Who wants an open operating system on a mobile phone?

The manufacturers? The telcos? The users?
*The manufacturers: yes to cover their bets, but not really to sell it; an open OS reduces their differentiation, which makes them more vulnerable to low-cost manufacturers.
In addition, an open OS
- increases the cost of hardware for the same service,
- allows the user to add third-party applications, which makes technical support almost impossible
- forces manufacturers to rewrite their legacy applications
The Linux case is slightly different: manufacturers need an OS in any case, so it might as well be Linux, for which there is a momentum right now. Because it is open source, they can have complete control over the software they develop with it. In addition to the hype effect that comes with Linux, that is probably the reason why almost all manufacturers have an internal project based on Linux. That doesn't mean the device itself will be sold - if it is ever sold - as an open platform.
In this regard, the Mobilinux initiative, launched by embedded Linux specialist Montavista, is interesting. Mobilinux is a consortium whose objective is to define reference platforms to facilitate manufacturers' job when they look for a complete software platform. It is a response to the issue that very often, Linux is only a starting point, not a complete environment. Such an initiative should also help integrate legacy applications. While technically appealing, however, the risk with the Mobilinux approach is that it tend to reduce the differentiation of each manufacturer, as is the case with Symbian or Microsoft Windows. As a result, one would expect that Mobilinux would be more successful with emerging asian manufacturers than with established players.

*The telcos: Again, they might want to cover their bets and leverage the marketing of some powerful open OS vendors, but is it really in their interest? An open OS decreases the switching cost for the users, reduces the possibility to create exclusive applications (e.g. Instant messaging) as this would violate the "rule" that the OS should be the same on all platforms. In addition, the integration of such functions is not very good, which means the user experience is poor, which in turn limits the usage.
- As in the case of the manufacturers, the addition of third party applications will make technical support difficult and expensive.
- The cost of the devices will reduce the rate of renewal, generating frustration among users as the obsolescence rate will not diminish. In turn, ARPU will be hurt.
Telcos have been involved in defining industry standards through the OMTP initiative. The evolution of the specs will soon give a good indication on what telcos have in mind with regards to open OS...

*The users: having a universal device modeled on the computer is appealing, but are the users ready to pay the price? Such a device is inherently multi-purpose and is not optimized for telephony anymore. As anybody who has played with a smartphone has noticed, each function becomes more difficult to use, and key features are gradually buried under several layers of menus. The increasingly large number of functions reduces the quality of the user experience.
- the devices are more heavy and more power hungry, which reduces the autonomy
- the cost per device is higher and more difficult to recoup for user
- the risk for bugs increases probably to the square of the number of features offered.
- the responsibility of the telco is diluted as the user doesn't know where the buck stops anymore: at the telco, at the OS vendor or at the device manufacturer? Who is accountable for the user experience? Who gets the brand effect?

As long as the mobile telephony remains a closed environment, it seems it can only run on a closed platform. If it chooses to become open, much like the PC where the provision of the device and the Internet access are done by two separate players, then an open platform would probably make sense and operators could loose their dominance. To say the least, that is not the way the industry seems to be heading now.