The Symbian Myth

Nokia’s in a free fall with smartphones at the moment. Having held on to the market leadership in volumes, with a slow and steady decline, for years, they may now actually have been overtaken by Apple in their market share, just as they were overtaken in both profit and sales revenue share by them before.

Now success in mobile is a multi-faceted thing. It’s not simply about having a better product, or better sourcing, or better advertising, or better sales channels. Why Nokia is failing is not a simple story, and a lot has been written about it. There are articles about when Symbian was killed, whether Steven Elop is a Trojan horse for Microsoft, whether MeeGo would have been ready before it was killed, the timing of the Microsoft/Nokia announcement, and all other kinds of aspects.

Through all this the question of whether Symbian would have still been a viable OS for Nokia going forward is usually a pure fanboy topic.  There’s little between “Symbian sucks” and “Symbian was and is the best mobile OS”, and actual arguments are scarce. Being a Symbian user, and having some experience with both Android and iOS devices to compare the Symbian experience to, I’d like to contribute to a more balanced and detailed picture here.

The question was posed here has two parts: There is the state of Symbian itself, and then there’s the state of Symbian development within Nokia.

The State of Symbian

When talking about the state of Symbian, not even the strongest proponents of Symbian argue that there aren’t at least some problems. The usual arguments here go along the lines of features and technical capabilities vs. interface, with the Symbian diehards maintaining that it is only with the later that there are any problems.

Admittedly, on a pure feature list comparison, Symbian may still be the leader. So let’s take it as a given that you can do more with your Symbian handset than you can do with one of competitors. Let’s also take as a given that Symbian handsets can have better battery life than those on other OSes. There is at least some truth to both. These are not the points I want to argue, because these are not the points that matter much out in the marketplace at the moment.

_Effective_ feature parity

The vast majority of users have a very limited base of knowledge when making a buying decision. Feature lists in consumer magazines and stores are very short. All the major current OSes have feature parity – as seen on these lists. Symbian hasn’t managed to get anything on there to differentiate itself. (USB on the go is not something on there, because it’s too unknown to be explained in two words. Ovi Maps is not a feature of Symbian itself – but Nokia should have marketed the hell out of it.) Battery life, while important in everyday life, is just one point on a checklist, and not something you experience when trying out a handset in a store – or trying a friend’s device.

Interface pain

What  sticks in these situations is the interface.  Now the Symbian interface certainly has come a long way since the release of Nokia’s first (mass-market) touch screen phone, the 5800 XM, and that itself improved vastly from the catastrophic launch state. Personally, I like a lot about the interface as it is. I’m not a fan of bling, and will take efficient over pretty transitions every time.

For example, once configured, the home screens are efficient and work for me. The ‘once configured’ is a big ‘but’ here, though. When my N8 forgot the entire homescreen configuration after a crash, I had to keep from screaming (no exaggeration here). Configuring the shortcuts widgets is a process from one of the deeper circles of usability hell. The choices made in designing this really do defy belief. Let’s take a closer look:

The UX hell of changing a homescreen shortcut

You can only have shortcuts widgets containing exactly four shortcuts – no more, no less. There are no single shortcuts. To change one of these four shortcuts on a widget you first need to enter the configuration mode for the homescreen (long tap anywhere, or via the context menu). Then you need to enter the configuration mode for the widget (tap, select option from pop-up context menu). This then switches to a different screen which gives you a list of the positions and the currently assigned shortcuts – in writing, without using the icons you just saw. You then select the one position to change by a tap. This brings up a question about the type of shortcut you want to replace the current one with: to an application, or a bookmark. Choosing an application brings up a scrollable list of installed applications. This list uses about a third of the screen, and so displays only five entries in a small font. Additionally, it only displays application names, no icons. You scroll through this list (painful because of the small size) and select an application by tapping on a radio button next to it. Then you choose back (no ‘ok’ or ‘save’!) to exit the configuration screen. Now all you have to do is tap ‘done’ on the homescreen configuration screen – and you’re done.

This is insane, pure and simple. That was nine steps to change a shortcut – across three different screens, none of which after the first one use the icons for the applications, which to me are their main identifiers. The first time I went through the process I didn’t believe it. This was clearly coded by an engineer – and not even using standard parts, since that ridiculous small application picker menu doesn’t appear anywhere else in the system, and doesn’t look like anything else in there. I believe even the font size on this menu is not used anywhere else. The engineer may be excused since he was assigned a task he obviously wasn’t qualified for, but the manager ultimately responsible for allowing this deserves to lose his job. It may be argued that this is not a very common thing to do in everyday use, and thus doesn’t matter too much to me now, but I can absolutely see a new user return the phone after first encountering this. After all, it is one of the first things you do with the phone.

There are other disastrous design decisions. Menus for configuration are still the same as they’ve been for years, nested three- and four hierarchical layers deep. I often need to go on a search expedition to find particular points that I know are there, and I’ve capitulated in respect to the SIP client that’s supposed to be in there and working. Using a third-party app is just less painful.

Just plain broken

This is the design issues part  – and as said before, Symbian fans do admit that there are some problems there. There are, however, a lot of parts of both the interface and the underlying system that are not misdesigned but plain broken. Examples?

The taps on the home screen that get registered (i.e. the icon changes to indicate this) but don’t have any effect. The application menu that gets massively slower to first display the more applications are installed (did somebody, a long time ago, try to save memory there by not caching the application icons?). The screen timeout, which does not work (it’s only the auto lock that turns off the screen, before that there’s just a dimming that somewhat imitates the behavior of an LCD screen with the backlight turned off).

Setting a browser other than the default one gets ignored even by some of Nokia’s own applications. The browser itself – which by now is so old that it has to be considered broken by today’s standards. The ‘back’ button in the email app which exits the app when you’re in a subfolder instead of returning you to the folder overview menu.

Payment and installation via the Ovi store – which, for me, has a failure rate of about 25%. The Qt smart installer which for a while crashed the phone each time during the installation of a new Qt app, and then needed manual removing before allowing any new installations at all. Nokia Social, which often takes a few attempts to start, and then often presents me with a black screen.

The network stack: some applications always ask for permission to connect, others connect via packet data when wi-fi is available. The stack even allows simultaneous wi-fi and packet data connections! Sometimes, for some applications, the connection process is not even triggered at all – but this doesn’t get communicated to me as an error. I once wondered for a couple of days what was happening since email and Opera Mini were still connecting, but the internal web browser always failed. I had to switch to manual approval of connections for that to get things going again. At this point most users would have either given up or contacted support. Accessing links from Nokia Social while on 3G? I’ve given up – I don’t see any pattern to the few times it actually works. Oh, and the fact that there is a single font that Symbian^3 uses for all font rendering, including web pages? I’ve never heard this mentioned by anybody in the Symbian world, but it is bizarre for a smartphone in 2011.

Two steps forward…

Other things are even more bizarre: Symbian^3 actually has less features in many respects than the previous versions. No more auto-redialing. No user-adjustable EQs in the music player. No podcasting – at all. No converter app, or stopwatch.

Some of these things may be cosmetic. Some may indicate shoddy development and quality control. Others seem to indicate some things are really fucked up under the hood. Taken together, they lead to pain on a daily basis when using a Symbian^3 handset. No, it’s not all pain, and there are a lot of good things about Symbian^3, but the level of bad design, confusion and frustration is simply unacceptably high.

S60 the edition – the skeleton in the closet

And I’m only talking about Symbian^3. What is often forgotten is that Symbian^1/S60 5ht edition is still what runs on all the low-end Nokia smartphones – which should have been at least a significant part of the 150 million Symbian handsets that Nokia still wanted to sell. And Symbian^1 with resistive touch screen on a 3 year old hardware platform vs. the current crop of low-end Android phones? No competition to the average user, not at any prices where Nokia could still make a profit.

Developers: There’s no pain like Symbian

Since today users are not the only people whose interests are important, let’s take a quick look at the other big group: developers. I’m not one myself, so I’ll have to go by what I heard. A friend described the experience of developing for Symbian as the most perverse, painful thing he’d done in over 25 years of programming. The sentiment has been echoed by others. Well, this was supposed to change with Qt. Which is currently in some transitions under the hood. Where the toolkit still doesn’t support kinetic scrolling. Or have a standard widget and icon library, apparently. So no, despite significant improvements, the average developer will still choose iOS or Android over Symbian without a second thought. Now in theory the strategy of taking the pain away from developing by establishing Qt for development, and then easing the transition to MeeGo because these Qt applications would run on that as well (with small changes for anything non-trivial, but still) was brilliant. The best in the industry, as a matter of fact. But then there’s nothing theoretical about the handset business. And the market doesn’t forgive delays in execution.

Symbian in the State of Nokia

That brings us to the second question: would Nokia have been able to fix the problems with Symbian? My consumer view is shaped by the fact that the promised update for my N8, the mystical PR 2.0, the one with the real goodies like then new browser, a portrait QWERTY keyboard, and some real fixes for the current problems, is already between 5 and 7 months late – depending which non-announcement from Nokia you believed. The only thing they’ve managed so far is a change of name – it’s now called “Anna”, but still projected for “the next few months”. There are some puzzlers here as well. That the portrait QWERTY keyboard everybody in the blogosphere demanded has not been delivered yet independently of “Anna” speaks of gross mismanagement regarding development priorities. This particular point can’t be a technical problem with Symbian – and plenty of third-party developers have implemented such keyboards in their own applications, so it can’t be a matter of resources either. Or perhaps the update process itself is broken, and there are bizarre, deep interdependencies that don’t allow such changes on their own?

Regarding the resources for Symbian development at Nokia: There are blog posts out there from former Nokia employees who tell of entire new UIs and frameworks for various things being scrapped, of double- and triple developments, of reverse engineering UI patterns from the already written source code (!). There are references to fractions inside Nokia that effectively sabotaged, and sometimes entirely ignored, instructions from the CEO and higher-ups. How much of this is true matters little.

In the end there is a simple truth regarding Nokia and Symbian:
Nokia had years to fix Symbian. Whether you count from the time the iPhone was released and upped the game regarding UIs, or from the release of the 5800 where the public reception must have told them there were some very big problems, its years either way.  They had 6500 engineers working on it. If you think that is an incredible number of people – it is. There’s nothing like it in scale in the entire handset industry. With all this time, and all these resources, we got Symbian^3. An update to Symbian that was delayed, and then delayed again. An update that improved things, but not by enough. We got promises of delivering upgrades and fixes quickly. This simply hasn’t happened. Nokia has proven that it isn’t able to deliver software of necessary quality, and to deliver it on time. It has been proving it for at least the past two years. Whether this is the result of gross mismanagement, the result of an OS that is no longer maintainable, or, most likely, a mixture of both is an academic question. Changing any of the reasons to enable it to catch up with the market would have required time. A lot of time. In today’s market, this is not something that any handset maker has.

The Symbian myth: Symbian as a viable OS at Nokia

So I think it is a myth that Symbian, at Nokia, was still a viable OS in today’s market. It is a myth that, without Steven Elop, Nokia could have found a way to soldier on with Symbian, and maintain its market share. Nokia would not be at 30%+ market share with Symbian at the moment, no matter what. Symbian handsets would not be considered market leaders by anyone. Not in this universe, not with where Symbian stood at the end of 2010, and not with the track record that Nokia has at this point in time regarding software development. This truth was a slow and painful realization for me. Things seemed OK for most of 2010, and it seemed like Nokia had a chance to turn Symbian around. They missed it. I am deeply saddened by this, but it’s no use pretending the situation was any different.

 

Advertisements

16 Comments

Filed under interfaces, mobile

16 responses to “The Symbian Myth

  1. marco

    I think you are correct about many things here. But did Elop have to pull Nokia off the shelves of all major operators on Feb 11th?

    Why did it take him so long to recognise his mistake and announce that there would be support for Symbian until 2016?

    Symbian might not have had a bright future, but right now I should be able to walk into a store and get an N8 or a C7, and its good enough and competitive enough right now for the price that I could live with it for 2 years until its time to replace. But Nokia phones have been pulled off the shelves and if I ask the salespeople they tell me that Nokia has announced that symbian phones are obsolete. They say I must get something else because Nokia doesn’t make any other phones for at least another 6 months.

  2. Graham

    Interesting post, though whatever the technical issues with Symbian, one has to feel that at least some of the issues are pure management intransigence. Nokia managed to launch the Astound with an OS which must be something more than halfway towards what they are touting as “Anna”. And that was, what, a couple of months ago now? Making that same approximate build available for other handsets would be perfectly in line with their stated change to more rapid, incremental improvements. It’s quite impossible to understand their decision making.

  3. RobDK

    Great article! Very true! The N8 is an appalling phone!

  4. Matt

    I completely agree with your analysis. I think nearly everyone inside Nokia also acknowledged (if grudgingly) that a transition strategy to another platform was necessary, and that Symbian was not malleable enough to move forward with.

    Given that, the problem with Elop’s actions were two-fold: first, he chose a forward direction that denied a migration strategy to the vast numbers of users and developers that were linked to Symbian. And second, he announced these decisions immediately, with the effect that all Nokia products until the arrival of WP7 devices are effectively DOA.

    Personally, I think he is an egomaniac. He wasn’t satisfied with making the correct decisions to save the company behind closed doors – he wanted the whole world to know that he had taken charge and made the tough decisions, and would thus be hailed as Nokia’s saviour. Well, he will certainly be mentioned in future textbooks!

    • James

      Matt: “He wasn’t satisfied with making the correct decisions to save the company behind closed doors”

      Me thinks that MSFT had something to do with this. An announcement ASAP would help their Win7 sales effort.

  5. Nice post. I see you read mine – it is true – you can probably still find the built and abandoned remains of the crappy replacement UIs that Nokia created.

    This is the important bit that people miss from those that argue Symbian was still a viable operating system for Nokia… the user interface is not part of the operating system, it sits on top of it. Symbian has had multiple different UIs in its history and it could have had another. The move to name the whole OS + UI layer Symbian^3 help compound this common misconception.

    S60, the application and user interface layer is a total mess, completely unmaintainable and utterly beyond salvaging. The sort of UX problems you’ve described are the tip of the iceberg. The comms stack is flexible, as it should be – proper access point priroty management is implemented on top in the middleware and then has a hopeless settings UI and lots of the apps (even though developed by Nokia) completely ignore it and make their own choice of access point – mostly for historical reasons but that’s not an excuse for the not fixing it!

    So, the argument from those in the know that say Symbian was still viable, is that the UI layer needed completely replacing, not fixing. A plan to do that with Qt was botched an abandoned, then a better plan to do the same has been downgraded to patching up the worst of the problems with new Qt versions of some of the core apps, since the new plan is Windows Phone.

  6. In answer to the comments:

    marco:
    By all rights you should be able to get the entire range of Symbian^3 phones in any store right now. Symbian not being competetive in terms of future developments doesn’t mean it’s completely hopeless now. It just has to priced very low, and the right kinds of aspects have to be marketed. Free, world-wide, offline satnav is a trump that I never say played up enough. Marketing social integration instead was a mistake.
    And to be clear: Steven Elop did not pull anything from the shelves. The February 11th announcement certainly contributed to this, but I believe it wasn’t intentional.

    Graham:
    That is the maddening thing. I believe there are deeper issues with Smybian that made it unviable for future development. A lot of the changes that would have allowed to stand it in the market are relatiley small though – and some of them are in “Anna”. Not having them out in all the Symbian^3 phones is a management failure – like so many with Symbian, and Nokia in general, in recent years. E.g. can anybody explain to me why the N97 (original, not mini) was sold for two years, with it being a phone that drove Nokia fans away in droves?

    RobDK:
    If only the N8 were so appaling all round. The design, the screen, the way it feels in the hand, the standby screen clock, USB on the go, the HDMI out, the unlock slider (once quick gesture instead of two across two elements to unlock the phone), and, of course, the camera: there are a lot of things I like about the phone, which makes its downsides all the more maddening.

    Matt:
    I’m not speculating about Elop’s motives. The announcement should have been handled differently – and given at another time. Killing Qt as a migration path after pushing it so hard was catastrophic, especially since developer interest was starting to come back, and the Ovi store (bad as the user experience is) was starting to gain real momentum and to make money for people. Unfortunately, I think the damage here is done. According to the new VisionMobile developer report, developers are moving away from Symbian fast now. And with all the strategy, toolchain, UI etc. changes of Nokia in the past, I don’t really think there’s much in the way of goodwill out there to support yet another, hypothetical change of direction.

    Marc Wilcox:
    I don’t know much about the internals of the core of Symbian, but two points regarding continuing to use that:
    First: Why maintain another kernel, network stack and device drivers, when the UI is that separate and is what you differentiate on most? Sure, Symbian was designed in times when memory was measured in KB and every processor cycle counted.It can run efficiently and on low-end hardware. For now, for the transitioning period when every dollar in the BOM for the low-end smartphones that replace (lower) mid-tier feature phones counts, this is an advantage. In the long run, having to deal with all this, especially with the device drivers, when due to Android, webOS, MeeGO (whatever becomes of it) a working Linux stack for SoCs will be something that chipset manufacturers contribute to, is an unnecessary sap on resources. A transition to something Linux-based would have been wise in any case. Not because it is open source, like Tomi Ahonen keeps repeating. Open source does not have value in itself, not to handset manufacturers, but because it would have enabled concentrating on the important parts.
    The second point is that whatever was going on with Symbian on a technical level, the management level has problems that are unlikely to be fixed in the short term, or even mid-term. Whatever merits Symbian still had, and however brilliant the Qt transition story was – without proper execution it wasn’t worth anything. Just look at how badly Nokai handled that announcement back then, when at least half the blogosphere, plus mainstream journalists, where confused regarding what to think of Maemo (the MeeGo) and the plans.

    • Absolutely yes. At an OS level Symbian has two big problems (1) Comparatively very few developers know it so it’s hard to hire engineers for a new project – this hurts Nokia too if they want to switch to a new silicon vendor that doesn’t have a strong Symbian team (i.e. anyone) – which leads nicely into – (2) No-one else (outside Japan on homegrown hardware) is using it, so there isn’t a wide hardware choice and no-one else is sharing the OS maintenance costs. Because of these two factors, most silicon vendors use Linux to bring up new hardware and thus Linux-based platforms have a time-to-market advantage.

      I’ve long felt Symbian’s days were numbered and it would be transitioning gradually down the value scale until hardware got cheap enough or Linux got trimmed down enough to phase it out completely at the low end. Nokia needed something that would scale down to compete for the feature phone consumers trading up. If you’ve tried the really cheap Android devices you might not think even the current mess of Symbian was that bad. Linux is quite a way from running well on cheap phone hardware. Switching to M$, which doesn’t solve the hardware porting problem (although M$ will be paying for that I suspect) and doesn’t scale down (yet) either doesn’t seem very smart. It’s probably true that Windows Phone can scale down quicker than Linux. Killing Symbian prematurely for “cost savings in R&D” is going to be the most expensive thing Stephen Elop ever does in his career. The thing is, as I wrote, they’

      Regarding the Symbian management issues – they existed in MeeGo too – they’re Nokia management issues and have to be solved whatever platform they use.

      • oops, lost a bit there. “The thing is, as I wrote, they’…”
        As I wrote, they’re trying to do the low-end smartphone transition with Symbian anyway – they’ve just told the world it’s going end-of-life first.

  7. @Mark Wilcox
    I think we agree in principle about a Linux stack as the way forward. I also agree that for the time being, in the sub-$ 100 segment, Symbian^1 is still the best choice among existing smartphone OSes.
    Android not running smoothly on low-end devices is not necessarily an argument against Linux, but I guess more about the additional Jave-VM layers they put on top. The power to jump through that many hoops just isn’t there yet at the low end. The N900 by now is relatively low-end hardware, and it runs a Linux-based system just fine – and would be even better without having to run the X-Server.

    • The N900 isn’t even close to low-end yet. It’s a very expensive bit of hardware and needs a lot of RAM & Flash. Linux + some kind of fully featured smartphone platform is yet to run well on low cost hardware, Java-VM or no. There’s a whole community funded primarily by ARM #(Linaro) looking to make Linux work better for smartphones on the Cortex class processors like the one in the N900 – they aren’t really doing anything to help it scale down to older cheaper things like an ARM11, which will still be in use on new low-end devices for a couple of years I expect. Symbian’s window was the gap while we wait for the Cortex class SoCs to get cheap enough for the low cost masses. It got closed somewhat prematurely. Instead we have folks trying to run Android on a single core ARM9/11 device with software hypervisor to run the modem on the same chip – these devices are nothing short of terrible.

      It’s not really the Linux kernel itself (although that is bloated these days as Linus himself admits) but that the common open source middleware that the various Linux-based phone distros have used was designed for the desktop without low cost phone power and memory constraints in mind. Those constraints are disappearing as hardware improves – so Linux based something is definitely the future.

      • I think the N900 IS close to low end now. I fully expect to see entry level phones built on N900-and-similar hardware by next year.

      • Not sure why I have no option to reply to Texrat’s comment directly… but we must be talking about seriously different definitions of “low-end”. I’m talking about where the mass billions of feature phone users will trade up to smartphones, at least a sub-$50 BoM. True entry-level phones now are still not quite up to the specs of high end phones I was working on 7 years ago.

        Lowest of the low-end Android is currently running on an ARM9 core with the modem (using a simple RTOS) sharing the processor via a software hypervisor. It’s also a QVGA screen which makes web browsing completely unusable and isn’t supported by a majority of the apps. These still don’t make it to true entry level pricing yet because they need hardware graphics acceleration and a minimum 256MB RAM to run Android.

        N900 level hardware (Cortex A8, large screen, 256MB RAM + lots of flash for virtual memory, very capable hardware graphics accelerator) will certainly be credible at the mid-range next year.

        • (Fixed the setting for the nesting level for comments. Hadn’t had a look at that before.)
          Thanks for clarifying regarding low-end Android specs. Pity about the window for Symbian. Wonder what is filling that market demand now, with Symbian sales crashing so completely at the moment.

          • To some extent, still Symbian – I guess without seeing any sales figures that mid-to-high-end Symbian sales in developed markets are what’s really fallen off a cliff. In the slightly lower cost segments Samsung’s Bada and low-cost Android will be the main gainers from “Nokia’s gift to the other OEMs” as Tomi Ahonen called it.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s