Building Connected TV Apps
I'm Roux Joubert, Head of the TV & Mobile Platforms team based in the ´óÏó´«Ã½'s new offices in MediaCityUK, Salford.
I am privileged to lead an amazing team responsible for building some cutting edge and award winning products, who brought you the ´óÏó´«Ã½ iPlayer on TV and mobile, as well as News, Sport and Red Button on emerging TV and mobile platforms.
Today the ´óÏó´«Ã½ iPlayer, News and Sport apps are available on an astonishing 650 , from internet-enabled or Smart TVs and set-top-boxes to media players and games consoles, delivering more than 45 million videos to 2 million users every month. Most recently, the ´óÏó´«Ã½ Sport app has been used by more than 200,000 users a day to watch the phenomenal London 2012 Olympic Games coverage on connected TVs alone, having only launched a few short weeks before.
While this is a remarkable achievement in itself, it certainly wasn't easy or straightforward, and I would like to share with you what challenges we have encountered, what we have learnt in the process of solving them and what we believe is important to consider for anyone looking at building applications for connected TVs.
The Challenge of Massive Fragmentation
From the moment ´óÏó´«Ã½ iPlayer first appeared on a TV for Virgin cable users back in 2008, it was clear that deploying the ´óÏó´«Ã½ apps on connected TVs wasn't going to be simple. It has actually been a tremendous challenge to build multiple applications, in different presentation technologies, with little or no standardisation, and across such a wide range of devices.
Some have TV tuners, some don't. Some work with remote controls, others with pointers or even voice and gesture controls. Some run HTML 3 and CSS 1, others HTML 5 and CSS 3, still others FlashLite 3.1 or Adobe AIR 3.0. Every device seems to have its own way of playing back video, and many devices have memory constraints we haven't seen on a PC or mobile phone for years, with some having to cope with as little as 1MB!
We're delighted when our audience doesn't even notice the hoops we've jumped through to deliver a graceful experience across so many devices, but the engineers among us know what a mammoth task this has been.
A big part of the challenge is perhaps surprisingly due to the emergence of browsers on connected TVs. Earlier versions of the ´óÏó´«Ã½ iPlayer were built in Adobe Flash or MHEG-IC ( is an ISO and ETSI ratified interactive TV standard used in the UK). But over the last two years, the vast majority of devices we have seen have shifted to using HTML and JavaScript ( standards), bringing browsers to the TV.
Now, while W3C standards include generally well understood technologies like HTML, JavaScript and CSS, building for a browser was initially the most difficult platform to engineer at scale. The problem is this: Flash and MHEG are either proprietary (the former) or a well-defined and mature industry standard (the latter), which should make it relatively simple to build an app once and deploy it to many devices. The HTML landscape, however, consists of a variety of standards rather than a single one, with many of those standards in varying stages of development. Two device manufacturers have yet to build their browsers and HTML support in quite the same way. For those familiar with building websites, imagine building a site that needs to support hundreds of different browsers, not the three or four we're used to on desktop sites today.
Scalable Architectures
The way to tackle this challenge is by approaching it from two sides: firstly by having a defined and accepted industry standard for browsers on TVs, and secondly by implementing a scalable architecture to simplify app development and deployment across the many different devices.
The first problem is that there is no widely adopted industry standard for browsers on connected TVs today. Most manufacturers do not follow any particular standard, although there are a few contenders for a new standard that are promising, such as the DBook 7 and HbbTV. The emerging DBook 7 specification - published by the (DTG) - is aimed at the UK market and supports the ´óÏó´«Ã½ interactive technologies used in Red Button. is backed primarily by European broadcasters and is starting to get some good traction in the industry. The ´óÏó´«Ã½ will continue to feed into these specifications and provide compliant applications to help make the case for greater alignment and adoption of standards.
But it can take a long time for a standard to be fully adopted. In the meantime, hundreds of new devices flood the market every year, and we want to make sure that ´óÏó´«Ã½ applications are available on as many of these as possible. The solution lies in decoupling the applications from the underlying device complexities.
The TV Application Layer (TAL)
This is done by creating an abstraction layer between the applications (iPlayer, News and Sport), and the devices they are intended for (consoles, TVs, media players). This abstraction layer, which we call the TV Application Layer (or TAL), takes care of all differences between devices, such as remote control key-mapping, media player interfaces, performance, networking and storage.
This means application developers do not have to be aware of all the specifics and idiosyncrasies of each device, and can build to a common, well defined interface, safe in the knowledge that all the devices supported by that version of the TAL will be easily addressable. Likewise, when a manufacturer adds a new device to the list of ´óÏó´«Ã½ certified devices, all the ´óÏó´«Ã½ applications will become available without having to write new versions of each app.
This approach was validated again most recently when the ´óÏó´«Ã½ Sport application first made its appearance on a number of devices in time for the London 2012 Olympic Games, but using only a single application, integrated with the TAL. This proved that we had found an elegant solution to a very tough challenge, and one which we believe is the only way application development on connected TVs can be done at scale for the foreseeable future.
Better Standards and More Content
We hope to make our TV Application Layer available to content providers and manufacturers in the near future, and in the process help stimulate the market for connected TV applications at a time when its full potential is only beginning to be realised. In the meantime we are continuing to develop and use the TAL internally to improve the way we build apps and deploy them onto platforms, and working closely with manufacturers to improve our tools and processes.
But creating new abstraction layers to get around device fragmentation and lack of standards can only be a short term solution. To achieve real scale for TV-based applications, manufacturers need to adopt a common standard for TVs based on HTML, JavaScript and CSS, and content providers need to get familiar with the challenges of building apps on TVs and, to give the manufacturers more incentive to solve this problem. The ´óÏó´«Ã½ will continue to play a major role in helping to define the direction the industry is taking, as well as providing valuable insight into what makes for a good TV-based experience.
Simply Building Apps
We started off a few years ago thinking it would be easy to build apps for connected TV. We quickly realised that it wasn't, but also that we had an opportunity to make it a lot simpler for ourselves and the rest of the world to do so, while also helping the industry understand what content providers want. I believe we are making great strides in achieving both. So, the next time you watch ´óÏó´«Ã½ programmes on a games console, smart TV, set-top-box or media player, whether it be through ´óÏó´«Ã½ iPlayer, News, Sport or future Red Button services, spare a thought for the work of the ´óÏó´«Ã½'s TV application teams.
We've faced these challenges and made it possible for you to enjoy the ´óÏó´«Ã½'s fantastic programmes on your sofa in the living room on whatever device you choose, and hopefully on even more devices in the future.
I hope to build on this topic as we deploy our products onto more platforms, but in the meantime I'd be very interested in your comments on how we have built the ´óÏó´«Ã½'s connected TV apps. Let me know what you think.
Roux Joubert is Head of TV & Mobile Platforms, Programmes and On Demand, ´óÏó´«Ã½ Future Media
Comment number 1.
At 28th Sep 2012, fr wrote:Standardisation is urgently needed, currently smart/connected TVs seem to become obsolete far too quickly. For example I have a Sony NX713 which was only released in the second half of 2010 and already seems pretty much abandoned with iPlayer stuck with the old basic interface and no HD content, no ´óÏó´«Ã½ News or Sports apps, and no other significant additions or improvements to the available UK content since it was purchased.
Complain about this comment (Comment number 1)
Comment number 2.
At 29th Sep 2012, mrango wrote:If MHEG is a "mature industry standard" and HTML "consists of a variety of standards rather than a single one, with many of those standards in varying stages of development", why is there such an emphasis placed on developing for HTML for TVs?
It seems the ubiquitous and highly conformant MHEG is largely ignored - see Olympic coverage.
Complain about this comment (Comment number 2)
Comment number 3.
At 29th Sep 2012, Craig Smith wrote:Roux, came across an interesting article on NewsView about how - for me the most exciting aspect of this is what you and your team are working on... connected apps. There is no area where the benefits of tech convergence is more evident.
A similar challenge of fragmentation you mention was there on the mobile platform many years back... made it quite a nightmare for both content providers and app developers for mobiles. Convergence to some standard I believe is inevitable as some dominant platform always emerges. I think the greater challenge lies in design for useability and quick take-up.
Complain about this comment (Comment number 3)
Comment number 4.
At 30th Sep 2012, spych102 wrote:1. Stop supporting crappy, slow boxes.
2. Invest in a decent high quality Android app.
3. Start supporting set top boxes that run Android apps. Android apps are the new standard.
Complain about this comment (Comment number 4)
Comment number 5.
At 1st Oct 2012, steve smith wrote:Thank you for this. I am writing a market report on Smart TVs. One section is about standards, and it's great to hear some first hand experience about some of the frustrations arising from this.
Complain about this comment (Comment number 5)
Comment number 6.
At 1st Oct 2012, ZeShaan wrote:Some interesting comments, if you want to be part of the solution, Roux is recruiting for his team: bbc.in/SwUxQ8
Complain about this comment (Comment number 6)
Comment number 7.
At 1st Oct 2012, eConundrum wrote:I have to say why I've heavily criticised your mobile platform support I think you've done a excellent job of coping with the much more fragmented Smart TV platforms.
The ´óÏó´«Ã½ iPlayer app on my Samsung Smart TV works really well and is becoming a fairly common way my wife and I watch your programmes.
I wouldn't be supprised if Streaming didn't replace broadcast TV in the UK entirley within a couple of decades.
Complain about this comment (Comment number 7)
Comment number 8.
At 1st Oct 2012, eConundrum wrote:It's interesting that the approach you went for mirrors microsoft direct X abstraction in many ways. It's a good idea, but relies on the implementation of the TAL layer being consistant on all devices. This is often an issue with graphics cards on direct X slightly different driver implementations of a standard feature often means an application that runs on one card will fail on another.
I assume at the moment you create the TAL driver as it where for each device, but if manufacturers take over this aspect themselves how will you drive strict complience?
Complain about this comment (Comment number 8)
Comment number 9.
At 1st Oct 2012, Paul Jakma wrote:I think the ´óÏó´«Ã½s' current path, where it is leading the way in building future TV systems from which small manufacturers and hobbyists are excluded, systems which likely will rob the end-user of their fair-dealing rights in copyright law, runs counter to the ´óÏó´«Ã½s' public service remit.
I'd expand on that, but if I do I suspect this post will be censored. Instead, I've written a blog "Dear geek, the ´óÏó´«Ã½ is not your friend". See: . Thanks.
Complain about this comment (Comment number 9)
Comment number 10.
At 2nd Oct 2012, Roux Joubert wrote:Thanks for all the comments so far - I'll try to respond to some here.
@fr - I couldn't agree more. Some early connected TVs have very limited capabilities making it very hard to build applications that would work reliably on them, many of which require at least a firmware update from the manufacturers to get it up to a workable level. Fortunately it seems that this is gradually improving as the manufacturers on new devices are using better software and browsers as well as getting used to providing a better level of on-going support over the life cycle of the device.
@Craig Smith - Thanks for your link - a compelling read. I too expect the connected TV space to follow mobile browsers in the way it has matured over time. I am just too impatient I suppose! I also happen to agree that usability (and interface design) is a massive challenge. Just try to navigate your way around any connected TV's internet portal and see for yourself.
@eConundrum - Very glad you are enjoying the ´óÏó´«Ã½ iPlayer on your telly! I think the difference between DirectX and what we are doing with the TAL is that we set out to make life a little simpler for ourselves, without mandating or requiring manufacturers to adopt any proprietary middleware or driver. Instead, we expect them to comply with a fairly pragmatic and reasonable requirements specification that most of them are implementing anyway. By sticking to a specification we are keeping things simple for both content provider and manufacturer.
Complain about this comment (Comment number 10)
Comment number 11.
At 3rd Oct 2012, eConundrum wrote:Thanks for the reply, I suspect you specification is rather simple compared to an abstraction layer like direct X. Be interesting to see how widley adopted it becomes.
Paul Jakma,
I've read your blog and while I do agree with most of what you say I think it somewhat inevitable the ´óÏó´«Ã½ and other large media players would attempt to go in this direction. Being as the future of media generally is probably streaming not broadcast they are bound to want control.
I do agree with what you say about current stream security they are well aware of the issue, and do take the approach of trying to avoid discussion of it to try and lower public awareness.
Complain about this comment (Comment number 11)
Comment number 12.
At 4th Oct 2012, Paul Jakma wrote:eConundrum: I don't think it's inevitable. I've had this discussion a lot, with a number of people. Including with ´óÏó´«Ã½ people (some senior).
I respect their views, and I believe they have made their decisions in all sincerity and with the best of intentions. However I disagree with their conclusions. I disagree that the reasoning they have given logically leads to that conclusion. I believe it only does if certain aspects of reality are ignored AND if the contracts they cite state what they say they state (I am not accusing the ´óÏó´«Ã½ of lying, rather I suspect many of the ´óÏó´«Ã½ people justifying these decisions may only have 2nd hand knowledge of those documents themselves - they may be mistaken in various ways).
Finally, even if their argument were logical and rational as a matter of short-term pragmatism, the contracts as they say and non-negotiable, I believe that, as a subjective matter, they have grossly under-weighted the importance of the public interest in fair dealing and in having a competitive IPTV market. The ´óÏó´«Ã½ Trust didn't agree with me on that argument though, it seems, in their ondemand strategy review.
My feeling is that the ´óÏó´«Ã½ could easily have chosen that the public interest demanded a non-discriminatory policy for IPTV/ondemand interface access. My feeling is that there is evidence and other reason to believe the content makers would have gone along with that: a) The ´óÏó´«Ã½ is a big player in UK commissioning (including having significant ownership) b) Other rights-holders have gone along with no-DRM in other places.
Why the ´óÏó´«Ã½ did not, I do know. Why the ´óÏó´«Ã½ has gone even deeper into building a heavily restricted IPTV market, I find hard to fathom. The only explanation I can think of is that, despite the good intentions of the people who work there, the ´óÏó´«Ã½ has become institutionally captured by private media interests. Perhaps because people working at the ´óÏó´«Ã½ may move between private media and the ´óÏó´«Ã½, and/or perhaps because the ´óÏó´«Ã½ has a commercial interest in private media companies. (Including now YouView, a maker of IPTV devices - being able to control new entrants, and block any potentially market-disruptive newcomers is the dream of many, and the ´óÏó´«Ã½ has this power).
Complain about this comment (Comment number 12)
Comment number 13.
At 8th Nov 2012, Robert Stenson wrote:Is there any reason that my mac OS Lion can't download ´óÏó´«Ã½ iplayer Desktop? Love to able to download programmes to look at them in Spain, my second home. I can't get iplayer there. The ´óÏó´«Ã½ should give each TV License holder a number so that they can put in to their PC and watch the ´óÏó´«Ã½ away from the U.K.
Complain about this comment (Comment number 13)
Comment number 14.
At 2nd Dec 2012, eturk001 wrote:to ´óÏó´«Ã½ & Roux:
Please update your compatibility list for the Samsung TV ´óÏó´«Ã½ app. There's a new TV model UNxxES models of 2012. The ES lines has been out most of the year and you're not supporting them, according to your list: TV
Complain about this comment (Comment number 14)