In a recent report ARCchart postulated that Nokia was going to move away from Symbian and use Linux as its underlying OS for the Series 60 platform. The report has been picked up by others and there has been a somewhat suprising amount of speculation about the subject. Such speculation is erroneous and shows a lack of understanding of the Symbian - Nokia relationship and the position of the industry.
ARCchart do allow that the porting process would be possible if technically not an easy feat. This rather understates the difficulty involved. The strength of Symbian is and always has been the fact it has been designed as a mobile OS from the beginning of its life. From release 6 onwards it has been designed with mobile telephony at the heart of the OS. As a result the Symbian OS is structured is some fundamentally different ways to other OS's. Power and performance management are key considerations in design from the kernel upwards. As a result the Symbian OS is the most powerful mobile OS available. It would require fundamental changes in Linux's core to achieve similar specifications. Furthermore the higher layers of the Symbian OS would need to be recreated, one example of this is the security model. This is far from trivial. It is a common misconception to under estimate the amount of work that the Symbian OS does. In particular release 8 and 9 with its single chip architecture, built in security model (more important than most people ever realise), and multiple architecture and standard support are some way beyond the nearest competitor.
The investment that Nokia has made in Symbian OS, both in terms of finance but also human and knowledge resources should not be ignored. The large majority of Nokia software developers are working on Symbian and Series 60 related projects. Indeed Nokia has made clear its desire to become a software company as well as a hardware company.
While the license fees that Nokia pay are large it is important to realise that they are paying for something that they would otherwise have to fund themselves. The question that would need to be asked is how much would it cost Nokia to do the same thing internally? Furthermore as volumes expand it is very likely that the license fee will be reduced. While Symbian has yet to make a profit the subject has been moot, but once it does so the shareholders of Symbian are very likely to move to reduce the license fee for larger volumes. The first sign of this i the fact that the fee is reduced to $5 from $7.5 after the first 2 million units for each licensee.
The Bill of Materials for an average smartphone is still some way beyond $150 (and considerable more for the higher end models). With this in mind it seems unlikely that Nokia would look to cut the $5 licensing fee for software when there are undoubtedly are cost savings to be explored. Even when the Bill of Materials reaches $100 (and this is not expected to happen until 2009) the software will still (even assuming there is no licensee fee reduction) only represent 5% of the total cost. This could be compared very favourably to other industries.
ARCchart point to the Nokia 770 as an example of Nokia exploring Linux options. This is perfectly true, but it important to understand that Nokia have been doing this for a long time. The 770 project itself goes back to at least 2002.
Part of the lesser known history of smartphones is that about 5 years ago Nokia was making a choice between several platforms as to what its future platform would be. At the time Linux, Palm and Symbian were all evaluated. It is also important to note that the device does not include any cellular connectivity which is at the heart of the Symbian OS. Nokia were stating an obvious truth that they considered Linux to be the most suitable platform for a wireless tablet device.
ARCchart also point to the separate licensing of ActiveSync and the Windows Media DRM. This is to misunderstand how the licensing arrangement works. Symbian have licensed the technology so that they can create a 'plug-in' that all their licensees can use to allow their phones to inter operate with an Exchange server. The license only covers use by the Symbian licensees. In order to do their own work with ActiveSync (principally with Series 60) Nokia were obliged to license ActiveSync themselves regardless of whether Symbian had a license or not. This situation has arisen because Symbian and Series 60 are separate, and are owned by different companies. This might be contrasted with UIQ which as a owned subsidiary of Symbian does not need to individually license ActiveSync. Furthermore it is worth noting that Symbian did not license the Windows Media DRM (Play for Sure) - only Nokia have licensed this.
A major point of the ARCchart report was that because most Series 60 applications are written in Java and consequentially the pain involved in porting would be small. This point is simply untrue. The majority of program are written in C++, and, perhaps more importantly, an even larger majority of the best selling programs are written in C++.
Symbian figures shown at various press conferences and presentations show that there are at least twice as many C++ applications as there are MIDP 2 designed for Series 60 (2,000 vs 1,000 at the beginning of the year). One of the problems commonly identified when counting MIDP 1 and 2 programs is that claims of 'write once, run anywhere' do not exist in reality - instead Midlets must be tweaked for each phone. There are many 10,000's of MIDP programs written in many variants for different phones, but relatively few target Series 60 specifically and relatively few can therefore be regarded as Series 60 compliant. With other languages also considered such as Crossfire (AppForge), and OPL the difference is even more marked.
Aside from a simple numbers issues this ignores the fact that there is a stark contrast in the richness and complexity of the applications. The very nature of MIDP (it runs in a strictly controlled environment (sandbox) within the phone) the more powerful features of the phones are often inaccessible, and performance can be an issue. Although JSR's do allow access to more features of the phones (such as Bluetooth) than the MIDP core functionality these do not cover everything, are not available on all phone and are sometime poorly implemented.
This is perhaps best illustrated by looking at the Top Ten Best selling applications for the platform (as shown on Handango) - 9 out of 10 of the best selling programs are written in C++. In the top 50 there are only 4 MIDP programs. Given the top selling programs are the most important by revenue volume the dominance that C++ applications enjoy on Series 60 is clear.
The large majority of MIDP programs fall into two categories - games and enterprise specific programs. Games are often ported from other phones to Series 60 (Gameloft is a good example of a company that does this). However all the more complex games tend to be written in C++. For example Gameloft produce several Series 60 only (C++) games and they are their best sellers. Developers often comment that to move beyond basic graphics and sound programming in C++ is a necessity.
Enterprises will often use MIDP programs to interface with back end systems (mainly for data reporting and retrieval - a relatively simple function). MIDP is regarded as a good choice because despite some issues it is relatively device agnostic which is very important in a multi device Enterprise environment. Even then the most powerful Enterprise programs are often written in C++.
The claim that Java programs are indistinguishable from native programs is also dubious. While this is true at first glance - any experienced user will tell you that the differences are apparent in use of UI components, performance and installation reliability.
Clearly then applications and the investment by the ecosystem is this area by both Nokia and 3rd parties is probably one if the single biggest reasons a switch is very unlikely.
Speculation about the relationship between Nokia and Symbian is inevitable. However the chance of Nokia dropping Symbian as the underlying OS for Series 60 is non existent. To think otherwise is to ignore the reality of the costs and the industry and the large investment Nokia has made in Symbian, both in terms of financial, human and knowledge resources. It ignores the correspondingly large investment that would be required to move away and successfully pursue a Linux strategy. Such a strategy would be a massive risk, not only because it would threaten Nokia's market position, but also because there is no guarantee it would be successful. The Symbian strategy has already been very successful for Nokia (some 70%+ smartphone market share) and is likely to remain so for the foreseeable future.