Android Versions by The Millions

Last month I blogged about the missing market share progress graph that Google used to publish. And also provided some extra graphs based on the collected data, with much interesting facts to extract from them. I updated these data/graphs with the latest Play Store stats. But there was something significant missing and what really matters for app developers: how many actual people are on the versions, in millions.

To get this information, we need to know how many active users there are. We never had this information until Google I/O 2014 where Sundar Pichai announced that there are 1 billion users active on the Play Store at that time. The other regular piece of information I could find was the quarterly worldwide shipment of Android devices since 2010 up to April 2014. Using this data and a lifetime of 18 months for each shipped device, I managed to reconstruct the progression in millions of active devices and get to the 1 billion number we have now. The math may not be all sound, but in the end the growth is pretty linear from the beginning and the milestones from each IO keynote seem to coincide (number of activation vs active device). All these data are added to the original spreadsheet in the page "Active Users".

Given these grossly accurate data I could build the graph of each version progression in millions of users, not just in market share.

The road to 1 billion has been pretty linear. The last quarter global shipment are unknown yet. And they also take in account an explosive growth in China where the Play Store is not available.

Another interesting graph, and the real information I was looking for is how many users are currently using each API.

To compare with the original one

You can see the story is very different.
  • KitKat was the fastest growing platform in recent Android history and it's showing even more by million of users. If the growth continue like that it may reach 300 million users in the next 3/4 months. Before Android L comes out.
  • Although API v16 has been slowly declining for a while, the platform was still growing and so the number of users was still growing, the market share alone is not a good indicator. The number of users are in free fall though, despite still being the dominant API.
  • API v10 has still 140 millions of active users, these are not Chinese users.
  • There were never more Ice Cream Sandwich users than Gingerbread users. It topped 200 millions, compared to 300 millions for Gingerbread.
  • The growth of API v17 is more significant when taking in account the amount of users, it's still growing well.
  • On the other hand, API v18 is still not very meaningful in the number of users.
  • There are still 7 million users using v8.
The good news is that supporting v17 and up gives you a very good amount of users. But failing to support v10 or v15 gives 250 million more potential users to your competitors.

We can assume that in the next 3 or 4 months v10 and v15 will drop below 100 million users each. And v19 should reach 250 to 300 million users and might have more users than v17.


Android Version Distribution

A long time ago, Google used to provide a graph of the evolution of version distribution with their monthly update of the Android Dashboards. I missed this graph ever since because it was giving a good indication of where we're at and what to expect in the coming months. This is especially important when planning a new project, to know what most users will have when your product ships.

After digging in the Internet Archive Wayback Machine, I reconstituted all the data published by the Play Store since December 2009 up to now (June 2014). The result is this bare spreadsheet table with a link to the source I used for each line.

Reconstructed Play Store Statistics
With the data in hand, it's now easy to create the graph that Google used to publish. But as soon as you see it (see in the Google Doc spreadsheet, after the numbers), you realize it makes more sens to group the Android versions by their codename (Froyo, Gingerbread, Honeycomb, Ice Cream Sandwich, Jellybean and Kit-Kat). This gives this chart:

This graph looks familiar, just with more versions in it. Each major version seems to have the same life cycle and you can see that Jellybean is currently very dominant, that you get from the official monthly pie chart. What you don't get is the idea of how things are probable to move in the next 3 to 6 months. For example Gingerbread and lower currently represent 15% of the active Play Store users. When will it reach 10% ? Judging by previous Android versions, it took 4 months to get from 15% to 10%. So that would be in October of 2014.

This area view is nice, but there are other ways to represent the evolution by simply plotting the numbers in a line graph, as follows:

This graph, IMO, gives a much better idea of how things went for each version and the real importance of each version. Here are some noteworthy points:
  • Each major version has a similar life cycle. A rapid growth and then a slow "logarithmic" decline.
  • The decline starts 4 months after the next main Android version is released (4 for Froyo, 5 for Gingerbread, 4 for ICS, 4 for Jellybean)
  • Honeycomb never had much of an impact
  • Ice Cream Sandwich was never more popular than Gingerbread
  • During Google I/O 2013, the most used Android version was Gingerbread (so much for minSdkVersion=14)
  • Around April 2013 Froyo and Gingerbread lost a lot of market share at the benefit of Jellybean (harder to see in the area graph)

There's still another way to plot the data, from the time the version was introduced and counting how many months it was in use. That gives the following graph:

There's plenty of extra information that can be found from this graph.
  • Since Gingerbread, the evolution during the first months of each version is very similar
  • Older versions of Android were growing more rapidly to their peak
  • After a slow start Kit-Kat has caught up with the growth of Jellybean
  • We can project that Gingerbread will be below 10% in 4 months, ICS will be below than 10% in 2 months 

With Google I/O 2014 on the way and Android 5.0 likely to be revealed there, we can get an idea of the fate of Kit-Kat. It will be similar to the one of Ice Cream Sandwich. Of course, Jellybean is a bit cheating here, since it had 3 major versions. Here is a version of the graph, not grouped by main versions and starting at API v7.

More points can be found from this graph:
  • Obviously the older versions had less versions to share with so had more market share.
  • In the recent years API v16 is the one with the most market share by itself.
  • Kit-Kat (v19) is growing faster than all the versions since Gingerbread (except for API v18)
  • API v15 has reached its peak 15 months ago
  • API v16 has reached its peak 6 months ago
  • API v17 is growing very slowly but still growing
  • All popular recent versions reached their peak around 15 months of existence
  • Versions like v14 or v9 almost never existed

Conclusion / TL;DR: Know Android Users Through Fragmentation Graphs



A while back I created a company in France on the side of my day job in hope to make a little cash on the side. I created GAWST. My first idea was to make some programs for iOS and OS X as people to pay more for softwares there. I had various projects lined up. The more advanced one was a replacement for the crap Finder, something more in tune with Windows Explorer. Then Apple introduced the App Store and plenty of rules and restrictions. It not possible anymore to sell a file explorer there. The App Store being the easiest and most obvious way to sell content there, I decided it wasn't worth the efforts (working on OS X was never my cup of tea).

So I switch to Android. I wanted something like Ghostery for Android, a way to see the kind of traffic goes to sites I don't want and a way to block it. So I started working on a proxy written from scratch in java. Something lightweight and fast. It took me a while to get it right but now it's pretty stable. I can see and block all the traffic just like I wanted. And I can do a lot more with this little tool. But then last week Google removed all the ad blockers from the Play Store. I planned to sell a restriction unlocker via the Play Store (while basic functionalities would be available to everyone). And once again all the time I spent on this project is wasted because a large company wants to have more control with what we can do with our machine.

I don't know yet what I will do with the code of gawsttp (my proxy). I use it every day on all my devices and it's useful, even a debug tool (or to test Twitter's API blocking). But what is pretty clear is that I won't try again to make a commercial project of my own anytime soon. I'm tired of all the rules that are only good for the big players, despite real user demand.



When I read this discussion on Mozilla Dev thinking about supporting H.264 I felt like I had to say something. Not just as the creator of Matroska (on which WebM is based) but also as a general net citizen.

Ever since I started working on video technology I realized how vast and encompassing the subject is. In this particular case it's about letting patents dominate a fundamental web technology. It's about who is controlling the technological stack when you go on the Internet (aka market shares). It's about a technical decision that is actually changing the whole principle of the web: the technologies it uses have to be Royalty Free, be it on the user side, on the server side or on the creation tools side.

So far Mozilla was opposed to using H.264, not for technological reasons but on the principle that it didn't fit the Royalty Free web. They were also saying it was “not good for free softwares”, that if used “you [couldn't] have a completely free software Web client stack”, that would be “honoring the letter while violating the spirit” or that it “is not a game [they] are interested in playing” and eventually “it pushes the software freedom issues from the browser (where [they] have leverage to possibly change the codec situation) to the platform (where there is no such leverage)”. None of what was said then has changed. So what has changed must be Mozilla.

When BootToGecko was announced my first reaction was whether they would support H.264 when it's hardcoded in the phone chipset. Because answering this question has a lot of repercussions. The amount of reaction to their answer is proof that the issue is deep. In fact it goes further than just H.264 which is hardcoded and paid for on the hardware (I suppose a lawyer should double check on that to be sure). It also means supporting the MP4 file format, which is also patented but not hardcoded in chipsets, unless they plan to only support H.264 inside Matroska (they already have the WebM code to do it easily), but I doubt that. At some point they will likely support MPEG DASH which made the W3C uneasy because it's not yet a Royalty Free technology. And there may be other patented technologies in the future.

The W3C will never endorse the use of the H.264 codec, it doesn't fit its rules. Vendors using it for the <video> tag are just using a backdoor to put patented technologies on the web, where it should have never been in the first place. Now the message is sent loud and clear, there are no reason big companies shouldn't try to impose their technological dominance on the web (Skype and Facebook come to mind). Everyone will surrender in the end. The bigger you are, the biggest momentum you have, the more chance you have to succeed. And patents are just as virus like as the GPL license: once you touch them you can't escape from them...

Maybe Mozilla is just acknowledging the fact that the web is going away from its core roots and what made it such a global success. But it still seems shortsighted to me. The timing of their announcement comes just weeks/months before new devices with VP8 hardware decoding are widely released. If they wanted to emphasis their continued support/favor for WebM they could at least have first released B2G on a device that has VP8 capabilities. Instead the message sent to those wondering if they should support WebM is: don't bother anymore, in the end we'll have H.264 everywhere. Such a decision to kill WebM should have been taken collectively. Especially since it's just to satisfy the need of a project that currently has 0 users and will have to compete against Apple, Google and Microsoft on the smartphone/tablet market. Billion dollar companies that heavily depend on this market for their future. Whereas Mozilla's future depends on product placement from the same companies. But that may change if I read this correctly: “We will not require anyone to pay for Firefox. We will not burden our downstream source redistributors with royalty fees.“ Payment not required, but possible ?

To be fair, Google deserves a big share of the blame as well. Even though they provided VP8 to WebM, they didn't put all their weight in the battle. They have done too little too late. And there is no sign of that changing. They even failed to deliver on their promise of abandoning H.264 in Chrome to favor the technology they built and is perfect for the web. If there ever was a strategy for WebM I have never heard of it.

Good reads on the subject:


The Open Web Conspiracy

So Google has announced that it will drop support for H.264 in future iterations of its Chrome web browser. WebM will be the favored format for video rendering inside the web browser. Until this announcement very little people really knew about WebM. It was considered as a pet project of Google, like so many others. But with all the press of the last few days, it's impossible to ignore WebM anymore.

What people didn't realize is that Google paid a good amount of money for this technology. Not to make even more money out of it, but simply because the video inside the web browser was going nowhere. It was either the expensive H.264 or the "low" quality Theora for those who can't afford it. So Google bought On2 and VP8 with it and made it free and completely open. The exact kind of technology that is mandatory in W3C standards. They also converted most of their YouTube library to WebM (in 360p and 720p IIRC) which probably also costs a lot in storage.

The tone of most reactions to this move was "how dare you Google?". Which is pretty astonishing when you think about it. People blamed Google for abandoning H.264 in a web browser. Completely ignoring the fact that Mozilla and Opera have always been on that position. AFAIK the web has not collapsed because of this decision. In fact FireFox is now the dominant browser in Europe and Chrome is the fastest growing one. I don't see how this move is going to hurt anybody already serving video on the web.

Not only that but Mozilla and Opera are the ones to thank for the existence of HTML5, the ones that stood against MSIE and it's proprietary way of interpreting the CSS standards. Once again there's going to be a war on standards. This time Google is joining the party. And once again the standard should win in the end. It's not a matter of who has the biggest one, but what an open web means and doesn't mean.

Everyone who has been involved in the Audio/Video business knows how much patents are a nightmare to deal with. A real way of stopping innovation. It also costs a lot of money for small players, to the point were real business is almost impossible if you don't have a size large enough. At CoreCodec a large amount of our revenues were going each year to the MPEG LA for the use of MP3, H264 and AAC. On many products we didn't make any money other than what we had to pay for patents. If there are better business alternatives, everyone should embrace them.

In the flow of articles/comments I read a common argument coming back was that H.264 is already there. Yes, but almost noone is using the <video> tag right now. It's only at the experimental stage. They confuse the web browser and the entire video ecosystem. Everyone who's interested in the story of that <video> tag knows that although a great idea it was a problem due to browser fragmentation and the lack of agreement on the codec, H.264 being out of the question for many key players and likely the W3C who approves the standards. Now there's a solution in sight. Again a solution for the web browser not for the whole world of video. It's not a first arrived/first served system. If H.264 doesn't fit the bill there is no point continuing (or ever) supporting it. Because in the end something else will have to be used. This fact has always been known by all users of the <video> tag from the beginning.

Also it's not because something is there that any future development should stop. In fact everyone working with the <video> tag knows that it's an evolving technology that is bound to change until it stabilize. There are a lot of things coming like adaptive streaming (IMO the most important part of streaming that will make browsers useful players), transparency (for better integration of video in rich UI), 3D (3D WebM files are already playable on YouTube) and even likely some form of DRM if the browser is going to become the universal "cloud" video player. All of these changes will require many investment for people providing online video. The sooner they know what are the possible choices and the ones that will be the best investment, the better.

Plus the video on the web is still in infancy. Adaptive streaming is coming (and IMO the real stopper right now for real use of the browser as a video player), live streaming is coming, 3D is coming (should browsers support MVC too like on Blu-Ray?), transparent videos (for better integration in UIs) and likely DRMs will come too if video rentals are ever going to work without Flash/Silverlight on a desktop. So don't assume that was exists now is set is stone and should never change. All these technologies will require work and money from the people serving videos. The earliest they know the options and the likeliness of success, the better choices they can make. And there is no reason investing time and money in something that will not last in the end. What is clear now is that WebM is not a pet project, it's here to stay.

I've been involved in the development of WebM even before it was out (no surprise that I'm a supporter then). And one thing that stroke me so far was that the most active in its development were Mozilla (3D and live streaming) and Opera (live streaming). Chrome has always been trailing with new features of WebM or even bug fixes. Now Google is finally putting its money where its mouth is. It's being pragmatic AND ideological, not one OR the other. The support of Flash being the pragmatic part here. Not only that but Adobe has been a supporter of WebM since day one (and even before). I wouldn't be surprised if Flash would support WebM in the future. That would make sense for them too. If that ever happens, WebM will be the format that would play on all platforms, unlike H.264. And would likely be decisive in making WebM the first choice when putting video on the web.

In the end one can always think of it as a big conspiracy from Google, Mozilla and Opera to free the web from audio/video patents, and keep the World Wide Web utopia alive and kicking.


2011: The Year Apple Went Back To Its Niche

Hello I'm Steve Lhomme. You may know me from the Matroska (.mkv and .webm) format I (mostly) created, or working for years on CorePlayer from CoreCodec on platforms like Windows Mobile (5 & 6), Symbian, PalmOS, iOS, Android, etc or lately from my work on the Plume Twitter client on Android for LevelUp Studio.

For many years I've followed and been involved in the rise of mobile computing and smartphones in particular. I've used and developed for about any of these OS'es (that would allow some form of native code). In that time I've seen the growth of the iPhone (from the early version without a SDK where we had CorePlayer already running) up to the point it became almost a world of its own. And with great power comes great responsibility. Except that Apple has been acting very aggressively, for their own profit.

Recently I've started noticing an unfair bias towards Apple from the press and from friends. Either because one gives credit to Apple for more than they deserve or because it gives advantages to the Apple platform that IMO is not good (more on that later). While Apple was mainly the only player with a modern platform, it was fine. But now that Android is on level with iOS and exploding in market shares, it's time to rethink the old habits.

The current situation is that on the OS level Android is more or less at the level of what iOS offers. The OS was not meant for tablets and thus the ones sold so far leave to be desired. It seems that Honeycomb (Android 2.4 it seems) will leap forward and make Android more than decent on tablets. I personally think the tablet market is overrated, but that's another debate. On the hardware level there is certainly a lot more experiment & innovations on the Android side. We now have dual-core, tiny devices, up to 4.3" screens, 4G/LTE, car dashboards, 5" and 7" tablets, tablets as TV companions from Vizio or Panasonic, etc. And depending on your needs, there's usually one device that is exactly what you need. Just like not everybody wants to be dressed the same, not everybody needs the same from their phone/tablet.

But with all the CES announcements, I've seen plenty of news/comments on Apple fanboy sites, that whatever, the iPhone/iPads are better. Even before Honeycomb was demonstrated and devices were tested by real persons. For many Apple has become a religion to follow, and denying all other "gods" that are not theirs. This is not new (Mac vs PC). But it's always surprising when it comes from bright minds. And it's even more dangerous when a whole economy has been built on feeding solely the iOS ecosystem (and 30% of it in Apple's pockets).

Things have changed radically since the last 6/12 months. Android is now dominant globally and even in the symbolic USA market. Unlike iOS it's completely free and open. That's why hardware and software innovations are now happening there. And the trend is not going to change. The head start that Apple had is now gone.

The freedom in Android means anyone with a Windows, Mac, Linux computer can develop for free their software and run it on their devices without paying anything to Google. That is not possible with iOS. One can easily see which one is going to be picked by coders in developing markets. Not only that, but Android doesn't require a PC for synchronization or system updates. In doesn't require you to put your billing information in iTunes before buying apps. In a world where more and more people use a (smart)phone and not a PC, that's an important growth factor. One may argue that poor people don't buy apps. But they are likely to buy food, detergents, gas, etc. So the advertising model can work in these areas. So I think the smartphones are going to be a lot bigger there too. It's a lot better if it doesn't require a PC.

Because of all this, I think Android is going to explode even more this year. And unless Apple has something special about to be released (rumors don't seem to show that), it's going to lose even more significant market shares. And like the Mac vs PC war, it's going to end up in a niche for trendy/hip/rich people. Which is probably fine for Apple as long as it has a good share of the #2 position. History is just repeating.

NOT sent from my iPhone (yes, I still use one)


Upgrade your music/movies

So a lot of people are talking Steve Jobs comments on the DRM issues. I haven't read it yet. But it made me think about the digital economy for content/entertainment.

People have been buying music at 128 kbps (MP3 or AAC or WMA) or DVDs in 480i resolution, and with DRM. If they want to get a better quality music/video or without DRM, they would probably have to buy it again. In software you usually get free upgrades or pay less than the full product for an upgrade (even Microsoft does it). So it should be the same for music. If you want to buy the same music in lossless or transparent, you shouldn't have to buy the whole thing again. That was necessary when you had to buy a physical object to upgrade your qualiyt (tape/LP to CD, VHS to DVD). But from digital to digital, there's no need. The only problem is to do that you need some DRMs :(

So the non use of DRM might actually mean there won't be cheap upgrades... How ironic !