Symbian unloads UIQ, and the mobile apps situation gets clearer -- and uglier

As you've probably heard, SonyEricsson bought the UIQ user interface from Symbian this week. The only surprise in the deal was that it didn't happen a long time ago -- Symbian had been trying to unload UIQ for at least four years (they tried to sell it to Palm at least two times that I was aware of).

I haven't seen any online information about the terms of the deal, but at some point I hope it'll come out in a public filing. I'll be shocked if it turns out that SonyEricsson paid a lot of money. Knowing what it costs to create an operating system layer, I think it's very likely the UIQ team has been losing money.

With the UIQ acquisition, three of the top mobile phone companies now have their own operating system layers -- Nokia with S60, SonyEricsson with UIQ, and Motorola with whatever it's building on top of Linux.

Years ago, we were expecting to see the development of one or at most two operating systems that would run across all mobile devices. The idea was that that phones would become like PCs -- hardware decoupled from software, with low costs for consumers because the hardware was standardized and the cost of the OS and apps was spread across a huge base of hardware.

Instead, the mobile market is moving in the opposite direction, at least for now. Most of the major phone companies are creating their own OS layers, and they're not compatible with one another. I think this raises big questions for the mobile market:

What will Samsung and LG do? The other two big mobile phone companies don't have coherent software strategies, as far as I can tell. They specialize in copying others -- Samsung copies Motorola and Nokia, while LG copies Samsung. As the implications of the UIQ deal sink in, I think both Samsung and LG will start to feel they need software layers of their own. They could try to buy one, they could cut a deal with Microsoft, or they could try to write their own software layers in house.

None of the three options is attractive. There aren't a lot of mobile software layers left on the market to buy, Microsoft is well known for acting like a predator on its licensees, and creating user interfaces isn't exactly a well-known core competency of Asian hardware companies. My guess is that both Samsung and LG will continue to play the field by working with multiple operating systems, while they hope that mobile Linux will magically mature and solve their problems. But I wouldn't be shocked if one of them decided to take a chance and partner deeply with Microsoft.

What happens to Palm and RIM? RIM is going strong at the moment, while Palm is under moderate financial stress. But both are relatively small phone companies that rely heavily on software innovation for their differentiation. All that software investment makes their cost per unit a lot higher than the big guys. Their have to keep their products dramatically differentiated in order to maintain the price delta that funds their businesses.

Can they keep that edge as the big phone companies start adding more and more software value to their products? It's by no means impossible, and I wouldn't bet against the smart guys at either company. But they're both like small movie studios that fund their next movies directly from the profits of the last one. One major dud and the process could fall apart.

What about Microsoft? Theoretically, Microsoft should be cackling right now. Microsoft (and its then-buddy Intel) destroyed PC companies that were dumb enough to sell proprietary systems against the army of PC clones in the 1970s and early 1980s. Microsoft has heavy ties with the Asian phone manufacturing world, and should be able to help them produce devices which pair sophisticated software with commodity pricing. I think of this as the "HTC=Compaq" scenario.

But there are three problems with the scenario...

1. No IBM. In the PC world, Microsoft benefited immensely when IBM set a hardware design standard and then let everyone else clone it. Without IBM's help, Microsoft might never have created the clone PC market, and we'd be dealing with a much less standardized (and more expensive) PC market to this day. The closest thing to an IBM in mobile phones is Nokia, and it ain't about to let anybody clone anything, and certainly not for free.

2. No apps. The other key to Microsoft's power in PCs was the huge base of software applications that developed for DOS and Windows. People wanted Windows in order to run the apps, and Microsoft was the gatekeeper. In case nobody's noticed, mobile app sales for smartphones are miserably low, due to app distribution problems and the operators' unwillingness to allow truly open phones to be sold in mass volumes. If there's no big applications base, Microsoft doesn't have the leverage to force its software onto the world's phones.

3. Microsoft doesn't really "get" mobile. Although Microsoft's mobile software has definitely improved, the company still shows the reflexes of a PC company. It relies on cool technology to motivate customers, overloads its products with too many features for the average user, and counts on Moore's Law to bail it out over time (if you doubt this, check out David Pogue's review of the Zune media player in the New York Times). The mobile market rewards minimalist design in both hardware and software. I think that's not in Microsoft's DNA.

Is this good news for Symbian? Sure, in the sense that they've unloaded an irritant that complicated their relationship with their sugar daddy Nokia. Symbian seems to be on a path to try to make itself the embedded OS of the next generation of mobile phones -- kind of a super Nucleus. That's the only path its owners were going to tolerate anyway, so the Symbian folks might as well smile and make the best of it.

But there's no law of nature that says S60 or UIQ has to always run on top of the Symbian OS. In fact, as Nokia and SonyEricsson define more of their own application interfaces, and as the development of native Symbian applications continues to lag, there's very little to stop Nokia or SonyEricsson from moving their interfaces to run on top of a different OS. Maybe onto Linux, which doesn't charge Symbian's several dollars a unit licensing fee. The expense of supporting an OS layer team is very substantial, and only grows as the code gets older and you have to maintain it. That cost on top of the Symbian license fee is going to be pretty large even for a big phone company. The more units that Nokia and SonyEricsson build Symbian into, the more the Symbian licensing fee is going to bother them.

So Symbian is on a treadmill in which it will be under constant pressure to match the pricing of Linux. Unless Symbian can develop a big native apps base of its own, it'll have very little market power to keep its licensees loyal.

The most likely outcome in the near term will be a divided mobile market in which each of the major phone companies has its own operating system layer, with its own base of applications. Microsoft will continue to plug away with its Asian buddies, but without the critical mass to take over the phone market. We're looking at a minimum of five major OS layers (S60, UIQ, Moto Linux, Microsoft, Qualcomm Brew), plus smaller contenders (Palm OS, RIM). Oh, and don't forget Java.

This situation will drive independent software developers insane, because they'll have to rewrite their applications for every phone platform. This will drive a lot of the most creative software developers away from the mobile market; they'll continue to focus on creating web apps because they face lower barriers to entry, there's a single fairly unified platform, and they'll have more control over their own destiny.

Mobile phone enthusiasts like to point out how many mobile phones there are in the world, a much larger market than PCs. They say that larger market means the future of apps innovation will inevitably happen on mobiles. But if the mobile market is divided into five or more incompatible camps, and it's not easy to make money in any of them, the mobile apps market will be stunted indefinitely.

At some point, I believe one or more of the mobile operators will become so anxious to access the wealth of innovative web apps that they'll set up a fairly open garden for web apps developers. I don't know what software that garden will be built on -- maybe it'll be Adobe Apollo, maybe something else. But when that happens, I think we'll finally see a critical mass of third party mobile apps that will then force the other companies to deploy the same environment. It'll be a giant layer cake -- the web apps layer on top of the phone vendor's layer on top of the phone's native operating system.

The whole thing sounds incredibly baroque and inefficient, and it is. But it says something about how messed up the mobile phone apps market is, that this is the optimistic scenario for the future.

_____________

PS: Thanks to the folks at AllAboutSymbian for linking to my post on Sprint's Ambassador program in the latest Carnival of the Mobilists.

19 comments:

Anonymous said...

I'm a developer for mobile applications. I directly feel the pain from all of the different OS's. You described the environment pretty well.

I took off four months to learn and become fluent in Flash. It seemed to me that this new layer was the common solution capable of providing professional looking applications without the over-the-top porting issues of making native apps work on all the OS's. You touched on this near the end of your posting where you talked about layers and Apollo. Somehow, I think there is more there there (duplicate intended).

I've written some play-around FlashLite apps but they just aren't ready to be used professionally yet. Maybe Apollo (lite?) will do the trick. There seems to be a lot of waiting around for Adobe and I wonder at times if they know what they have. They could eliminate a layer or two if they wanted to (per your description). They also need to reduce some of the brain-dead issues that FlashLite has that are stopping developers like me from using it for real.

What will it take for Adobe to realize that they could be the Developer Studio for mobile apps across all platforms?

Anonymous said...

Great post. Although Symbian claims the largest installed base of smartphones of any OS today, I can see it going to zero in five years.
What matters to Nokia (and to all the other OEMs and operators too) is the GUI and application layer; that is where they differentiate.
Currently I run Aqua on top of unix on an intel box; that is to say I have a MacBook Pro. In the long term I think Nokia will do what Apple did, and develop some version of S60 that runs on top of mobile linux on standard hardware, and differentiate through the look and feel of the box and the GUI/app layer.
That would be cheaper for Nokia, better (although not ideal) for the developer community, but death for Symbian.

Anonymous said...

Great article as always. It's unfortunate how things are shaking out, and it seems like operators, phone makers, consumers and developers are all facing a tough situation.

But I'm interested in how you see Access (ALP) fitting into all of this. Are they a legitimate player now? And when you refer to PalmOS, are you meaning Garnet, or Garnet + ALP?

I'd also like to hear more about your thoughts on flash, Apollo, web apps, hybrid apps and how that might all evolve. And whether it has a place in the mobile market for the next 5 yrs.

Admin said...

I don't think Sony Ericsson will move UIQ to another OS, but if they did, Linux is the only realistic choice. That way Sony Ericsson wouldn't be dependent on Symbian OS, that's very much controlled by Nokia.

"Unless Symbian can develop a big native apps base of its own"

They can't. The only option is that Nokia does, but that likely doesn't help Sony Ericsson, and also: Symbian OS is used in just a small fraction of all phones.

You almost forgot to mention that >99% of all mobile applications are currently developed for Java ME / MIDP, and MIDP is also supported by smartphones, so there's actually a common platform of sorts. Even though we mostly hate the differences in Java implementations between devices, it's at least the only broadly deployed platform that is somewhat OS and phone independent.

Anonymous said...

Great post and clear analysis.
I dont understand why a great application base is a decisive factor. Palm has the largest base of apps, and it doesnt make it to sell more palm-powered-phones.
In my opinion, it is more important the carriers aptitude: they are configuring the market, subsidizing terminals, putting pressure on manufacturers, and charging oversized rates for traffic data.

Michael Mace said...

Excellent comments; thanks.

I want to come back and respond to the questions properly when I have a bit more time, but meanwhile here's some of the information people asked for:

More information on Flash/Apollo. Adobe has very aggressive plans for Apollo in both the PC and mobile spaces. I wrote about them a couple of months ago in a post here.

What do you mean by Palm OS? I was being deliberately vague, because the situation there is so confused. When I said Palm OS I was thinking of: the Palm OS Garnet APIs (now being licensed by both Access and StyleTap), the Access Linux Platform, and whatever Palm is going to run on its devices in the future.

I have no idea which of those are going to be most successful in the long term. It'll be up to the licensees and users to vote with their wallets.

Anonymous said...

The sleeper platform in this persistently multi-OS environment is SuperWaba. I continue to be amazed that someone with some money and marketing skills hasn't scooped this little Brazilian company up and given its platform the visibility it deserves, because their product clarifies the muddy mobile platform waters like nothing else I've seen.

* It's the only mobile Java platform that offers a truly "write once, run anywhere" development experience. "Everywhere" means Palm OS, Windows Mobile, Symbian, PSP, and Linux (with a BlackBerry VM supposedly in the works).

* Unlike Java MIDP it supports native libraries, so developers can use platform specific-APIs if they want the full power of the native OS at their disposal.

* It's fast, small and amazingly rich, with SDL graphics, a powerful sync-able SQL database engine, and a rich, snappy user interface that with just a little work can perfectly mimic the look and feel of the underlying OS.

* It's Java language so the tools support rivals Microsoft's for power and productivity (thanks to Eclipse).

* Unlike the competition, it's open source, so it can be distributed for free to end users.

SuperWaba's greatest liability has been its low profile outside of Brazil, where it's evidently all the rage among developers. Here in the US it seems to be mainly used for vertical applications. I doubt that most of your readers here have ever even heard of it, although if they download third party software they could very well have it running an app on their smartphone without even knowing it's there.

I realize these are rather gushing remarks, but the result of my work with SuperWaba has been that neither I nor my customers have to care about mobile OS fragmentation or shifting market shares any longer. That's been a substantial advantage for my business so when I see people write about the mobile app situation getting "uglier" it's hard not to pipe up about it.

Anonymous said...

My felling is that the problems Mike is alluding to will drive more and more people to Flash Lite once handset adoption speeds up.

Linux *will* rule the low level OS. The number of players looking at the platform coupled with the economic advantage in what is a cuthroat margin-sensitive business makes this inevitable.

Standards bodies like OSDL-MLI and LIPS are mandated by companies and a number of operators to define lower level standards that foster interoperablily. Its important to remember that operators need apps to be successful as it will the drive renvenue they are desperate for and app support only pulls down their bottom line.

JMH

Anonymous said...

"But there's no law of nature that says S60 or UIQ has to always run on top of the Symbian OS." - MM

This is where having a modicum of clue would help you Michael and all your commenters.

As much as they might want to, and as much as they might threaten to whenever it's time to revisit their Symbian license fees, S60 and thus Nokia (and SE/UIQ) are totally locked in to the Symbian OS way of doing things (Descriptors, Active Objects et al.) and therefore totally locked in. They might be 'just the UI' but when your UI is built on an architecture that is patented to the hilt until 2017, you can't just go writing a compatibility layer for Linux.

Not to mention the 2000+ Symbian OS engineers and hundreds of millions of sunk investment at these companies. Even these billion dollar behemoths are accountablte to their shareholders, who might not smile kindly on 'starting from scratch' versus using what's already finished.

Symbian C++ is not your father's C++, to do it well takes years of learning. Crap tools don't help either. These firms have already been through the curve though, so they can zoom ahead while everyone else plays a futile game of catchup (which is exactly what they are doing).

Nokia is already building an alternative platform based on Linux, called Maemo. But it's not for phones (for the foreseeable future anyway), and it's not S60 (and it's not very good either).

Dean Bubley said...

Michael

Lots of good stuff in this post. I've been saying for a while that I didn't believe the handset OS space would consolidate that much.

As well as all the platforms you mention, it seems likely that low-end phones will be based on proprietary embedded OSs for at least 10 years (can you imagine a Symbian or Linux ultra-low cost $20 phone? I can't....)

And despite the last poster's comments, I remember asking an S60 representative about whether it could work on Linux, 2 years ago at 3GSM. The guy went white as a sheet, and said I'd have to ask their analyst relations folk...

For me, the winner will be whichever platform can deliver true multitasking (but not necessarily full open-ness) on a phone priced right for 3G on prepay. The more I look at this market, the more I think that multitasking has more inherent value than "smartness" per se.

Michael Mace said...

Anders wrote:

>>You almost forgot to mention that >99% of all mobile applications are currently developed for Java ME / MIDP, and MIDP is also supported by smartphones, so there's actually a common platform of sorts.

If mobile Java can ever get to the point where "write once run anywhere" actually works, I'll be glad to consider it a platform. But right now it's more like a programming language – a shared syntax that you use across different platforms, but with lots of customization to the particular system.

I think there's a very good chance that Flash/Apollo is going to take the "write once/run anywhere" position before the Java world gets its act together.

David's comment on SuperWaba gives me some hope. And I think Access could change the world, even at this late date, if they open sourced Palm OS Garnet.


Oscar B. wrote:

>>I dont understand why a great application base is a decisive factor.

You're right, it's not a decisive factor at the moment. I believe that's because mobile apps are hard for the average user to discover and install. I also believe that the apps would matter a lot if those problems were fixed. DoCoMo's iMode is my favorite example of a mobile platform done right. I wrote extensively on the subject here.


Anonymous wrote:

>>S60 and thus Nokia (and SE/UIQ) are totally locked in to the Symbian OS way of doing things (Descriptors, Active Objects et al.) and therefore totally locked in. They might be 'just the UI' but when your UI is built on an architecture that is patented to the hilt until 2017, you can't just go writing a compatibility layer for Linux.

Maybe. But since there's very little native software available, you wouldn't necessarily have to write a full compatibility layer. In fact, that's the last thing I'd do. It would be quicker and cheaper to just have a team re-implement the core apps and UI on a new platform. That would take time, but it's eminently affordable for Nokia and marginally affordable for SonyEricsson.

>>hundreds of millions of sunk investment at these companies. Even these billion dollar behemoths are accountablte to their shareholders, who might not smile kindly on 'starting from scratch' versus using what's already finished.

I think investors don't care about sunk costs nearly as much as they care about forward-looking revenue forecasts. If Nokia figured out a way to cut $2 a unit out of its bill of materials, the investors would form a conga line all the way through downtown Helsinki.

In mid-winter.

At night.

Naked.


Dean wrote:

>>it seems likely that low-end phones will be based on proprietary embedded OSs for at least 10 years (can you imagine a Symbian or Linux ultra-low cost $20 phone? I can't....)

Ten years is beyond my planning horizon. But I can't picture it anytime in the next five years.


>>I remember asking an S60 representative about whether it could work on Linux, 2 years ago at 3GSM. The guy went white as a sheet, and said I'd have to ask their analyst relations folk.

Oh, yeah. I think Nokia has been playing with all sort of heretical layer-like ideas (they didn't toy with buying Palm OS because they wanted the kernel). But whether they'll actually implement anything is another question. I think the economic incentive is there, and will grow over time.


>>For me, the winner will be whichever platform can deliver true multitasking (but not necessarily full open-ness) on a phone priced right for 3G on prepay. The more I look at this market, the more I think that multitasking has more inherent value than "smartness" per se.

Let me put it this way – having an independent "open" OS in a phone, with all the associated APIs and documentation and overhead, is overkill unless you also open up the developer ecosystem. If you're not going to open the ecosystem, then yeah you might as well use an embedded OS/kernel.

But then give up on the fantasies about getting an Internet-like explosion of mobile data activity. Won't happen without the ecosystem.

Anonymous said...

Very interesting post. I would agree that there is no law that says S60 / UIQ has to run on Symbian (MOAP-S vs. MOAP-L being the best example of this). However I do think that the argument gets over simplified into Linux is cheaper and better and will therefore become dominant.

My comment would be there are two things that need to be considerd with regards to OS - cost and technical ability.

I would add a third - defintion, but I think most here would agree with the kernel + device driver and middleware in the context of this discussion.

The cost issue seems simple - Linux is free and therefore must be cheaper. However Linus is not free - the kernel maybe, but the middleware layer is not. Moreover refinements to to ther kernel (e.g. Access changes for memory management, boot time and power consumption for ALP) are necessary. The middleware component needs more works, though some svaing through the use of opensource components may be possible (especially with device drivers). However even using free stuff is not free (there's still integration etc etc). Do people really think that the Linux Platform companies (ACCESS, TrollTech etc.) are not going to charge a license fee. Of course the debate as to how much this will be ins open. Even Motorola who is doing their own thing will have costs associated with developing and maintaing the platform. How much will this be per linux phone they produce - open to debate perhaps? Then of course there are questions as to whteher Linux will platformise sucessfully in the same way S60 has for Nokia. (You could look as MOAP-L in Japan for the cost issues here). Thus to say Nokia could save the license fee to Symbian merely by switching to Linux ignores the cost of using Linux (not free), the switching costs (cost ?), and the impact cost (within the ecosystem). The switch is not nearly as trivial as it would first appear - you have to understand both S60 and UIQ are tied more closely into the OS than traditional UIs. Symbian actually provides a UI framework and many of the underlying data store mechanisms. The middleware part of software that supports various technologies and standards is very susbstantial. Yes it may be possible to create a saving, but it is not a simple calculation.

Nokia looked at Mobile Linux back in 2000 (and rejected it then) and I don't doubt they will continue to do so. They will almost certainly look at ways of taking advanatge of 'free' opensource as they have with the browser. Sony Ericsson by choosing UIQ have committed to Symbian at least for the next few years.

The second thing to consider is technical ability or fitness for purpose. Linux was not designed as a mobile OS. The kernel is not optimised for mobiel usage. It can be modified of course, but is that a good as solution.

Symbian's micro-kernel model versus the Linux kernel. Built to fit vs modifieid. Same applies higher up the stack too.

The importance of that is hard to judge of course. The technical issue is an important one and will become more so. Mobile is hardly likely to stand still. I wonder which is better equipped to move with it - Symbian or Linux.


And it is not just the kernel - there is a lot more to the software stack. This is the bit that creates a lto of value. This is not all going to be free via Linux. It will cost a lot to create and cost may come down to volumes. If Linux was homogeneous it might be able to match Nokia + Sony Ericsson's market share (43% ish), but its not.

I do find this is an interesting topic, but I find it frustrating that it often gets simplified to Linux is better because it is free. Its equally valid, in some ways, to say Symbian is better because it is technically better.

I do think Symbian being outside of Nokia / Sony Ericsson et al. may be an issue. Is it more expensive running it this way if you're a shareholder of Symbian? It depends how much synergy and expertise you see in Symbian... Moreover I don't think Symbian is unaware of this (e.g. see the license cost change if earlier this year). Indeed you might argue that Symbian, being owned by its major customers can afford to have less of a margin on its licensee fee than say TrollTech or ACCESS.

Admin said...

"If Linux was homogeneous it might be able to match Nokia + Sony Ericsson's market share (43% ish), but its not."

I might be missing your point, but are you implying that Symbian OS would be those 43%? Rather Symbian OS' market share is very small. Almost all Nokia and Sony Ericsson phones run in-house developed platforms on top of simple RTOS's.

Anonymous said...

Anders - yes sorry I should have been clearer there. What I meant by this is that companies who currently have 43% market share seem to be committed to Symbian. This is not to say all their phones use (or will) Symbian, just as not all NEC / Motorola / Panasonic etc phones use Linux.

There is of course the arguement that Motorola is going to use Linux in more of its phone than Nokia will use Symbian.

Actually it seems Linux is and will do best in the mid tier phone space (feature phones), but may have a harder time in high tier space.

I belive both Nokia and Motorola have talked about the 50% level. In anycase the future the portion of phones that run on an 'open' OS (Linux or Symbian) will be higher whatever the manufacturer - that much I think is certain. I think the 100m prediction has been used quite widely for open OS phone in 2007... that's a sizeable portion of the market (say 10%) and growing quickly.

So what I meant to say by this is that at least two manufacturers who represent a very significant portion of the market plan to use Symbian - homogenous at the OS level. The nearest to this will be Motorola (20%), but even they are using two differnt Linux implementations as I recall. I think this is worth bearing in mind when Symbian is written off (by some) where as Linux numbers are added to something much bigger by aggregation.

I do appreciate it is easy to get caught up in numbers (I am as guilty as anyone of this), but they do have implications for volumes and the impact on development costs per unit etc etc.

Anonymous said...

Michael,

As usual, your blog entries are the equivalent of strong paid research. Speaking of which, there's one point you didn't make: with Symbian divesting itself of UIQ, any new licensees are stuck going to either their #1 competitor or their #4 competitor for a UI layer... Yes, S-E and Nokia are locked in, but outside growth is locked out.

-avi

Admin said...

"If mobile Java can ever get to the point where "write once run anywhere" actually works, I'll be glad to consider it a platform"

I'd argue that you are a tiny bit unfair, as it equally could be said about the variants of Symbian OS, Windows Mobile, Palm OS and Linux, that are available on only a small fraction of phones, Linux being the worst among these in terms of fragmentation between variants.

That's also not WORA, however compatible e.g. Windows Mobile might be between phones that run specifically Windows Mobile.

I fully agree with you that the mistakes made with Java ME / MIDP - the biggest in my opinion is that Sun outsourced KVM development to several providers - must not be replicated for a future platform.

Jason Shao said...

In the US unless the mobile operators can move off the walled garden mentality I don't see a common platform evolving. It's counter to the goals of the operators. Afterall, American Cell companies are phone providers customers, not mobile consumers.

skierpage said...

I want and will even pay for phone apps like a music player, password storage, and a remote control.

But most of my application needs on a phone could be fulfilled by a browser running Flash and AJAX applications. Look at usage on a PC. I don't run a TV app or connect to my ISP's music store, I just browse a wide-open Web, watch Flash videos on sites like YouTube, occasionally save multimedia content locally, put my content on the Web, and increasingly run Web-based apps (blog, Writely, Google Calendar, Google Maps) in the browser. NONE of that usage is O.S. or browser-specific. I want to use my phones (Samsung sph-i500 Palm and Sanyo MM-9000, maybe a Treo or Nokia N93) the same way, but:
* the browser can't save content to memory card or upload from it
* the phone can't tell the browser my current location
* Flash doesn't run in the browser

So I'm forced to run local phone apps, I'm forced to sync data to them, and the cellular company tries and fails to sell me ludicrous "channels" for video, fee-based map programs, and a stone-age online photo service.

I don't know what the technical problems are. I hope that a phone O.S. + browser like Opera Mini or the Mozilla-based Minimo solves them and gives me the NORMAL browser-based usage that I employ for 80% of my computing.

Cheers!

Anonymous said...

I heard that UIQ is out of the picture now. Linux and BSD are just creeping in by stealth (except iPhone OS).