YouView: the future of British TV or another Domesday Project?
In which Telco 2.0 reviews the YouView specifications
Delegates at the recent T2 events would have seen a fascinating couple of sessions on future TV. Will we (as LG suggested) have multi-screen TVs, with screens dedicated to high definition video, meta-data, and to social content? Or will the role of "social TV" be fulfilled by an independent "companion device", as former YouView CTO Anthony Rose suggested?
[Ed: We'll be discussing this and other issues at our forthcoming Digital Entertainment 2.0 events. For more Telco 2.0 thinking on this, try our recent YouTube: Recent Improvements Change the Game note and our new Analyst's Note on UltraViolet,the content industry's answer to iCloud and Amazon Cloud Drive]
On the other hand, as we've been saying for years, the move of mass-audience TV onto the Internet is constantly testing the technologies and business models involved. Even if the growth rates are not as ferocious as first predicted, they are still higher than overall traffic and the content itself is getting higher quality. How will we push all those packets?
In the UK, whenever there has been a technology transition in broadcasting, there has always been one institution that has acted as a leader - the BBC. Its great rival in this has to be BSkyB - think of Sky+, Sky HD, and Sky 3D. On the BBC's side, there's decades of work in the core trades of TV, developing the basic infrastructure, pioneering TV on the Web with the iPlayer, and some interesting side projects like the BBC Micro. The two of them have contrasting and perhaps complementary specialities - as the pay-TV challenger, Sky is fascinated by adding more features to TV, while the BBC as a public service is all about infrastructure and universal reach. When a little-known Racal division called Vodafone needed a radio planner to build their GSM network, they poached John Causebrook straight out of BBC Research.
After much tortuous negotiating with the regulators and the other TV stations, they have finally got a specification out for a common platform for the next generation of STBs, YouView. If it gets deployed, it will shape the future - so is it any good?
Telco 2.0 recently had to make a long train journey, so we grabbed the 229-page technical specification and got stuck into it.
The Good Points
This spec is basically sensible and that's always preferable to the alternatives. It describes a low-power, media-optimised Linux device based on well understood, widely used open-source software.
Best of all, the spec is sound on networking. YouView intends to tackle the problem of dumping enormous quantities of video on the UK's creaky copper-line infrastructure in the following ways:
- CDN the hell out of it
- Adapt to the limitations
- Shave the peaks
- Integrate broadcast and broadband
- Multicast everything
We already knew, of course, that a big part of YouView would be BT's spanking new content delivery network, Content Connect. CDNing is a well-tried and reliable way to deliver video without blowing things up. So the point at which the firehose of telly pouring out of the BBC gets turned on the ISPs is pushed well towards the end user.
Perhaps that's not enough, though. This is one of the reasons why an intelligent STB is useful in itself. YouView intends to obviate the idea of throttling or deep-packet inspecting streams of video by managing them itself from the network edge. The specification requires that the video bitrate adapt to the available bandwidth - the BBC iPlayer client already does this and manages to fit live streaming Rugby League onto the 400 or so Kbps Telco 2.0 enjoys in the wilds of remotest north London.
Self-managing the flow of video is a theme in the whole project. The specification requires that YouView devices provide several edge-defined traffic classes - the highest being for live streaming, the lowest being a scavenger-class for downloads that uses all bandwidth left over after other flows are satisfied and gets out of the way otherwise.
The mention of downloads should remind us that a big part of dealing with online video distribution is what we described as peak-shaving, borrowing a term from the electricity grid. Much more load can be handled if some of it is shifted from the peak hour to the rest of the day when spare capacity exists. This is especially important as wholesale bandwidth is typically charged for by peak utilisation (the famous 95th percentile). YouView intends to queue much of the heavy content ahead of time and stash it on the device's local hard disk.
The spec also foresees the integration of broadcast and broadband - a substantial chunk of the disk is reserved for content pushed out by broadcasters and stored locally. Clearly, this is one way of keeping Coronation Street off the Internet and on the airwaves, where it belongs.
Finally, the specification requires IP multicast to be on by default. BT has recently announced that at long last their backbone network is going to be multicast enabled, and in fact they're incorporating it into Content Connect. Multicasting essentially makes every router into a CDN node, drastically reducing the volume of data that has to be moved over the whole end-to-end path. Somebody has to originate the multicast streams and count the viewers, and BT is going to be offering this as a wholesale service.
Moving on from the packet pushing side - and we find everything seems a little dull after you leave the networking stuff behind - YouView also foresees that there will be apps for the system. Being a full-featured Linux device based on consensus technology, there will be no shortage of potential developers for it, especially as its web browser is WebKit. But we aren't completely sure about the distribution model.
The Worse Points
It's not surprising that a specification drawn up by TV channels would tend to give TV channels a special role. YouView foresees three kinds of apps.
- Core YouView
- Platform Provider
- Content Provider
The core YouView system is what it says on the tin, plus any applications the device manufacturer wants to include. The Platform Provider is the party responsible for YouView itself, for the electronic programming guide, for a variety of crucial approval and cryptographic signing processes, probably for the broadcast infrastructure, and is very likely to be the BBC. Among other things, the PP signs Content Providers' keys that allow them to publish software that gets installed on the YouView STB. It's possible that a pioneer device maker might take on the role of PP for their own devices, but the specification gives the strong impression that there will be only one.
So, to a first approximation, you've got to be a TV station to get your app in the app store. Further, this quote from the spec should stand in its own right:
Chapter VI Consumer Device Software Management. This chapter is not relevant to application developers
Really? How could the process of getting your application out to the public and onto their machines not be relevant to application developers?
Obviously, there's a possibility that one of the TV stations would decide to set up an app store that would facilitate the process. Our money would be on Channel 4 - they already invest in a variety of digital projects including ScraperWiki.
But there's another problem here. YouView hopes to be the centre of all your TV and video content, whether it's sideloaded from DVDs, downloaded or streamed from the Internet, or broadcast over the air and stored locally. But the fraction of this content that comes from a traditional TV station, however it gets to you, is falling all the time. Can YouTube be a YouView channel? Would it get code-signing rights?
There are ways round this. One, described above, is that a content provider would act as a proxy for independent developers. Another would be to do it all in the browser - the specification states that an AppPlayer object (the base type for a running application) can be a web browser. As we mentioned above, the browser is WebKit, which should provide plenty of scope, especially as a lot of web developers are intimately familiar with it from the iPhone.
More fundamentally, the user interface paradigm is very, very TV-centric. You wouldn't expect anything else from TV channels, of course. But the specification insists that the user interacts with the system through a classic TV remote control handset. It's very explicit about this.
Obviously, support for the remote is a must - there are public service obligations to think about and a lot of deep expectations and legacy kit to deal with. But what if you wanted to control the system - which gives you a lot more options and far richer forms of interaction than a TV remote - from your laptop or mobile device?
Even more importantly, what if you wanted to use your Kinect or Nintendo Wii or the accelerometer in your iPhone to do something genuinely exciting with it? All sorts of things are happening in this line right now. Even Microsoft has managed to make its peace with the Kinect hackers. And even Nokia's doing interesting stuff with TV, video, and gesture interfaces...
But if you want to do it with YouView, you're missing an API that lets you drive the user interface with something other than the classic IR remote you lose down the back of the sofa. Why isn't there a Web service API for this? In fact, why isn't there a user script API, the equivalent of Firefox add-ons for television?
Finally, we have some concerns about the networking side. When we've thought about broadcast-broadband integration, we've always worked on the principle that as much stuff as possible should travel over the broadcast system. Broadcast, whether satellite or terrestrial, does one thing well - moving hit content in high quality, cheaply.
But we're struggling to see the use of the Push-VOD over IP functionality in YouView. The TV infrastructure broadcasts a list of available programmes, and the user device requests something from the list over the Internet. Isn't this upside down? The metadata, a text file, is travelling on the broadcast link, while hundreds of megabytes of video are travelling over the Internet. It's worth remembering that Virgin Media never bothered with Push VOD and Sky has tried it and changed their mind in more recent iterations of Sky+ - also, YouView foresees that the device might automatically pick out content predictively, based on observed usage or on user-specified criteria, which doesn't make any sense at all with Push VOD.
However, in the light of the BT multicast announcement, we may have been too dogmatic about this. Apparently, BT originally didn't do multicast because they were primarily thinking in terms of supporting their BT Vision IPTV product, and they hoped to move much of the video involved over the UK DTT broadcast infrastructure. Now, BT has got a lot more live streaming video to shift - they've succeeded in getting wholesale access to Sky's football coverage. They've also got to think in terms of YouView and in terms of preparation for the 2012 Olympics and associated Global Packet Pushers' Pentathlon. (London 2012 is likely to be a key driver of YouView.)
And it seems that they've decided that the additional DTT multiplex capacity would cost more than the IP multicast. This is an odd blow-back from the mobile video traffic boom - people started really using mobile data, the carriers lobbied frantically for more spectrum, and they won the lobbying war and got the additional spectrum ahead of TV. You pays your money, and you takes your choice, but it's worth pointing that the original multicast concept foresaw global content distribution and it provides for some interesting future possibilities.
Further, as the spec stands, it doesn't let you cache anything that was delivered over an SSL-encrypted link due to DRM issues. It is likely that most things on the Web will be SSL before long. This is a bug.
Telco 2.0 Final Thoughts...
Our conclusion is, essentially, "Good - as far as it goes". YouView has obviously benefited from the work of network engineers with Internet clue, and is a fine and regrettably rare example of the content industry not finding ways to make everything more complicated. It represents an intelligent response to the online video challenge, and incorporates many of the approaches that we recommended in 2007-2008 as the BBC iPlayer revved up to disrupt the UK Internet.
It's another reminder that, if you're looking for the "British Google", it's BBC Internet Services. However, there's another, darker story here - that of a whole lineage of interesting, elegant British tech projects that didn't really happen. The classic case is the BBC's Domesday Project, a massive crowdsourcing exercise from the 1980s that eventually just ended up in a filing cabinet at Broadcasting House. More recently, another BBC idea, Project Kangaroo, hoped to create a public archive of the UK's TV history but failed miserably amid rancorous disputes between the partners.
As the surprisingly close choice between more broadcast and IP multicast shows, it really is a question of crafting the right digital logistics package for the job. YouView provides a lot of useful options
However, whether any of this actually helps depends on end user adoption and the quality of the products supporting YouView as a standard. And we're seriously concerned that the TV-centric thinking behind YouView is going to cripple it as a platform for innovation. For example, we doubt that the notion of a TV "channel" will mean much, and especially that anyone will get software from one.
It's also going to be critical to get the content providers of the future in there. YouTube is obvious (again, are they going to become a UK TV station?) but there's Ultraviolet and Lovefilm to think about.
Also, will people really interact with a future "social", "connected", or "smart" TV through the classic, nine button remote control? For the first time, the percentage of households with a TV at all is shrinking. This is a critical issue - we don't actually know if future TV will be delivered by something that even looks like a TV. Clearly there's going to be a big display of some sort, but beyond that it's anyone's guess. Baking the remote into the system tends to enforce a TV-like design and a TV-like interaction pattern.
The danger is that YouView is too much like TheyView - a project inextricably linked to a media model with not much of a future, doomed to join Concorde, the Domesday Project, and much else in the halls of the Science Museum.
It's not foredoomed - it's not certain. But there is a serious risk that it will go that way.