Symbian

Several readers pointed out that in my discussion on the market share of modular vs. inter-dependent market shares for smartphones, Nokia was incorrectly classified as having an inter-dependent software architecture since the Symbian platform is/was a modular component.

The problem is that the relationship between Symbian and Nokia is not that of independent modules. Nominally, the two are independent and mutually exclusive, but, in practice, Symbian has always been so heavily dependent and influenced by Nokia that it's never been possible to declare its governance fully independent.

I won't describe the full history since there are many other sources (good example here.) But I will point out that since Nokia licensed Symbian exclusively and Symbian obtained the vast bulk of license fees from Nokia, the relationship was one similar to a bilateral monopoly.

Nokia's bargaining power over Symbian was always high enough to affect its operations. "Ownership" does not only infer a formal control over equity, and many would agree that Nokia "owned" Symbian even when it held less than 50% of the equity. The relationship was so tight, that even though there were other members of the consortium with board seats, the licensing by Nokia's competitors was always very limited. The only significant non-Nokia revenue for Symbian came from Japan where Nokia did not have a presence.

After Symbian was effectively acquired by Nokia, the ownership became both de jure and de facto. The open sourcing of the Symbian code made it no less a Nokia operation than the open sourcing of Android makes it no less a Google operation.

So for all intents and purposes, the relationship between Symbian and Nokia makes Nokia's smartphone hardware inter-dependent on Symbian's software.

Now, given that the way the products are built in practice may be far from integrated, it would seem that Nokia's management captured the worst of both worlds: Modular implementation of an inter-dependent business. Nokia captured all the downside of modularity (poor integration and polish) without any of the benefits (e.g. flexible supply chain or lower R&D.)

ZDNet reports Sony Ericsson are abandoning Symbian for Android, and Samsung headed down the Android and Bada road a while back. There are precious few device manufacturers remaining as foundation members, e.g. ZTE, Sharp and Compal, none of whom are exactly trend-setting industry leaders.

via The Symbian open source experiment has failed [Gartner].

I was going to ask what happened to 'Open always wins?' but decided against it.

What I will say is that open sourcing Symbian was not a new beginning for the platform but the beginning of the end. I don't think anybody seriously considered it a viable multi-vendor platform, least of all Nokia.

Following up on survey data showing that up to 25 percent of Americans have moved to smartphones, here is another survey (Comscore) which shows that US smartphone users are at about 23 percent.

Comscore also surveyed European countries, and we can compare the popularity of smartphones vs. the US.

I also indexed the share to population to show the relative populations of smartphone users across these countries.

It may come as a surprise to some that smartphones are more popular in the UK, Spain and Italy than in the US. Considering that these are countries with lower levels of disposable income, and that in Italy and Spain pre-paid plans are overwhelmingly more popular. Buyers in those countries are much more likely to pay full, unsubsidized prices plus 20% or more VAT. Overall the price differential for an Italian buying a smartphone vs. an American is likely to be a factor of 5. It gets even more peculiar when you consider that many Italians have more than one phone, with overall phone line penetration above 100%.

The explanation for this remarkable appetite for smartphone goodness is the early lead that Symbian had in Europe. Buyers who entered the phone market ten years ago became accustomed to upgrading their Nokia phones. Symbian phones were aspirationally positioned as feature-rich camera and messaging devices. Many were also purchased without data plans and were thus used as high-end feature phones. In other words, consumers in those countries were comfortable paying full price for unlocked Nokia devices and using them with multiple SIM cards.

This can be seen in the share of Symbian in these charts:

The data points to how normative behavior evolved and how different that can be even among culturally aligned Western nations. When looking at Asia and South America, it gets even more interesting.

Whereas in Europe it was Symbian, in the US RIM got the ball rolling.

The contribution of these platforms in shaping expectations, not to mention pricing, for the iPhone and Android should be noted.

Market share is an existential threat to Symbian, it imperils the very existence of the platform. And the main reason Symbian is losing share is the user experience which isn’t competitive with Apple or Android. Based on the early previews I’ve seen Symbian 3 looks to have polished a few of the rough edges, but doesn’t fix the problem. So if the weak UI is threatening Symbian’s very survival the Foundation ought to be seriously worried, right? Wrong. I just looked on the Foundation web site and blogs at the roadmap and features for future releases. What I see is too much effort on stuff that really doesn’t matter. For example: Audio policy packages for Symbian, WIFi direct, support for an “open cloud manifesto”, an accredited Symbian developer program for China, better multitasking, multiple personalised home screens, HDMI connection to external TVs, better web runtime support, better internal architecture and so on.

Forget elegant architecture, forget better multitasking, forget Chinese developers, forget release schedules that don’t deliver S4 devices with a new user experience until 2011. None of these matter. People will never use the features if they don’t buy the phone. The situation is now serious enough that any developer who isn’t working on something directly related to a new UI is wasting their time. The S4 UI is a “bet the platform” project. For any organisation to be in a situation where its survival depends on one project is very dangerous, especially when their track record in the area isn’t outstanding. I think the Foundation needs a contingency plan in case the planned S4 interface isn’t radical enough or good enough. Maybe redirect some developers and start a couple of skunkworks projects to create new competing UIs for S4, or perhaps announce a competition with a $1M prize for a new Symbian UI to encourage some radical ideas.

I think the Symbian foundation is just re-arranging the deck chairs on the Titanic and ignoring the Android iceberg ahead.

In June 2008, Nokia made its first big move to turn around the platform, and announced that it was acquiring Symbian, with the intent of turning the OS into an open source project.

Two years later, the move to open source has proved to be a miscalculation that is slowing down Symbian's development. It would be better for Nokia to take full control of the OS, according to Wood. A lack of support from other vendors means Nokia has to do most of the work itself, while the open nature of the platform allows competitors to keep a close eye on its progress.

via Nokia on long comeback trail after smartphone misses - Digital Lifestyle - Macworld UK.

Then there's LiMo foundation open source mobile Linux. Maemo is/was open source, Openmoko and Qt Extended and PalmSource/Access moving to open source and there was the Motorola Linux OS that launched years ago. If Open always wins, whatever it wins, it's not market share.

The number of Android devices is rising steadily; it's already up to 135.  Android devotees should rightly rejoice.  However, Android is not the first mobile platform with an open licensing strategy. A quick visit to pdadb.net lets us count the number of devices that shipped for every mobile platform in history.  We can also see the current market shares as listed by Gartner for these platforms.

The numbers of SKUs (stock keeping units) that have shipped historically vs. the market shares of the mobile phones running those platforms are (see Footnote below for some caveats):

The same data in a scatter plot:

(SKUs on the x-axis and share on the y-axis)

The thing that jumps out is how uncorrelated the two quantities are.  The Windows Mobile/CE number is such an outlier that it breaks any attempt to link the two variables.

What's likely to happen is that since the licensing of Android is even less problematic than it was for WinCE (i.e. it's both free as in beer and free as in speech) the number of Android SKUs is likely to overtake WinCE.  In other words, we're very likely to see Android grow from hundreds to thousands SKUs.

However, the real question for the Android platform isn't whether it will have licensees. WinCE proved that there is vast demand from many companies to build cloned and undifferentiated devices. The real challenge is whether Android can achieve significant volumes without fragmentation to keep the platform/network effect going.

The WinCE platform was actually less fragmented through its life than Android is today. This is understandable since the source code was decidedly not open and thus not modifiable and new versions were released far less frequently. And even with a high price tag, it was wildly popular with licensees. Nevertheless, the volumes never came and Microsoft was forced to abandoned the whole strategy.  The next version of Windows Mobile (Windows Phone 7) will follow an "iPhone lite" model of highly restricted configurations and device and ecosystem specification.

So will Android follow the fate of Windows Mobile with an even more fragmented, bazaar-like licensing model? Or will thousands of undifferentiated Android devices swamp their integrated competitors?

Footnote: The total SKU number includes non-cellular devices whereas the market share number includes only cellular devices. Omitting non-cellular devices will not detract from the pattern or change the correlation meaningfully.  The argument can also be made that the "platform" value lies in both cellular and non-cellular devices (e.g. iPad) and they should both be counted.  Unfortunately there are no market share statistics that blend both device types into one market.

MM Research does not count Symbian as a smartphone platform. This makes them inconsistent with any other analyst for counting smartphones.  So shouldn't Symbian be included?

In a comment to iPhone has 72% of Japanese smartphone market | Asymco it's been pointed out that 12 million Symbian sold in the same time frame as Apple sold 1.7 million phones in Japan.

It would seem then that the correct market share for iPhone would be 12%, with Symbian having 83% and "others" having less than 5%.

However...

Symbian in Japan is not the same thing as Symbian elsewhere.  Symbian in Japan is used as a low level OS by Fujitsu, Sony Ericsson Japan, Mitsubishi, Sharp and others to provide devices running the MOAP(Symbian) software platform.  MOAP (Mobile Oriented Applications Platform) is the software platform for NTT DoCoMo's FOMA (Freedom of Mobile Multimedia Access) service.

Unlike Series 60 and UIQ MOAP(Symbian) is not a open development platform.

MOAP is also supported by Linux with Panasonic and NEC using it in something called MOAP(Linux).

MOAP(Linux) is also not an open development platform.

So the "72% share for iPhone" in Japan must be stated with this important caveat: that Symbian and Linux are not included because, due to not having exposed APIs, they are classified as feature phone platforms.

Following up on my last post on how misleading US-only share comparisons can be, I decided to draw charts to visualize the comparison.

As Android and iPhone compete in various ways, it's hard to see which is the preferred choice given a direct comparison.  In other words, iPhone and Android devices rarely are placed next to each other with similar terms.

Take the US market for example.  The overall data from NPD suggests that last quarter Android reached 28% share vs. 21% for iPhone.  Many of those Android devices were new to market or at least newer than the iPhone which in Q1 was coming to the end of its product cycle.  Second, pricing for Android devices seems to have been quite aggressive with buy one get one free sales.  But I won't dwell on tactics now; what I do want to note are the differences in share between AT&T users and non-AT&T.

Note that within AT&T, iPhone outsells Android over 4 to 1.  iPhone also outsells "others" (mainly RIM) more than 3 to 1.  However, outside AT&T, where the iPhone is not available, Android does not outsell "others".

If we exclude the US altogether, we also see that Android does not have a great distribution.

Outside the US, the iPhone also outsells Android nearly 4 to 1, but it has a way to go before challenging Symbian which makes up the bulk of "Others".

So in markets where Android is head-to-head with the iPhone (AT&T and non-US markets), iPhone's lead is quite high (still).  The possibility still exists that Android will overtake iPhone given the broad licensing and distribution, but it's not necessarily a given.  And in any case, iPhone is not the market share leader today and that leadership does not seem to be their objective (note the pricing).

The bigger question is what will happen to RIM and Symbian as Android grows.

Smartphone impressions triple in a year « Asymco

Following up on the previous Admob numbers, it seems that Symbian impressions grew by 22% vs. overall growth of 43%.

In absolute terms, total impressions a year ago for Symbian reached about 1 billion vs. 1.22b in February.

Symbian's share of requests therefore fell from 43% in February 2009 to 18% in February 2010.

When thinking about the number of devices shipping out of Apple, and the relative value of those units compared to the competition you have to always think of the platform.

The market leader Nokia claimed to have sold 16+ million smartphones in the quarter. When comparing platforms, to whatever Apple ships in iPhones you have to add the iPod touch units. I see that number being about half of the iPhone numbers or about 4 million. Let's say 10 to 12 million as a range for the platform. Already this is within striking range of Nokia.

It may sound that Apple has some catching up to do, however the important thing is that most of the Nokia devices are not uniformly addressable by developers because they are different platforms and the Symbian platform itself will be broken next year as it has been broken many times before. This is also true for Blackberry and Windows Mobile and will become true for Android as vendors fork and splinter the code to differentiate.

This already puts Apple in the pole position today in terms of contiguous addressable units volumes.

I cannot over-emphasize the importance of this platform effect. It's what made Windows dominant and it should be the most important issue in the planning of new mobile products, but it clearly isn't for Apple's competitors. Either product planners are ignorant of history or completely hamstrung by other constraints on their businesses (I expect the latter).

As a result it's inevitable that Apple will have a dominant platform. The numbers in apps and consumption of apps already are telling this story, and devs are voting with their code in a landslide. But going into 2010, this will become evident in the units and Apple share numbers will accelerate toward lead position.

The bogie therefore is to look for contiguously addressable installed base. On this basis of competition, I expect at least 100 million for Apple at the end of next year and less than 10 million for any competitor. At that point the game will be officially over.