24 June 2008

Welcome to the future of Symbian

If you’re reading this then the chances are you already know that today marks the most radical transformation of Symbian since we started the company 10 years ago. It’s going to take a while before all the implications of the announcement become clear, and things will still be in the planning stage for several months ahead, but I thought I would make a start on answering some of the biggest questions now. Here goes...

What’s happening?

Let’s start with the technical details. First, subject to approval by the relevant competition authorities and various other conditions being fulfilled, Nokia will acquire the shares in Symbian that are currently held by Sony Ericsson, Ericsson, Siemens, Samsung and others.


Secondly, Nokia’s hope and intention, described here, is that this transaction will pave the way for the next step: the creation of a new unified platform, consisting of Symbian OS and S60 along with technology from UIQ and MOAP(S). Sony Ericsson, Motorola, NTT DoCoMo and a whole army of others are collaborating in this venture, contributing money, time and technology to help make it happen.


To that end, there will be a new not-for-profit entity to host and guide the platform: The Symbian Foundation. Anyone who agrees with the basic principles and by-laws of the Foundation will be able to join (you can register your details for more information on membership here), and the source code to all the components that the founding members have contributed will be available to Foundation Members from Day 1 (which will be a short while after the deal is closed). In due course, the platform source code will be made available to anyone who wants it under the Eclipse Public License. And the platform will be royalty-free.


At the time of writing, the aim is to get the Symbian Foundation operational during the first half of next year (there are a few legal hoops to get through first) and while there is still a mountain of stuff about how the Foundation works to be planned, it is clear that all the parties share a strong agreement on a few core principles: it should be open, it should be transparent and it should be meritocratic. Contributing money or people doesn’t buy you a right to dictate what happens with the platform: if you want to influence its evolution then you just need to be smart, committed and active at the project level. I’ll say more about what these principles will mean in practice in the future.


So much for the facts...they still leave the question: Why?

To help answer this I need to share two insights about the future of mobile computing that have been prominent in the minds of those who work within Symbian and our shareholders over the past year or so.


First, the community will be a vital source of innovation. There’s nothing surprising about this to anyone who works in software. But the scale and the diversity of the community developing for mobile - in terms of geography, knowledge and target audiences - is likely to dwarf anything that’s been seen before.


Second, the economics of the handset market - where the cost and complexity of making a phone has been rising dramatically - beg for some kind of radical change in the efficiency of production. It’s just too damn expensive to make phone software. And incidentally - for reasons I’ll go into in a later post - this is currently true whatever platform you use and whether it carries a royalty or not.


These observations suggest a few conclusions. Namely, that the most successful platform will be one that has the broadest reach geographically, the broadest reach in terms of contributors and the broadest reach in terms of environments. This is why the Foundation members even at the outset cover so many regions and parts of the value chain. This is also why we need to create a new organisation with clear principles and rules - because to build the broadest cross-industry community possible the platform needs to be properly neutral. (I’m sure there’ll be some scepticism about that in some quarters but we’re happy to be judged by our actions.) And finally this is why the vision includes platform support for runtimes as well as the familiar Symbian OS and S60 environments - it just won’t be adequate for a platform to restrict people to C++, or just Java, or just whatever: we need to cater for all.


The other conclusion we’re led to is that the most successful platform from an economic perspective will most likely be one whose heart is mobile. That is, one whose raison d’etre, from the kernel upwards, is to incorporate features and modifications that are needed for mobile computing devices.


If you take these points together I think you can begin to see how we’ve arrived at what’s happening today.


What does this mean for Symbian’s developers and partners?

There is of course an army of individual developers and commercial organisations who create products to work alongside Symbian OS, who rely on us and have a huge interest in what happens to the platform. The most important point to make to these people is that it’s ‘business as usual’. Whatever you rely on Symbian for today - DevKit access, support, SDN and SDN++, Symbian Signed - is going to continue without missing a beat in the near and medium term. To be clear, we’re not even allowed to make any operational changes prior to the transaction receiving approval.


In the longer term clearly many of these activities will naturally migrate to the Symbian Foundation. More than that, we’d hope that the same developers and partners will join the Foundation, participate in the discussions about its evolution and contribute.


What does this mean for Symbian’s roadmap?

Absolutely the same principle applies here. The core technologies that are being introduced in current and future releases of Symbian OS - some of which we have talked about, like Freeway, Screenplay and SMP - will continue with the same target delivery dates. Making sure these stay on track is critical.


What will change in due course is that as we move over to the Foundation model - following the completion of the transaction - we will shift to unified platform releases, and that will mean we move over to a roadmap which looks somewhat different, since it will include everything from the OS up to the application suite. As I said before, we intend to be as transparent and open as we can about this stuff so expect more details as they become clear.


Why the Eclipse Public License?

This is the kind of question that has some people settling down for a debate and others looking for the exit, and which could in any case run to several pages. I’ll keep to the two clear pragmatic benefits which pointed towards the EPL for now. One, it’s ‘weak copyleft’, which means that it gives commercial entities some confidence that the decision about when a given feature is a ‘differentiator’ and when it is something that belongs in the core platform is something that they have a degree of authority over. I regard that as important because competition drives innovation too. Two, the EPL is familiar to everyone involved in the Foundation - everyone has already been through the process of adopting the Eclipse license in some context. This really helps to accelerate the whole process.


Why not open source on Day 1?

Being able to do so would be great, but we want to make sure of two things. One, that all the contributors have time to go through the necessary steps before contributing their code to the Foundation, because they will in effect be granting patent licenses to their competitors in doing so, and on the whole if you’re granting patent licenses to your competitors it’s best to check with your legal department.


And two, that what we open up makes sense and is really usable - i.e. that it is consistently commented, well-documented and so on. And of course that we have the rights to open source it: today some of the code in Symbian OS and the other platform components is licensed in from third parties whose rights we will obviously take care to respect. These processes are going to take a little while, but we will try to work out a way of doing it in phases so that really useful bits of source get prioritised and released earlier. It’s a big task though - there are open source projects with a larger code base, but I’m not sure that any body of code this size has been open sourced - if you can point me to examples otherwise then please do because we’re eager to learn.


What happens next?

The first thing that happens next is most likely that a large number of Symbian, Nokia, Sony Ericsson etc. employees go to their respective bars/pubs/whatevers around the world to absorb the news and probably other substances besides. For most people inside the companies as well as out this will come as a surprise, so there will be a lot of work in the next few days answering questions about what the future looks like and what it means for everyone. But once the first dust cloud has settled we will get on with planning the details of how the new platform will be developed, how the Foundation itself will work in practice once it becomes operational and what the transition process should look like.


Which means there are still a few thousand questions to be answered, but it’s going to be fun tackling them. To that end, I’ll post back here as we nail more things down and are able to fill in some of the blanks. If you feel like adding to the list, please post a comment here or mail me and I’ll do my best.

6 comments:

Tonethefone said...

Unless my memory serves me badly, I believe that IBM made some of its products open source and being IBM those products must have been millions of lines of code.

Anonymous said...

John, that's the best explanation from a development point of view that I have read. Making Symbian open source is a great idea. I still wonder about the application singing process though. From a security point of view this is essential but how does that equate to the cost of development. So I hire some devs we build a great new app, it still costs a significant amount of money to get that approved to run on a symbian device. Do you see a way of this changing without increasing the chance of someone writing an app that runs in the background and calls an international sex line in Hong Kong..lol

best Rob Knight - GoMoNews.com

John Forsyth said...

Rob, I appreciate your concerns: of everything we currently do in Symbian, application signing has got to be the activity that most firmly belongs in the 'damned if you do, damned if you don't bucket'.

I generally assume that, as you say, some form of application signing is probably necessary, and secondly that it's better done by a central and relatively neutral body - rather than, say, network operators. (I have had no shortage of debate with operators on this point...) All of which suggests the Foundation will need to provide this service.

Which raises the issue of cost. The current costs associated with signing have come down a lot - for certain types of application it can be as low as $20. (Ewan Spence has a good summary of the last redesign of the signing program here:

http://www.allaboutsymbian.com/news/item/6092_Symbian_Signed_Announce_New_20.php

). And there are channels for freeware developers to distribute their apps without obtaining a publisher ID (which also costs).

But I admit we still need to think from a global perspective: $20 is a rather different figure depending on whether you are a developer in the US or Europe vs being a developer in, say, the Philippines, where per capita income is about 1/30th of that in the US.

In short - although the cost has come down some way, we probably do need to look at what more can be done here.

John Forsyth said...

Hi Tonethefone...IBM has certainly had a lot of involvement in open source projects and has run its own initiatives. I figured the largest by lines of code was probably Eclipse - currently this is around 10M lines of code, which is enormous, but was a lot smaller when it was open sourced. Either way the code base that we're looking at here is significantly larger than 10M(!)

Anonymous said...

Thanks Jon, I take that all on board, it certainly is much cheaper than it was!!! and I totally back the open source stance, but I was looking at it from a business point of view.

Yes its not a lot per device even with the Publisher ID but put 50 devices in the release schedule and it starts to add up especially for the single developer application which undoubtedly the open source stance will now attract. But like you say you are damned if you do and damned if you don't. There is no way an operator is going to allow an unsigned application to run on its network which is a stance no one can hardly blame them for. It's ironic don't you think that only last week Dr Ari Jaaksi made his comments on open source business models!!

thanks for taking the time to reply.

best

Rob GoMoNews

Anonymous said...

Should have known you were involved in all this John ;-)