Google updated the stats in its Device Dashboard a couple of days ago and the results make for encouraging reading. As noted in my last blog Android 1.5 penetration surged unexpectedly between January and April 2010. That trend has been reversed this month however with Android 2.1 taking just under 10% share from both 1.5 and 1.6 to become the most dominant version of Android OS with 37.2% penetration. Developers still cannot afford to ignore 1.5 and 1.6 however as collectively they represent over 60% of devices visiting the Android Marketplace.
The trend in Android OS distribution can be seen in the following Google pie charts for January, April and May 2010 respectively:
As noted last time, the odd feature is the surge in Android 1.5 usage in the April 2010 pie chart, that trend has now been reversed with both the 1.5 and 1.6 slices shrinking in size and the 2.1 slice increasing in size.
Here's an update of my own rough "changing fortunes" bar graph from last time charting distribution changes January - May 2010. Some people suggested a line graph may better show over trends so there's one of these also. Now you have the full set, pie graphs from Google and bar and line graphs from me.
The Cyan bars in the bar graph represent where we are today:
Now onto the good stuff. In addition to providing a breakdown of OS distribution Google is now also providing a breakdown of the screen sizes and densities that it is seeing visiting the marketplace:
This is particularly useful information. Android 1.5 supports only normal sized screens (320x480) with medium pixel density (160 dpi). Android 1.6 and later on the other hand supports 9 different combinations of screen size and pixel density. Our advice has typically been that developers should target Android 1.6 to future proof an application for the introduction of various screen sizes going forward. That advice still stands but what the Google data shows us is that there is very little immediate cost to ignoring this advice and simply targeting Android 1.5 using the normal screen size/medium density combination because this combination is used by nearly two thirds of all Android devices accessing the marketplace today and even though Android 1.5 does not provide the tools to support other screen size/density combinations, since the only other combination worth supporting is also a normal screen size (hence the same aspect ratio) those devices will simply upscale the medium density image for the high density screen. Sure it won't look as crisp as a high density image, but it will work.
So provided you're not using any Android 1.6 or later specific features your Android 1.5 app will work and look good on 99% of Android devices accessing the market today. Interesting huh!
Back in February I talked about the Android Dashboard which shows the distribution of versions of the Android OS on devices visiting the Android Marketplace. Very handy for deciding which versions of the Android OS to target. The stats at that time told a fairly straight forward story, the majority of users using Android 1.6, some marooned on Android 1.5 and a growing number on the latest and greatest Android 2.0/2.01.
Google recently released updated stats which give a rather more confused picture when contrasted with the earlier statistics:
The first pie graph is for January 2010, the second for April 2010. Rather incredibly Android 1.5 penetration has surged. There are a huge number of Android devices sold daily so the total size of the pie has increased over the four months separating the two graphs. This means that the increase in Android 1.5 usage cannot be explained by users of the HTC Hero and similar devices being marooned on Android 1.5 but instead must mean vendors continue to ship large numbers of Android 1.5 devices. In the PC world that would be akin to PC vendors shipping large numbers of Win 2K machines today .
Relative Android 1.6 usage has shrunk consistent with existing 1.6 devices having been upgraded or more likely, simply less 1.6 based devices shipping. Not surprisingly the relative share of 2.0/2.01 devices has all but evaporated as those devices have now almost all been updated to Android 2.1.
Here's my own rough bar graph of the changing fortunes of the various Android platform versions between January and April 2010:
The purple bars represent January and yellow April. You can clearly see the unexpected Android 1.5 shenanigans.
This makes life interesting for Android developers such as ourselves. Our advice this year has been to upgrade apps to Android 1.6 and target new Android apps at Android 1.6 to take advantage of the screen size and other capabilities introduced in that version. It was solid advice when Android 1.5 looked to be on the way out. Despite these new stats I still think its the best approach but the arguments for targeting Android 1.6 in the very immediate term aren't as clear cut as they were just a couple of months ago!
The last time I was gripped by this feeling, and perhaps the last time anyone was moved by this feeling for a Nokia product, was before I bought my N95 8GB. An absolutely brilliant phone which naturally I just had to have. The fun part was that the excuses I used to justify the purchase of this device were unexpectedly borne out when its TV-out proved to be the perfect no hassle way of blowing clients away during an app demo. The N95's successors have never quite managed to generate the same cachet although some may argue the case for the N900. Maybe Nokia's PR just wasn't up to it but mostly I think its devices and the aging Symbian OS powering them simply couldn't cut the mustard. Silk purse and sows ear anyone?
Nokia announced it's latest superphone contender last week, the N8:
The N8 has all the hallmarks of a classic Nokia superphone - serious hardware. A 12mp camera, HD video recording and playback, HDMI connectivity etc. It's styling is solid if unimaginative (the black and silver look good, the orange and other colours, questionable but my aesthetics are my own). Best of all Nokia has finally shifted from the incredibly tired 434 mhz CPU's that its been using forever. The N8 uses a 680 mHz ARM 11 with a dedicated graphics processor unit. While this still isn't the 1 ghz ARM Snapdragon packed by Google's Nexus One it is on par with the iPhone 3GS' 833 mhz (underclocked to 600 mhz) ARM Cortex-A8. I know, mhz, like money, isn't everything, but more is often better!
The two single biggest complaints levelled against Nokia's smartphones over the last couple of years has been sluggish performance and a tired OS. With the hardware the N8's shipping with, particularly the processor upgrade, hopefully sluggish performance will be a thing of the past. The big question mark with this phone really hangs over it's OS - Symbian^3.
I have a couple of concerns about the N8 and its shiny new OS:
Firstly, Symbian^3 is a stop-gap release, the stepping stone to Symbian^4 which both Nokia and Symbian view as their pitch at a high impact OS. With so much attention focused on Symbian^4 just how much love is Symbian giving Symbian^3? Nothing's been said to my knowledge about the ability to upgrade your N8 to Symbian^4 in due course but on past form that won't be possible. If you buy this device you'll be marooned on Symbian^3 forever. Not fatal as you'll still be able to run QT based apps that target both Symbian^3 and Symbian^4 but you won't be able to take advantage of any apps that just target Symbian^4 such as any app taking advantage of the improvements Symbian^4 introduces. I'd be feeling a whole lot more comfortable about buying this phone if I knew that it was going to be upgradeable to Symbian^4 in due course.
Secondly, Nokia doesn't have a great recent track record for releasing solid devices on a new OS - the N97 and 5800, its first Symbian 5th edn devices, were riddled with bugs which have had to be addressed with many many successive waves of firmware updates. Nokia itself has acknowledged that it did a shocking job, the devices were launched when the hardware, not the software was ready. Comfortingly these devices work like clockwork today but it was a rough ride for early adopters. If Nokia had released the devices with the firmware they use today then they would have been much more successful but then of course by the time the firmware was finally bug free the devices may have been too outdated to launch I suppose. It's worth noting in this regard though that by all reports Nokia's N900 launch (using Nokia's Maemo 5 OS) was brilliantly orchestrated. Perhaps Nokia should consider getting some devices out to reviewers before the big day to reassure those with N97 scars that this time things will be different.
So is the N8 the phone to bring me back? Has Nokia done enough? Enough to give me itchy fingers at least, yes I can feel this phone in my hand already, but its not something that I'm lusting after yet. It's a while yet before this phone launches so there's plenty of time for me to catch the bug however and this is a phone I can see myself falling for if Nokia can just give me a little reassurance. Ah hell, Nokia, why don't you just send me one? I'll put all our history behind us and write nice things about it . promise!
How is everyone else feeling about this phone?
After many years treading water the mobile world has evolved at a staggering pace in the last couple of years. RIM have made a valiant attempt to keep up but for my money haven't done a great job of it. I've said it before, even though I don't use an iPhone it truly is an iconic device. It has shown all other vendors where they need to go with their devices. RIM however has struggled on the device front. It has a very loyal following and posts unbelievable increases each quarter so its in much better shape than a lot of vendors but outside of its traditional enterprise market it hasn't been able to make much of a splash. This is because RIM and its Blackberry devices are basically defined by the enterprise market. This is why you can't just enter in your pop3 mailbox and password and get your email and this is why RIM is struggling in the consumer market.
Increasingly I've been thinking that enterprise users would eventually tyre of RIM's ugly duckling devices and ossified OS and go for the cool "consumer" devices with ever improving enterprise features such as the iPhone or HTC Desire or the ever dependable Symbian devices from Nokia.
So when a YouTube video featuring highlights from the upcoming Blackberry 6 OS was posted I wasn't expecting much. Boy was I surprised, it looks awesome:
This is night and day different to what I use on my present device. It includes all that you expect from a modern mobile OS with touch features such as system-wide kinetic scrolling and pinch-to-zoom. I didn't see anything innovative - what I did see looked to owe a lot to the iPhone and Symbian. But anyone that's read my earlier blogs will know that I don't care about that so much. When Xerox Parc invented the mouse other people picked it up for use with their computers because it was a damn good idea. Similarly, just because someone else invented multi-touch doesn't mean you shouldn't try and get round their patents and copy the idea as well!
So with an OS to bring it back into the present day RIM now need to design some nice hardware for it to run on.
What do you guys think. Is Blackberry 6 OS looking good or have I just been taken in by a cool video?
We've used Skype in the office since it was in beta. It has been fantastic. Its a great way of firing files between workstations, chatting without disturbing other people, keeping track of what someone told me about something and naturally, a cost effective way of communicating with people remotely.
Skype on mobile makes so much sense to me. Why SMS when you can IM? If you're in wi-fi range why waste your call minutes when you can call for free?
So I've always been frustrated by Skype's inexplicable approach to mobile. For years Skype had a Win mobile client but no Symbian client. That's a pain since I've more often than not had Symbian phones. Skype explained that the user experience wasn't good enough on Symbian which never stacked up as you could just use Fring on Symbian which then used Skype. It made even less sense when in 2006 (yes that's right, four years ago) Skype announced it did actually have a Symbian client after all but no you couldn't use it unless you were on 3 (the mobile network).
Here's a pic circa 2006 of Skype running quite happily on an ancient Nokia 6680 using Symbian S60 2nd Edn:
Recently Skype announced that Symbian is now supported (great, I can throw Fring away at last) and that they've killed the Windows mobile version (bummer, tough break guys). Skype have also announced that they have an Android version, but no you can't use it unless you're with Verizon (again with this, really?!). There used to be a Skype lite Android version but that has been canned and so all non-Verizon Android users are out in the cold.
Unbelievably the new "Skype for Android users on Verizon's network" cannot use wi-fi. The word inexplicable does not even come close. How hard can it be to roll a version that just works on a the leading mobile platforms?
Not that hard apparently as Fring hasn't had any problem doing it. Performance a problem? Hire some smarter engineers from the guys for whom it obviously was not a problem or just reduce the functionality. Diminished functionality is better than no functionality. If a user doesn't like the app then they don't have to use it after all.
Skype I love you but you inexplicably keep frustrating my attempts to spend more money with you! Perhaps your mobile strategy is perfectly rational and I just missed the press release?
Back in December I blogged about "Which mobile OS to develop your killer app for?" and went through the various factors that we typically consider with a client. One of the factors was "What is their target geographic and which devices are prevalent in the geographic?".
I think people too often overlook the importance of regionality in their rush to pick whatever their favourite OS is as the target platform. The importance of regionality was driven home to me once again in some recent Admob stats.
I think this graph (courtesy of Admob) illustrates my point perfectly:
If you are developing an application for Oceania, naturally you would target iPhone OS and potentially Symbian OS. There would appear to be next to no business case for developing a WebOS app. Correspondingly, if you are developing an application targeting Africa, naturally you would naturally target Symbian OS and potentially iPhone OS. You definitely would not develop a Symbian app targeted at North America but you might develop iPhone and Android versions.
What this data tells us is that if your application is going to marketed in a specific region then the huge variations in market share between regions make using "global" statistics dangerous. If you are considering a regionally targeted application then you need to use regionally specific data when making your decision.
The obvious rider to this is that if there are such wide variations regionally, are their subregional variations? The answer of course is Yes! Admob also supplied data for South East Asia:
While Symbian OS has a 69% marketshare in Asia as a whole this increases to 77% in SE Asia. Similarly, while iPhone OS has a 27% marketshare in Asia as a whole this falls to just 19% in SE Asia. In this particular case the data probably would not alter your decision which would be that if you are developing an app that targets SE Asia you should definitely develop a Symbian app and consider an iPhone application. The data does however illustrate that its not always a good idea to even use regional or subregional data. If you are targeting just a single country then you should use data for just that country as part of your decision. Match your data to your target geography.
So regionality is a key consideration when considering which mobile OS to choose. Having read this post however don't jump to the conclusion that its the only thing you need to consider! Remember, as my original December post pointed out, regionality is just one of the things that you should consider before making a final decision.
One last rider, when looking at any data, regional or otherwise, do take the time to understand how it was compiled. I have been rather cavalier in referring to Admob's data because to talk about how it was derived in the middle of things would have just distracted from the whole point I was trying to make. In Admob's case, their data is an OK but not great proxy for the actual mobile OS marketshare as it is compiled from "ad requests we receive from our network of more than 15,000 mobile Web sites and iPhone and Android applications". So of course Admob's data will be skewed in accordance with its collection methodology, self selection bias etc. Admob discuss their methodology here.
I can just hear people now, "Perhaps it would just be easier to choose my favourite OS"!
Earlier today I received an email from "Windows Phone News", I assume because years ago we signed up to the Microsoft Developer Program. It included a link to a Microsoft Demo showcasing the new Windows Phone 7 UI. To be honest we haven't taken much notice of Windows Mobile over the years. Mostly because we've been kept busy enough on the other platforms but for sure there was also an element of none of our clients ever asking us to develop for it as well!
So I thought I'd take a look at this demo. And it actually looks pretty good. I know most demo's do, so we'll just have to wait until devices are released, but if they're true to the demo then Microsoft may have found a way back into the game after its time in the mobile wilderness. The touch based UI seem responsive and intuitive and the graphics not completely ugly.
Best thing by far about the demo was the ability to customise the video backgrounds playing behind the phone in the foreground. Through good luck the first background to play was "Beach", showing a bikini clad nymphets run past and begin frolicking in the water during sunset. There were 7 other video backgrounds that randomly played as well, so something for everyone I guess including Yoga fans.
After luring me to their web page and tantalising me with nymphets Microsoft had me interested enough to look around a bit. A shame that the rest of the site was focused on Windows Mobile 6.5. Even the feature comparison table only compared Windows Mobile 6.0, 6.1 and 6.5. So aside from the nice demo, I'm none the wiser!
Gregs top 10 take-away's from this years Developer Day were, in no particular order:
- Ovi stats show that in Australasia "fremium" is currently the most successful business model for mobile applications.
- The majority of downloads from Ovi today are from touchscreen devices. Nokia's suggestion is that if handling touch and non-touch in the same app is a problem for you, just support touch.
- Ovi globally averages 12 downloads per customer.
- Be careful about introducing advertising into your applications. Mobile devices are intensely personal to users and they can react strongly to intrusive advertising messages. Advertising should absolutely handle orientation changes.
- Eye candy trumps features and functionality. So does useability. So feel free sacrifice features and functionality. Spend money on UI and user experience.
- Symbian Signing will get cheaper on June 9. We've never had a problem with Symbian Signed but it was clearly causing problems for some people and Nokia indicated it would be overhauled to become simpler still (June 9 also).
- The first version of Meego will be available some time in the 2nd quarter of this year, devices will follow in the 3rd or 4th quarter. QT on Maemo/Meego has a long way to go but Nokia are working really hard on it.
- Nokia are planning to bundle data plans with some devices this year (just like it already bundles music with some devices). Sounds interesting but no idea how that will work. Will I be able to buy the same device sans data plan? Presumably not or I could then just immediately work out how much the data bundle is worth to me ...
- Home screen widgets are de rigueur - 80% of the time that people spend looking at a device is spent looking at the home screen. If you want their eye-balls then add a home screen widget to your application.
- While the S60 UI layer (Avkon) may not ship in Symbian^4 you will be able to bundle it your applications for backwards compatibility.
By all reports it was a great day. A long day though since Greg hauled himself to Sydney on the redeye and then back home again on the last flight out!
To finish up, perhaps the coolest thing on the day was the video playing of the N95 power cube solving robot. Perhaps the only think cooler would have been if the robot had actually been there itself. Here's a youtube clip if your interested.
Its just a video of course so whether the reality matches this only time will tell. The video for the most part showcases the multiple home screen and gesture support of the UI. The multiple home screens look clean with the ability to organise and manage widgets. Gestures (swipe, pinch etc) are supported throughout the UI. The video claims that enhanced multitasking and graphics support will make the OS faster and feel more responsive, addressing another long time complaint of users of the Symbian UI.
After viewing the video many commentators have complained that aspects of the Symbian^3 UI appear to be derivative of other platforms. Like being derivative is a big problem. I don't claim that my car is derivative just because it has a steering wheel for interacting with it just like every other car. So when someone comes up with a compelling paradigm for interacting with a device I would expect all other vendors implement it. Over time you would naturally expect similar devices to converge on certain core hardware/software useability and functionality paradigms and then differentiate themselves around the edges. Hence we all interact with our cars using the steering wheel but car manufacturers still manage to differentiate themselves.
The complete list of features (borrowed from this Symbian Foundation press release) announced so far are:
- The Homescreen takes a big step forward with support for multiple pages of widgets and a simple flick gesture to move between them. The widget manager makes discovery and download of new widgets simple and support for multiple instances of a native widget means that consumers can monitor multiple weather forecasts, news feeds, social networking accounts or multiple email accounts simultaneously through a common interface.
- Usability enhancements across the user interface include the adoption of a direct “single tap” interaction model, making it much easier to complete common tasks on a device. Multi-touch support for gestures such as “pinch-to-zoom” forms the basis of a gesture framework that can be extended and leveraged by the developer community.
- One-click connectivity for all applications greatly simplifies the process of connecting to the Internet, without interrupting the user. New global settings allow the user to configure platform-wide behaviour, for example ensuring the device automatically switches from cellular to WLAN when a free WLAN network is available.
- More efficient memory management due to Writeable Data Paging allows more applications to run in parallel for a faster, more complete and efficient multi-tasking experience, especially on mid-range hardware.
- A new 2D and 3D graphics architecture takes full advantage of the hardware acceleration available to deliver a faster and more responsive user interface. Users, developers and device creators will all benefit greatly from the visual enhancements and smooth transitions that will significantly improve the look-and-feel of their applications and services. Combined with industry-standard OpenGL ES, the new architecture also provides a great platform for high performance games – all without slowing the phone down.
- HDMI support enables users to plug their phone into a TV and watch a high-definition movie at 1080p quality without a Blu-ray player.
- Music store integration embedded within the radio enables users to identify a song and learn more about it. The addition of a “buy now” button, which links with the user’s chosen music store, makes purchasing easy.
- The industry-leading networking architecture, ready for 4G networks, provides next-generation Internet experiences on today’s devices. Consumers will benefit from the architecture's ability to seamlessly balance each individual application’s needs regarding factors such as bandwidth, latency and jitter. This improves the consumer’s experience of network-dependent applications and Internet services like VoIP and media content streaming.
Aside from a crumby UI one of the long standing problems with Symbian has been the low hardware specifications used by the device vendors. These are superphones folks, and they need super processes. When you're selling something for top dollar you can't scrimp on the processor. You can talk all you want about a strategy of choosing the n-1 generation of processor for its smaller size and better power efficiency but if that results in a sluggish user experience on the device itself then it just isn't worth it. Don't know what I'm talking about? The iPhone 3GS ships with a Samsung S5PC100 ARM Cortex-A8 833 MHz processor underclocked to 600 MHz. The Google Nexus One ships with a 1 GHz Qualcomm QSD 8250 Snapdragon ARM processor. Nokia's N97 ships with a 434 MHz ARM11 processor. So even underclocked the iPhone has a processor around 50% faster than then N97 while the Nexus One has more than double. Don't get me wrong, I understand that its not just about mhz and that other factors come into play in terms of system responsiveness. But the fact remain that competing platforms are throwing a lot more horsepower at these devices and Symbian is renowned for being sluggish. So maybe its time for Symbian device vendors to ante up as well.
Back in December Raphaël Moll announced the Android Device Dashboard on the Android Developers Blog. The Device Dashboard is awesome because it shows you the % of devices using each release of Android OS. Naturally this is particularly useful when deciding which versions of the Android OS your application should support and whether or not its worth taking advantage of a certain feature etc.
The Device Dashboard draws its information from devices connecting to the Android Marketplace in the preceding two weeks. As at the time of writing the Device Dashboard was showing results for the two weeks preceding 4 January 2010:
As the graph shows you, around 2/3'rds of devices are now using Android 1.6+ which makes 1.6 an ideal target OS level, particularly since it was the first version Android OS to support multiple screen sizes and that it is likely that all devices, even the G1, will eventually be updated to this version of the Android OS.
The graph doesn't give you any specific percentages however so the Device Dashboard also provides you a table which I replicate here:
Android Platform Percent of Devices
Android 1.1 0.3%
Android 1.5 31.0%
Android 1.6 47.6%
Android 2.0 0.7%
Android 2.0.1 20.4%
How great would it be to have this sort of information available for the other platforms?
ald's profileAaron Davidson
Co-founder and CEO of SimWorks - New Zealands leading developer of mobile applications.
SimWorks Anti-Virus protects over 1.5 million mobile phones around the world everyday.
SimWorks phone backup enables mobile operators to protect their subscribers mobile phone contacts and provide their subscribers with unique and compelling new ARPU boosting services.
If you need a mobile app developed give us a call.