According to my web logs, my post yesterday on Nokia MeeGo and Belle was very popular. Some people agreed with me on Twitter. However, a tweet by Sebastian Brannstrom made me look at it from another angle. He said "I think an important factor is the staggering cost of Nokia R&D. They can’t develop at reasonable cost".
This is true. According to Bernstein Research, Nokia spent $3.9 billion, almost three times the average of its rivals’, in 2010. About a third of this was on Symbian.
While there were, no doubt, savings to be made in company inefficiencies, a lot of blame has been centred on Symbian. When I worked at Symbian in London in 2006/7, I was amazed that the people there were hit by the same problems as 3rd party developers. I had somehow thought that the difficult APIs and ideoms might have been mastered by the people creating the OS. Instead, a mindset persisted that things were difficult because they needed to be, to provide for an efficient OS. Things took much longer to do than, for example, on Windows Mobile on which I also had deep experience.
As I mentioned as long ago as 2005, and again more recently, all Symbian needed was something more friendly (e.g. dev libraries and UI) on top of Symbian. In fact, this is what Windows Phone became – a new UI and APIs on top of Windows CE. Nokia could have done the same thing (but much better with native rather than managed code) on top of Symbian. Nokia tried something similar with Qt but had problems agreeing a new UI and Qt mobility (device APIs) arrived too little, too late. Adapting the Qt dev environment acquired from Trolltech proved too difficult. A simpler native c++ dev library that both internal and 3rd party developers could use, and new Belle-style UI might have been all that was required.