to mention the million updates Windows sends out that sometimes requires a developer to update their app. That is the "definition" of fragmentation being used here, but you don't hear it because the money involved makes it a non-issue.

Is the ROI higher for IOS? Sure, but that's irrelevant, all that matters is if the ROI for Android is high enough to make the investment worthwhile, and clearly it continues to be.

Just to make an extreme example to illustrate the point - what's worse, 3 OS's on 4 D1's, or 5 OS's across 4000 DX's? Fragmentation, by definition, is greater in the latter, but fragmentation will be a much bigger issue with the former because you cannot ignore the installed base (it's an opposing force)

But we're not talking 5 OS's on one "D"...we're talking 5 OS's on 60+"D's". And at the end of 2011 it'll probably be 7 OS's on 80+"D's"...as it becomes more fragmented the ROI becomes lower not higher. No android dev is making millions off their apps, other than ones (usually companies) who already have an established base on their iOS version and simply ported it over. The ROI for regular, do it yourself in your basement devs is totally not worth it. The time spent to code, the time spent to support their apps, the time spent on maintenance all for scant return is not worth it, at least in my eyes.

The more that Google doesn't pursue some kind of standard the worse it'll get. It's not as if when we reach the K version of Android, that all phones running Eclair will self destruct. At that point we'll have 9 android versions across almost 100 phones (yes, there are people who still run 1.5, and 1.6), all of which will be able to install apps coded for every other version of the phone. How you don't see the mess that will cause is beyond me.

By the time Android Zebra Stripe Cheesecake comes along, it'll be a complete mess if Google doesn't give the manufacturers or the devs some kind of app/software creation standard. Open source is all great and dandy but it gets out of hand when there are a million different ways to do the same thing...

 

[harrellj]

what you dont hear in that blog is the statistics where when they say 90% of apple products are on the latest iOS version, they are saying version 4.x (so 4.0, 4.1, 4.2 and 4.3) and they compare that to one version of android. 2.3.....if you look at version 2.x for android, we are right up there with apple (i believe we were at 87%) then you have to take into consideration the ipod touch's and ipads as they all run iOS.
i read an article about it but cant seem to find it.

anyway, i agree fragmentation sucks, but to be honest, pretty much every phone out there could run android 2.3, but they dont take the time to money to do it since the latest and greatest is just around the corner....its silly to just throw money in something that doesn't get you any money. and i wouldn't say it is ANDROID (googles) fault for fragmentation, it is the manufacturers fault for insisting on running their skins over top of googles OS. if google forced manufacturers to run stock android, i would guess even a higher percentage would be on 2.x

i'm asking this because i hadn't seen this and not to be smart. where does it say that they're talking about 4.x when they mention the "latest version" of iOS and not the actual latest version?

The article you originally linked wasn't the original article. From it:

He then gives the following stats:

Binned by major rev:
4.X: 89.73 %
3.X: 10.25 %
2.X: 0.02

And 2 sentences later:

So, iOS 4 has 90 percent share amongst iOS device owners. What about Android 2.3? 0.4 percent, as of a couple weeks ago. Yes, that’s zero point four percent.

And where he admits that comparing to just that version is somewhat silly:

But for the sake of this being slightly more fair, let’s compare iOS 4 to Android 2.2 — an OS which came out well before iOS 4. The adoption rate there? 51.8 percent. That’s still pretty pathetic.

However, the iOS numbers are major revisions. By the numbering scheme, Android 2.2 is a minor revision of 2.0 and 2.1. The accurate comparison would be iOS 4.X to Android 2.X, which would be 90% compared to ~87%.

 

[czerdrill]

what you dont hear in that blog is the statistics where when they say 90% of apple products are on the latest iOS version, they are saying version 4.x (so 4.0, 4.1, 4.2 and 4.3) and they compare that to one version of android. 2.3.....if you look at version 2.x for android, we are right up there with apple (i believe we were at 87%) then you have to take into consideration the ipod touch's and ipads as they all run iOS.
i read an article about it but cant seem to find it.

anyway, i agree fragmentation sucks, but to be honest, pretty much every phone out there could run android 2.3, but they dont take the time to money to do it since the latest and greatest is just around the corner....its silly to just throw money in something that doesn't get you any money. and i wouldn't say it is ANDROID (googles) fault for fragmentation, it is the manufacturers fault for insisting on running their skins over top of googles OS. if google forced manufacturers to run stock android, i would guess even a higher percentage would be on 2.x

i'm asking this because i hadn't seen this and not to be smart. where does it say that they're talking about 4.x when they mention the "latest version" of iOS and not the actual latest version?

The article you originally linked wasn't the original article. From it:

And 2 sentences later:

So, iOS 4 has 90 percent share amongst iOS device owners. What about Android 2.3? 0.4 percent, as of a couple weeks ago. Yes, that’s zero point four percent.

And where he admits that comparing to just that version is somewhat silly:

But for the sake of this being slightly more fair, let’s compare iOS 4 to Android 2.2 — an OS which came out well before iOS 4. The adoption rate there? 51.8 percent. That’s still pretty pathetic.

However, the iOS numbers are major revisions. By the numbering scheme, Android 2.2 is a minor revision of 2.0 and 2.1. The accurate comparison would be iOS 4.X to Android 2.X, which would be 90% compared to ~87%.

That wouldn't be an accurate comparison by your own admission. Froyo, gb, eclair etc are major revisions. You can't compare three major revisions to one os. That's like comparing windows xp, vista and windows 7 collectively to ubuntu 10.10

Sent from my Droid using Tapatalk

 

[czerdrill]

i'm asking this because i hadn't seen this and not to be smart. where does it say that they're talking about 4.x when they mention the "latest version" of iOS and not the actual latest version?

The article you originally linked wasn't the original article. From it:

And 2 sentences later:

And where he admits that comparing to just that version is somewhat silly:

But for the sake of this being slightly more fair, let’s compare iOS 4 to Android 2.2 — an OS which came out well before iOS 4. The adoption rate there? 51.8 percent. That’s still pretty pathetic.

However, the iOS numbers are major revisions. By the numbering scheme, Android 2.2 is a minor revision of 2.0 and 2.1. The accurate comparison would be iOS 4.X to Android 2.X, which would be 90% compared to ~87%.

? No that wouldn't be an accurate comparison by your own admission. Froyo, gb, eclair etc are major revisions. You can't compare three major revisions to one os. That's like comparing windows xp, vista and windows 7 to ubuntu 10.10.

Sent from my Droid using Tapatalk

Sent from my Droid using Tapatalk

 

[PikaCane]

But we're not talking 5 OS's on one "D"...we're talking 5 OS's on 60+"D's". And at the end of 2011 it'll probably be 7 OS's on 80+"D's"...as it becomes more fragmented the ROI becomes lower not higher. No android dev is making millions off their apps, other than ones (usually companies) who already have an established base on their iOS version and simply ported it over. The ROI for regular, do it yourself in your basement devs is totally not worth it. The time spent to code, the time spent to support their apps, the time spent on maintenance all for scant return is not worth it, at least in my eyes.

The more that Google doesn't pursue some kind of standard the worse it'll get. It's not as if when we reach the K version of Android, that all phones running Eclair will self destruct. At that point we'll have 9 android versions across almost 100 phones (yes, there are people who still run 1.5, and 1.6), all of which will be able to install apps coded for every other version of the phone. How you don't see the mess that will cause is beyond me.

By the time Android Zebra Stripe Cheesecake comes along, it'll be a complete mess if Google doesn't give the manufacturers or the devs some kind of app/software creation standard. Open source is all great and dandy but it gets out of hand when there are a million different ways to do the same thing...

At the end of the day, though, it's technology. It gets outdated, and even if it still works perfectly well, it will no longer run the newest software or OS. My D1 may be running Gingerbread, but it will never run Dungeon Defenders. I accept that as the price of using an older device, just like I don't expect my PowerBook 3400 running Mac OS 7.6 to handle Flash. If I take that relic on the Web, and I do, I only go to carefully chosen sites. Similarly, someone running Cupcake or Donut on a Hero needs to understand that they are using seriously outdated hardware and that eventually it will be left behind. If they can't or won't upgrade, that's their option, but they shouldn't expect mainstream support and cutting edge apps forever. With some luck, there may eventually be a small group of devs who specialize in supporting outdated hardware. Maybe then we'll have D1's running Watermelon


Sent from my Droid using DroidForums App

 

[czerdrill]

But we're not talking 5 OS's on one "D"...we're talking 5 OS's on 60+"D's". And at the end of 2011 it'll probably be 7 OS's on 80+"D's"...as it becomes more fragmented the ROI becomes lower not higher. No android dev is making millions off their apps, other than ones (usually companies) who already have an established base on their iOS version and simply ported it over. The ROI for regular, do it yourself in your basement devs is totally not worth it. The time spent to code, the time spent to support their apps, the time spent on maintenance all for scant return is not worth it, at least in my eyes.

The more that Google doesn't pursue some kind of standard the worse it'll get. It's not as if when we reach the K version of Android, that all phones running Eclair will self destruct. At that point we'll have 9 android versions across almost 100 phones (yes, there are people who still run 1.5, and 1.6), all of which will be able to install apps coded for every other version of the phone. How you don't see the mess that will cause is beyond me.

By the time Android Zebra Stripe Cheesecake comes along, it'll be a complete mess if Google doesn't give the manufacturers or the devs some kind of app/software creation standard. Open source is all great and dandy but it gets out of hand when there are a million different ways to do the same thing...

At the end of the day, though, it's technology. It gets outdated, and even if it still works perfectly well, it will no longer run the newest software or OS. My D1 may be running Gingerbread, but it will never run Dungeon Defenders. I accept that as the price of using an older device, just like I don't expect my PowerBook 3400 running Mac OS 7.6 to handle Flash. If I take that relic on the Web, and I do, I only go to carefully chosen sites. Similarly, someone running Cupcake or Donut on a Hero needs to understand that they are using seriously outdated hardware and that eventually it will be left behind. If they can't or won't upgrade, that's their option, but they shouldn't expect mainstream support and cutting edge apps forever. With some luck, there may eventually be a small group of devs who specialize in supporting outdated hardware. Maybe then we'll have D1's running Watermelon


Sent from my Droid using DroidForums App

I guess I don't consider that the Droid X could be called outdated by the time a Watermelon version comes along (at the rate Google releases OS versions, that'll be in like a year and a couple of months). Of course, we can say "the user should just upgrade" but that's not the best or feasible solution for plenty of people especially considering the X will still be more or less "new"-ish in 2 years. I mean it's less than a year old, and I doubt it'll see anything past Gingerbread!

I think the problem lies with Google just releasing far too many Android versions in too little of a time. The "obsolescence" is not because of technology, rather it's almost a forced obsolescence because of the frequency of os versions and the inability of the carriers/manufacturers to keep up. And that's going to become a mess the more they do this...

 

[jbdow]

czerdrill-

i do love the way your mind works - way to hold and explain a concept!
thanks for making it a very entertaining thread!

r, john

 

[Amagine]

czerdrill-

i do love the way your mind works - way to hold and explain a concept!
thanks for making it a very entertaining thread!

r, john

More lack like of holding a concept. What he is seriously forgetting are two key points.

This issue has been going on since mobile devices came out. Look at Palm, RIM, Nokia... every OS for every device has this issue dude. I went through 3 blackberries with 3 different OS versions before coming to Android. On about the second one I learned I could up date the machine myself .
For devs its a simple fix.. all they have to do is state that this app works on X.X OS. At that point its up to the end user to know if it will work. Or Google can make a function in the marketplace to check both ends for compatibility.
If a dev doesn't support a widely used OS version it's his fault and his investment is lost. I fail to see where a dev HAS to support the older versions. I also fail to see where we argue that we want control over our mobiles, then you say the manufacturer has to keep updating in a timely manner.
Fact is most people get tired of their devices after about 18 months. Unless your super attached to it or other wise can't move.
But saying Google or the carriers have to support this is like asking your car dealer to replace your 2.0 liter engine with a V6 or a V8 in the same car.

 

[PikaCane]

Google has generally released two or three versions a year. Eclair was a bit of an exception, but they slowed the pace back down again for Froyo and Gingerbread. At that rate, we have at least five years. Carriers encourage a max lifespan of two years on a phone with contact upgrades; a five year old phone will be outdated.

However, I think you may have misunderstood my post. I never said the phone would be obsolete. Obsolete implies that it would be unable to perform basic functions and would be nearly useless due to changes in technology. 2G will be useful as legacy technology for a long time; even after it's gone, the phone could be used on WiFi, it just will never make another call. Apps already installed will continue to work. My point was similar to Amagine's: there is no obligation for devs to continually support every version on every device forever. With what, 87% of Android users on 2.X, I'm not surprised that I am starting to see either App and App: Legacy Version or just notes saying "for 2.0+ only" in the Market. No one is going to force users to update until 2G dies and their Dream is as useful as a DynaTac. That doesn't mean they'll be running the newest apps the whole time.

Sent from my Droid using DroidForums App

Edited for typos.

 

[kodiak799]

But we're not talking 5 OS's on one "D"...we're talking 5 OS's on 60+"D's". And at the end of 2011 it'll probably be 7 OS's on 80+"D's"...as it becomes more fragmented the ROI becomes lower not higher.

Again, there are hundreds of different hardware combinations on the PC. And you continue to talk about fragementation in a vacuum - becoming more fragmented is a result of choice and technological advancement, it is not an issue if the installed base is growing faster than the relative fragmentation. PC's prove this because, as you note, no one talks about fragmentation there eventhough it is CLEARLY more fragmented than Android.

Again, given how the Android market is growing fragmentation is not hurting the ROI, or at least development is not being discouraged. Facts are facts. It's a popular little buzzword right now that is, to date, making a mountain out of a molehill.

 

[czerdrill]

But we're not talking 5 OS's on one "D"...we're talking 5 OS's on 60+"D's". And at the end of 2011 it'll probably be 7 OS's on 80+"D's"...as it becomes more fragmented the ROI becomes lower not higher.

Again, there are hundreds of different hardware combinations on the PC. And you continue to talk about fragementation in a vacuum - becoming more fragmented is a result of choice and technological advancement, it is not an issue if the installed base is growing faster than the relative fragmentation. PC's prove this because, as you note, no one talks about fragmentation there even though it is CLEARLY more fragmented than Android.

Again, given how the Android market is growing fragmentation is not hurting the ROI, or at least development is not being discouraged. Facts are facts. It's a popular little buzzword right now that is, to date, making a mountain out of a molehill.

ROI for Android devs is a joke. They might as well not bother selling their apps because like I mentioned no developer is making hundreds of thousands of dollars (or even any appreciable amount ROI wise) on the Android market or anywhere in the android ecosystem. And that's because of fragmentation. More time is spent getting out all the bugs across the OS versions then there is fine tuning their product.

ROI is low as it is, and it's not going to go higher. The entire world can buy android phones and ROI is not going to skyrocket. ROI is almost irrelevant in Android.

Saying "people should just upgrade their phone" or "devs should just say no support for 2.X or 1.X" is a foolish suggestion and not a practical one. The change needs to come from Google, not from the individual devs or users. The fact that that's even being suggested is ridiculous in itself. The reason? Fragmentation. Let's say a dev says "Ok forget this crap, I'm only supporting Gingerbread"...so he goes ahead and codes his app for Gingerbread, and refuses support for any other version...hmmm...considering that what is it 0.4% of Android users have Gingerbread (and no ETA whatsoever on when more people will get Gingerbread) are his ROIs going to be amazing? Nah I doubt it. And GB is an extreme example, but let's say he decides only Froyo. Ok so he isolates those who have Eclair (35%!!), Gingerbread and any other version that comes out after that. An Android dev can't just say "Screw it, I'm supporting one OS version" when they're being pumped out at the rate that they are...growth of the market has nothing to do with it. Your PC issue is irrelevant and ROI is far higher on PCs because Microsoft is not releasing Windows versions 5 times a year. Devs don't have to scramble to support their software on Windows 7 and also try to quell the angry group of Windows 3.1 users who can't seem to make it work for them...

On the manufacturer/OEM side, you have to support the phones you put out. You can't just introduce a phone with Froyo and say "We will only be supporting our Gingerbread phones"...that a recipe for disaster for any company...consequently, costs to support all those versions rise too

Backwards compatibility should be a given...but we all know that it's not in Android. If it was we wouldn't have maintenance updates for all of our apps every time a new OS version comes out. Whatever changes are being made in the codes are removing compatibility from previous versions. It's not going to get better if they continue at this rate.

The only solution is Google has to slow down with the updates to the OS. We can probably argue that since the Android "boom" is relatively new that they're playing catch up to all the other OSs out there and they'll hopefully slow down and backwards compatibility won't be an issue anymore. That is the only solution.

Downplaying the problem just because you're not directly affected by it is not the solution.

Ask any dev if he/she wants to put out an app and then have to maintain it and support it across 5 OS's...you will get a No from all of them, trust me.

Wait and see when GB starts rolling out. You will be spending much of your phone time updating your apps in the market because they'll start force closing. It's fine and dandy if you're the user and just have to press a button that says "Update", it's another story when you're the dev who has to figure out what the heck is going on