Sunday, 14 November 2010

Is Symbian dead? And if so, who killed it?

"We should declare victory and go home."
--Apocryphal quote attributed to George David Aiken

I hesitate to write anything about Symbian, because it's a great way to get branded a parochial American, or an Apple fanboi, or a "member of the US-protectionistic mobs braying for blood," to paraphrase a comment from a tech discussion forum in the UK this month.

But there's been a huge cloud of smoke and very little light in the recent online discussions of the changes at Symbian. Is Symbian dead? Is it stronger than ever? What's really going on? I wanted to see if I could make sense of the announcements. Besides, there are some important lessons from the Symbian experience, and I'd like to call those out.

Here's my take on what's happened: The business entity called Symbian was originally designed to prevent Microsoft from controlling the mobile OS standard, without having Symbian itself seize control over the mobile phone companies that funded it. In that task it succeeded. However, as a company run by a consortium, Symbian's governance was politicized and inefficient. This left Symbian woefully unequipped to compete with Apple and Google. A different approach was needed, and Nokia's new management has finally come to terms with that. As a result, Symbian as an organization is now defunct, and Symbian as an OS is becoming background infrastructure that has little relevance to the mobile platform wars.

To explain why I reached that conclusion, I have to start with a quick refresher on Symbian's history, for readers who haven't been following it closely...

There are two things named Symbian: Symbian the company and Symbian the OS. Some of the confusion this month was caused by people mixing up the two things. Symbian OS began as EPOC, the operating system used in Psion's handheld devices. EPOC was spun out of Psion in 1998 as a separate company called Symbian, co-owned by Psion and most of the leading mobile phone companies of the day, led by Nokia. The idea was that all of them would use the renamed Symbian OS in their smartphones, enabling them to put up a unified front against Microsoft, which they feared would rule the smartphone market.

Over time Nokia came to be the dominant manufacturer of Symbian OS phones outside of Japan, largely (in my opinion) because the Symbian phones made by other mobile phone companies didn't sell well. Eventually the other mobile phone companies no longer wanted to pay for a joint venture that was mostly just supplying software to Nokia. Linux was gaining momentum as a free, open source mobile OS, so the Symbian partners, led by Nokia, decided in 2008 to convert Symbian OS into an open source project. Nokia hired most of the Symbian engineers, and gave away their code through the foundation.

Symbian the company was replaced by the Symbian Foundation, a nonprofit tasked with managing the open source process and encouraging other companies to sign up to use the software. The idea was that Nokia, the other Symbian licensees, and a growing hoard of academics and developers would work on various parts of the OS, contributing back their modified code to the shared base. The move to open source kept some level of engagement from several other mobile phone companies, most notably Samsung and SonyEricsson.

But both companies continued to have poor sales for their Symbian phones, and this fall they announced that they had no further plans to use the OS. That left DoCoMo in Japan as the only other major user of Symbian. Nokia was stuck with an open source foundation that mostly just supplied its own software back to it. That wasn't going to be viable. So earlier this month, Nokia and Symbian announced three significant changes:

--The Symbian Foundation is being dramatically scaled back to "a legal entity responsible for licensing software and other intellectual property, such as the Symbian trademark." (link). In other words, it's just a shell. Symbian is now truly Nokia's OS. Nokia will plan, develop, and manage the Symbian code base, and distribute it directly to anyone who still wants it (presumably DoCoMo). You can read a biting commentary on the changes here.

--At the same time, Nokia reaffirmed an announcement it made in October that it is focusing all of its application development support on the Qt software layer that it purchased several years ago (link). Qt will now apparently be Nokia's one and only application layer, deployed on both Symbian and the upcoming MeeGo OS being codeveloped with Intel (link).

--The EU is putting 11 million Euros into a new organization, called Symbeose (which stands for "Symbian – the Embedded Operating System for Europe"), which will help fund the development of advanced Symbian OS features, including asymmetric multiprocessing, dev tools, memory management, image processing, video acceleration, speech to text, mobile payment, multimedia formats, and embedded systems beyond mobile. There are two semi-conflicting explanations of what Symbeose is all about. Some people say it's aimed at turning Symbian into an embedded OS that can run in all sorts of devices (why Europe needs that instead of Linux is unclear to me, but you can hear some discussion of the wrongheaded North American mobile paradigm here). Others say the intent is to resurrect Symbian OS as a smartphone OS used by companies other than Nokia. In a presentation, Symbian Foundation said the investment is intended to "combat mobile device and service homogeneity exemplified by Android and iOS" (link). Apparently taxpayer support is needed because Nokia isn't willing to pay for some infrastructure needed by other phone companies (link). A Symbian Foundation employee explained: "I would say that the main focus of the developments will be advancing existing, as well as building new tools and services relevant for smartphone manufacturing at the beginning of the manufacturing process. We want to make it easier for any manufacturer to take the Symbian codebase and develop new smartphones" (link).

What it means

Symbian isn't dead. It's just irrelevant. After the announcement, Nokia professed its strong support for Symbian OS (link). Nokia has no choice but to support the OS because it's built into the whole middle to top end of the Nokia product line. Given all of the legacy Nokia code written in Symbian OS, the Symbian-based phones still in development, and all of the Nokia development teams who are used to working in Symbian, it would probably take years to flush all of the Symbian code out of Nokia's products even if it wanted to. Symbian at Nokia is kind of like Cobol at IBM -- you're going to go on tasting that particular meal for a long time to come.

But the decision to focus on Qt for applications means that Symbian OS is effectively no longer an app development platform. It's embedded software; the background plumbing that powers Nokia's smartphones (and maybe other embedded systems, if the EU has its way). There's nothing wrong with that, but it makes Symbian irrelevant to most of the folks who talk about mobile technologies online. We don't spend much time online debating which OS kernel a device should use, and that's now the world Symbian lives in. The real competition for developer and smartphone user loyalty in most of the world is now Qt vs. iOS, Android, and RIM. Plus that Windows thing.

What it means for Nokia: Hope. Nokia's app recruitment efforts have been hamstrung for years by what I think was an incoherent software platform story. What should developers write their software on? Symbian native, S60, Silverlight, Qt, Adobe Air, one time or another Nokia romanced just about every mobile platform on the market. Nokia said that was a strength, but actually it was a sign of indecision and internal conflict. Developers crave predictability; they want to know that the platform they choose today will still be supported five years from now. By flitting from platform to platform like a butterfly, Nokia sent the unintentional signal that developing for it was dangerous.

Many developers did support Nokia anyway, especially in places where the Nokia brand and market share were so dominant that the decision was a no-brainer. But I think their loyalty did a disservice to Nokia in some ways, because it blinded the company to the shortcomings in its developer proposition. When Nokia had trouble recruiting developers in places like Silicon Valley, it seemed to think they were just biased against it. Time and again, I attended Nokia developer events in California where Nokia concentrated on telling people how big its installed base was, and showing off its latest hero device (N97, anyone?). I can see Nokia's logic -- after all, developers in Europe seemed happy. But the reality was that developers in Europe had given it the benefit of the doubt, despite its poor overall proposition.

So the decision to focus on Qt (pronounced "cute," get used to it) is a positive one, in my opinion. This is one of those cases where making any decision is better than the status quo. Qt isn't perfect, but if all of Nokia aligns behind it, any problems in it can be ironed out.

Unfortunately for Nokia, this is just the beginning of the changes it needs to make, rather than the end. Nokia's Qt development tools still reportedly need work (link). And app developers don't just need a coherent technical story, they also need a coherent business story. How do they make money? Although Nokia sells a huge number of Symbian-based smartphones, most of their users seem blissfully unaware that they can add applications. That's why Nokia has a much smaller base of applications than iPhone, even though its customer base is far larger.

To attract more developers, Nokia will need to do a lot of marketing, both in advertising and on the device, to make sure Qt users know they can get apps, and are stimulated to try them out. Nokia has the resources to do this, but once again it'll need consistent and well coordinated execution to make it happen, something that the company has failed to deliver in the past. (For example, spamming people with SMS messages telling them to try other features is probably not the right approach (link).)

To give you an idea of how much ground Nokia needs to make up, Apple iOS has 60 million users and 225,000 applications, a ratio of about 3.75 applications per thousand users. Android is close behind, with 3.5 apps per thousand users. In contrast, Symbian has 390 million users and 7,000 native apps, a ratio of about .02 apps per thousand users. (link). Yes, I know, there are additional Nokia apps written in Java, but that kind of proves the point that Symbian is plumbing rather than a platform.

All of these changes need to be carried out against a backdrop of cost cutting, as Nokia brings its expenses in line with its revenues. One of these days when I get the time I'll write more about Nokia's overall situation, but for now suffice it to say that Nokia is working off the after-effects of several years of growing expenses while revenue was stagnant. Nokia's circumstances aren't quite as bad as the California state budget (if you are in Europe, think Greece), but it's ugly enough to distract from all of the other things the company needs to fix.

What it means for developers: Wait. First, the bad news: The switch to Qt means that current Symbian OS developers who aren't already using Qt will need to rewrite their applications. This is the latest in a series of rewrites that Nokia and Symbian have forced on developers over the years. If they had more developers it probably would be causing a big ruckus right now. The fact that you don't hear a lot of screaming speaks volumes.

The good news is that Nokia may be getting its act together for developers at last. But if I were working on a mobile application today...wait a minute, I am working on a mobile application today. So here's what I'm doing about Nokia: I'm waiting. If Nokia creates a great business proposition for developers and sticks to it, our team would be delighted to support Qt aggressively. Who wouldn't want to sell to a base of 400 million users? But given Nokia's history of whipsawing its developers, we won't take anything for granted. In particular, we want to see if Qt is actually the exclusive development platform for MeeGo, rather than just a secondary option. You've got to show us the consistency, Nokia.

Oh, and ignore Symbeose. I don't know exactly how the Symbeose initiative got started, but to me it looks like the Symbian Foundation lobbied for it for a long time, prior to the recent changes in the Foundation. For the old Foundation, Symbeose made sense, because it was a clever way for a nonprofit to get some OS development done in areas that Nokia didn't care about. But with the Foundation mostly gone, Nokia has no incentive to turn Symbian into a general embedded OS, and in fact it says MeeGo is its OS for use in non-phones. In that situation, I can't picture a lot of other companies committing to build Symbian OS into their products.

Lessons from the Symbian Foundation's demise

I'm seeing a lot of interesting rationalization online about Symbian's fate. For example, Tim Ocock, a former Symbian employee, wrote a fantastic post (link) in which he argues that Symbian was very successful as an OS for phones with PDA features, but was never designed for running browsers and lots of applications. That's a pretty shocking statement, considering how many times I heard Symbian advocates boast about the sophistication of their modern, general purpose OS compared to clunky old PDA-centric Palm OS. Remember, this is a company that until very recently was bragging about its superior implementation of symmetric multiprocessing (link), hardly something you need for a PDA.

But I think Tim is dead-on in most of his analysis. He did a great job of detailing the technical and attitudinal flaws within Symbian itself, so I won't bother repeating them here. Instead, I want to talk about the flaws in Symbian's governance.

Did Symbian fail? The companies that founded Symbian had two goals in mind: to prevent Microsoft from dominating the market for smartphone software, and to prevent Symbian itself from becoming a power that could dictate to the phone companies that funded it. As a result, Symbian's governance structure was designed with a complex system of checks and balances that wouldn't apply to a normal company. To make major decisions, Symbian had to negotiate a consensus among its owners the mobile phone companies, who understood little about the management of a mobile platform and were suspicious of each other and of Symbian itself.

This bureaucratic, highly politicized oversight process repeatedly forced Symbian into blind alleys, and prevented it from doing things that a "normal" OS company would take for granted. When Symbian was founded, there was talk of an eventual IPO. The prospect of an IPO is an important recruitment tool -- it lets you use stock to hire ambitious engineers and managers. But the idea was eventually shot down by the owners; it would have made Symbian too independent.

Crippled by design. Once the threat from Microsoft receded, the owners' second goal for Symbian -- preventing it from competing with them -- seemed to dominate their treatment of Symbian. I'm not saying there was some central evil plan to hamstring Symbian; there wasn't. But everything the company planned to do had to be approved by the handset companies, and on a case by case basis they vetoed the things that sounded threatening to them. Over time, this forced Symbian away from initiatives and features that would cause users and developers to be loyal to the OS rather than the handset.

So Symbian didn't create an app store, and Symbian's developer relations were very confused because Nokia wanted to do a lot of that itself. But the most egregious example was user interface, which Symbian worked on from time to time, but was eventually forced out of by its owners. When I was at Palm, the Symbian project I feared most was "Quartz," the effort to create an icon-driven touchscreen UI for Symbian. Quartz looked very nice, and if it had survived Symbian would have had a dandy iPhone competitor on the market before the iPhone launched. But politics between Symbian's owners forced it completely out of the UI business, and Quartz was spun out into a separate company called UIQ, which went bankrupt in 2009.

You can get more details on the whole sad Quartz saga here.

Quartz circa 2001

An OS without a single consistent user interface is a nightmare for software developers, because they can't write apps that run across the installed base of devices.

Eventually, in the face of all the restrictions, the most ambitious, nonconformist people at Symbian -- the ones who drive innovation in any organization -- seemed to drift away in frustration or were forced out when they irritated the owners. Symbian itself retreated into focusing on technological esoterica like symmetric multiprocessing -- things that didn't really differentiate the platform to users, but that the licensees wouldn't object to.

From one perspective I guess you can say Symbian was a complete success, because it fulfilled the two negatives that its founders wanted: Microsoft didn't dominate mobile software, and Symbian itself didn't exercise any control over its founders.

However, the cumulative effect of the handset companies pursuing their short-term interest was that Symbian was utterly unready to respond when Apple and Google entered the market. I don't think either Nokia or Symbian really understood how the game had changed. Apple designs phones as integrated systems, with the software and hardware tightly coordinated. Nokia could never achieve that level of coordination with an operating system managed through standards committees.

And as for Android, Nokia apparently thought that open sourcing Symbian would create a level playing field with Google's free OS. But I think the structure of the Symbian Foundation made that impossible.

The fatal flaw of the Symbian Foundation. Although Android is a free product, it's supported by a for-profit corporation that has massive resources. The attraction of Android to phone companies isn't just its price, but its safety -- Google stands behind it with marketing and technical support.

In contrast, Symbian Foundation was designed as a rigorously noncommercial institution banned from any business activity. People at the Foundation told me Nokia was adamant about enforcing the ban on commercial activity because it was afraid the tax authorities might rule that the foundation wasn't a nonprofit, endangering the tax credit that Nokia got for donating its Symbian code base.

Most open source companies give away their software in order to make money from some other mechanism -- consulting, or support, or a for-fee version of the same code. Symbian Foundation was banned from making money on any of these activities, meaning it could never become financially self-supporting.

Forget about marketing support; Symbian couldn't even offer enhanced technical support to licensees who were begging to pay for it. That was especially crippling because Symbian OS is notoriously complex and difficult to program (link).

Consider this quote from Tim Ocock's article:
"The difficulty of writing good Symbian code was hugely beneficial to Symbian as a business in the early days. For many years, 80% of Symbian's revenues were earned through consulting for licensees....Symbian’s licensees...each had their own proprietary telephony chipsets that needed to be integrated and their own customisations to the platform in mind....Despite talk of Symbian enabling differentiation, the reality was licensees' budgets were squandered on hardware porting and making the core platform fit for purpose."

Picture yourself as a manager at a handset company, choosing an OS for your smartphone. The Symbian option has no advertising support, requires customization, is hard to program, has few third party consultants to support it, and the company licensing it won't help you do the programming. Meanwhile, Google Android is more modern, is based on Java and Linux so it's easy to find programmers, has lots of support, and has user-friendly features like an app store. Which one seems the safer bet?

How could the Symbian Foundation ever succeed in that situation?

Although people advocating for a "European" mobile OS often complain that Android had unfair financial advantages, the fact is that Symbian was ripe for the picking, a situation that was almost entirely self-inflicted.

The lesson for other tech companies: Open source is not magic pixie dust that you can sprinkle on a struggling product to turn it into a winner. Open source is a tactic, not a business strategy. It has to be paired with a business plan that says how you'll make money and drive innovation.

This is the end, my friend, of our elaborate plans

Like an army refighting the last war, Symbian was designed to defeat Windows Mobile, but never came to terms with its new adversaries Apple and Google. There's no shame in that for most of the folks who worked at Symbian; they did the best they could to navigate the politics of Nokia and all the other Symbian licensees. But radical change was necessary. I hope Nokia's Qt strategy will be successful. And I'm sure that Symbian code will continue to serve for years as the underlying technology for millions of Nokia smartphones. But except in the dreams of a few EU officials, Symbian OS is now just legacy plumbing.

It's time to move on.

No comments:

Post a Comment