Views on the new media market

Flash? Give it a REST

With the recent Apple v Adobe spat over Flash, and the associated mudslinging as to who really is "open" and "web-centric", I thought it might be interesting to share some experiences in actually using Flash to build a front-end to a REST web service - in this case the Live Talkback servers.

Now a REST API is pretty much the definition of the "native Web" way of doing things, and everyone from Amazon to Yahoo via Google and Flickr have REST-style APIs these days.  So you'd think that it would be a snap to use Flash to create a beautiful front end talking to those APIs.

You'd be wrong.

It turns out that getting Flash to talk to a REST API is surprisingly difficult, and ended up with us having to create several Flash-specific hacks in the server.  Not a pretty sight at all.

The problems

• HTTP Methods:  To do REST, you need to be able to send GET, POST, PUT and DELETE HTTP requests.  Surely simple - any competent HTTP library has had method support for ever.  Flash has the URLLoader class which should do the trick. Except that it doesn't support anything other than GET or POST when running in the browser.  Fail.
• Go low-level: There's a workaround - use Flash sockets.  This can be made to work, but runs into the morass of socket security policies served on port 843.  Which is a port that lots and lots of places block. Fail
• Overload: OK, how about overloading POST?  This is a common pattern to deal with libraries that can't do PUT/DELETE. So stick a new header in e.g. X-Method-Override: PUT, and then on the server side treat the POST as a PUT.  Ah, but what's this?  Flash refuses to put in new headers.
• Overload in the URL: OK, this can be made to work- http://www.example.com/resource?method=PUT.  It's more than horrible, but hey - that's where Flash leaves you.
• Session management:  Like most web services, we use session ids that are stored in cookies.  Oh, but what's this?  Flash refuses to send cookies.  Fail again.

Can all these be worked around?  Yes - and we've done it for the Live Talkback.  But it rather takes the shine off Flash as a "web-centric" platform.  Next time, we're going to be using straight AJAX/HTML.

Symbian's (and Nokia's) structural handicap

Hi, I'm Malcolm Box, CTO at Live Talkback and new guest blogger here.  Thanks to Matt for a chance to share his soapbox, and thanks to you all for reading.  As ever, comments, brickbats and bouquets are all welcome.  Well, OK, not the brickbats so much.  Be gentle!  With that said, on with the post!

There's been a bunch of commentary recently around the Nokia N8 and whether it has what it takes to lead Nokia's recovery in the high end device market. There's also been questions as to whether the Symbian platform has the "horsepower" for the devices of today.  Leaving aside the question of how many horses it takes to run a phone, I think this question misses a fundamental point about the economics of Nokia/Symbian that will have a more profound impact on the future than any technical "horsepower" argument.

Symbian has a persistent structural disadvantage that could prove fatal.  Even if the N8 is a winner.

Today, Symbian is competing against 3 major platforms for creating smartphones: Apple's iOS (~16% market share), Google's Android (~10%) and RIM's Blackberry OS (~20%).  It's instructive to consider what a change this represents.  For most of the decade I spent in Symbian, the focus was on beating the proprietary real-time OSes that powered (and continue to power) the vast majority of phones sold.  These were OSes like Nokia's S40, Ericsson's OSE and Motorola's P2K etc. No other smartphone platform posed a serious challenge. Today the battle for the high end is entirely between these 4 platforms - and Symbian/Nokia are losing ground.

Compared to these 3 platforms, Symbian OS possesses significant technical disadvantages that make it harder for the platform to compete in the crucial race to bring out exciting new products.  This handicap acts at two levels: it's harder to build any individual feature and more expensive to build the platform as a whole.

First let's consider why it's more expensive to bring out a new feature on Symbian OS. If you're building a new feature, then the cost/effort is a function of the language, libraries, tools, and documentation for the platform.

Platform	Language	Tools	Documentation
Symbian	C++	Eclipse	OK
iPhone	Objective-C	XCode	Good
Blackberry	Java	Eclipse	OK
Android	Java	Eclipse	Good

At first glance this doesn't look too bad - everyone except Apple's using Eclipse, and the documentation is generally OK. There's 3 different languages in play, but surely that doesn't make such a big difference. Except that it does. One of the overwhelming lessons of the last 10 years is that the languages and libraries used for development make a massive difference to what is achievable. C++ takes 2-3 times longer to deliver the same functionality as in Java. Worse, the Symbian dialect of C++ is non-standard and lacks support for basic parts of the language. On top of this, the Symbian libraries are idiosyncratic to say the least, leading to longer learning curves and a need to re-invent things that are basic parts of other languages. Then look more closely at the tooling - the IDE and emulators. Symbian's comes last on these as well. Debugging takes twice as long in Symbian OS as in Android.  Adding up these factors, Symbian's development productivity is anything from 5-10x lower than Android or iPhone.

It gets worse for Nokia. Once we look at the overall costs of creating the platform and the devices that run on it, Nokia suffers further disadvantages. The cost of platform development to the "lead" player (Nokia, Apple, RIM, Google) depends on how much they have to build, how much they can reuse/('buy') from elsewhere and how much they can persuade the community to develop.

Platform	Build	Buy/Reuse	Community
Apple iOS	~100%	Some - Mac OS, BSD, Unix	None - Closed source
RIM Blackberry	~100%	Some - Java libs	None - closed source
Nokia Symbian	>75%	Low	Limited
Google Android	< 50%	High - Linux, OSS, Java	High

The clear winner here is Google - large chunks of the Android platform are simply reused from other OSS projecs, so that Google has to build and maintain less than 50% of Android. In addition, they've got a successful community contributing around Android, further lowering their costs.

RIM and Apple lose out with their proprietary stacks, although both achieve some reuse from adjacent communities, and in Apple's case there is substantial reuse from the Mac OS X products.

Nokia are moving in the right direction with the Open-Sourcing of Symbian, but are stuck with a platform which has very low levels of reuse - everything from the kernel, to the networking and multimedia are Symbian-specific. The idiosyncrasies of the Symbian platform make it hard to port code from other platforms, further reducing the amount of reuse that's possible. They have also not yet built a community around the platform that is contributing substantial value.

The net result is that Symbian has the cost structures of a proprietary platform with the added costs of being open source - without reaping many of the benefits.

So where does this leave Nokia and Symbian? Nokia is saddled with a structurally higher cost of development which is only partially offset by their undoubted prowess at manufacturing, sourcing and logistics. Symbian as a platform looks increasingly unattractive to anyone not already deeply committed (Nokia + some Japanese manufacturers), further calling into question the ability to generate a community.

Nokia have made some bold moves over the last year to address some of Symbian's shortcomings, from porting Qt to open-sourcing the platform. However, in the words of the old joke, if you want to get to where Android is, you wouldn't start from here.

Nokia's fightback: wielding a swiss army knife against a lightsabre

I enjoyed reading Anssi Vanjoki's blog post on Nokia's fightback. Nokia does need to fight back - once they were the kings of the mobile phone business - the sort of company where senior executives made world headlines by receiving the largest ever speeding fine

Since then things have gone down hill, in fact if we look at stock price performance from January 2002 (when Anssi got his record fine) Nokia's share price has dropped by 64% - over the same period the Dow Jones index is down just 5%, and Nokia have even underperformed Motorola (who are only down 55% in that time)

It's not just financial performance - consumer perception amongst the mass market for smartphones in Europe has shifted to Apple being the premium brand, technical early adopters are keen on Android devices while competition in the feature phone space from Samsung and LG is very strong - IDC are predicting Nokia will lose their number 1 status in western europe in 2010. This pattern is repeated at a consumer level too - I attended an event yesterday, and for the first time ever noticed that there was not a single Nokia device in the room. 4 years ago this would have been unthinkable - it would have been Nokia phones + corporate provided blackberries. Today it's more blackberries, iPhones and the odd Android phone - a spectacular fall from grace for the once trusted king of the pocket.

So what are Nokia doing about this?In 13 paragraphs Anssi sets out his vision for bringing Nokia back to a leadership position in smartphones.

7 of these paragraphs - that's half of the post discusses Symbian and MeeGo. In just one paragraph does Anssi address the consumer - he says

For consumers, it will mean true computing power in your pocket. Something that can deliver everything you want, but be with you all the time.

As a consumer I wonder what does that mean? What can I _do_ with these devices? In what way will a Nokia device be better, cheaper, faster, longer lasting than a competitor's device?

Nokia seems to have fallen for the engineer's trap of adding features to the platform because they can - rather than focussing on what consumers can do, and repeatedly asking the question "are we doing the best that we possibly could here?"

A year ago I set out a roadmap for Nokia on this blog - I challenged them to create three areas of focus where Nokia have a strategic opportunity to WIN.

Nokia continue to execute around these areas - Text messaging is led by the E series devices, Web is led by the N series devices and Mass market are led by their mobile phones and Ovi services.

As I measure their progress 1 year on, they seem to be falling behind in two of the three areas - the only area that they are being successful in is the Mass market.

To fight back Nokia needs to be brave - and to resist the temptation to add too many features, creating products that are swiss army knives - lots of small, useful tools, when the competitors are wielding specialised, focussed modern that look like lightsabres - you will lose the fight.

Could Nokia recover? Sure - they have the scale, marketing expertise and R&D capacity to win back their position.

Will Nokia recover? I'm less confident here - a cultural change is needed, and they need leadership that is prepared to knock heads together, and upset people, with a clear vision that can be communicated to consumers. This means taking greater risks with their core business, in the recognition that continuing to do what has got them into the position they are today, won't get them out - they need new focus, and clear execution.

Good luck to my Finnish friends.