The IP Development Network

Welcome to The IP Development Network Blog

Thursday, 26 July 2007



It's not even a month since the last i launch, but tomorrow sees the launch of another service that could disrupt its industry to an even greater degree than Apple promises to do with mobile telecoms. This time though, thankfully, we won't have to pay the homeless to wait in line for us to get hold of it.

The BBC launches the iPlayer tomorrow, but unlike the iPhone launch where all you could find was praise and hype, the BBC faces nothing but criticism, doomsday scenarios and even calls for a ban on the eve of it's big announcement. No wonder the folks behind it have decided to find pastures new.

The problem is that the BBC is publicly funded. It gets its money from everyone in the UK with a TV set because we all need a license to own a TV. The BBC's license revenue comes in exchange for a responsibility to deliver a universal service, free of advertising to anyone who pays the license fee. Foreign readers may find this curiously eccentric in the 21st Century, but the BBC is a national institution and we are British so that's the kind of thing we do.

This is where the problems lie. The license fee was designed at a time when the BBC was broadcasting: it had no competition in 1922 when the license was introduced to cover radio. The TV + Radio license was introduced in 1946. The Sky empire was still just a twinkle in the eye of James Murdoch's grandfather at that time.

The company (if you can call it that) is now operating in a very different world, but for many reasons (most of them sentimental), the BBC is still funded this way. As a result, it competes with other TV channels (and web sites) on an unequal footing because their funding model does not expose them to market forces.

Because the BBC is publicly funded, it has been free of the commercial pressures that competitors face on a daily basis. Has this given it an unfair advantage...? How many R&D departments would be given 4 years and £3m to deliver a project? Surely, anyone else in the same position would have lost the faith of shareholders well before now and management would be history. The BBC's unique position has shielded the iPlayer and given it breathing space in which to develop the service.

On the other hand though, how many R&D departments would face an Ofcom Market Impact Assessment, a Public Value Assessment, a full review by the BBC Trust and scrutiny by parliament before it could launch? The kerfuffle about the lack of service on Macs and Vista - there is a petition with 11,000 signatures with Downing Street asking the PM to ban it - is frankly pathetic. Do people really expect the BBC to be able to launch the service working 100% and available to everyone on day 1 with no testing?!?

Anyone who has ever been involved in product management will know that this is a recipe for disaster. The BBC cannot eat the elephant in one bite, but because of its funding model it will be forced (they might say "easily persuaded") to deal with standards issues like no other entity. The elephant will be consumed.

The Mac and Vista options might be addressed by making the content available through other media players as long DRM issues can be resolved. I suggested in my LUI Part 6 piece, where we described a prototype of the future of IPTV, these players are likely to include the likes of Joost. Because of its universal service obligation, the BBC is not in a position to say no.

The BBC's obligation extends beyond the internet however. For those without a PC, the BBC is investigating Virgin Media's on demand platform. This still leaves a chunk of people with no access to the service because of technology constraints on the user's side (no PC, no cable, no broadband).

Even though Freeview does not offer the bandwidth, the BBC is sure to get embroiled in how to serve these users, where other competitors would simply write off the niche as too expensive to serve. This is the flip side to the breathing space they have had to develop the service.

We already have video on demand from Channel4, an evolving service from Sky and a promised launch of a service from ITV that looks spookily like that promised by the BBC. So what's the big deal with the BBC's launch tomorrow? I've said it could disrupt its industry to a greater degree that the iPhone, so I had better explain myself...

Driver for IPTV Adoption
Ofcom's MIA states that by 2011, the iPlayer is likely to account for 3% of TV viewing hours, which doesn't sound like a lot. This is in fact about 45 mins per household per week, assuming total viewing remains as today at around 25 hours per week.

But, as with Freeview, the BBC gives this new(ish) technology the credibility to go mass market very quickly. There will undoubtedly be a knock on effect on all other broadband television services because there may not be a more trusted organisation anywhere in the world than the BBC. If IPTV is good enough for the BBC, it's good enough for me...

Looking closer at the Ofcom projections: 3% of total viewing is 9% of the BBC's current viewing. It would be reasonable to suggest that competitors services might grow in line with the BBCs. This would mean every household in the UK watching on average 2 hours and 23 minutes a week of IPTV by 2011. Over 3 billion hours a year...

The MIA also says "The costs of the broadband capacity required to support the services could in aggregate be between £399 million and £831 million over the next 5 years." Once the capacity is there "the additional capacity would also be available for use by a wide range of other services, including commercial on-demand services, [so] it would not necessarily be appropriate to attribute the associated costs to the BBC services in isolation."

Ofcom's model says that the average capacity increase from the iPlayer will be 3GB per user per month by 2011.

Assuming that other broadcasters follow the same adoption curve, you are looking at almost exactly 9.5GB extra per user per month to serve the 9% of viewing hours at standard definition. This will add around 46kbps per user to an ISPs peak traffic load (approximately doubling what they have today). This is low, because I am using data that shows that early iPlayer alpha trial users had web-surfing-like peak to mean traffic profiles.

TV usage profiles tend to be much more peaky than web surfing traffic. Where you might get a peak to mean ratio on web traffic around 1.6, on TV viewing profiles, this looks more like 2.8. Cutting a long story short, this would push the traffic impact of the iPlayer from 46kbps per user up to around 81kbps additional traffic (easily tripling today's usage, from just one application).

Reverse engineering Ofcom's 3GB per user per month figure from the 3% penetration rate shows that they assume a 2Mbps encoding profile in their models. This suggests that high definition is not being taken into account.

If the BBC were to deliver at 1080p instead (as in the US have announced they will), you might want to multiply the total capacity requirement by 5. With all content (ITV, Sky etc) as HD, the 9.5GB might be 45GB extra for every house connected to the broadband network. This would push the incremental peak load per user up by between 220kbps and 385kbps depending on peak to mean profile.

Where there is demand, there is money, right...?

Actually, no. This is the other major problem with the BBC, the license fee and the universal service requirements. The BBC's iPlayer will not generate money from adverts (the BBC does not do ads), from subscription (the license fee already covers the service) and any other creative sources of income (including abroad), are likely to be relatively trivial.

This is not an issue for the BBC because the content is paid for already (its a catch up service of stuff already produced for broadcast). The service creation costs have been kept under control at £3m and rather than having to pay a big hosting bill, Kontiki's P2P client is being used, theoretically relieving the BBC of the burden of distribution costs.

The big losers are the networks who have to carry all this extra traffic and have no way of monetising it. This is again a BBC-specific problem because with other commercial broadcasters, the ISP is in a position to do an ad-revenue share agreement based on the unique element that the ISP can provide - the postcode. (We are going to come back to this point and the revenue opportunity from commercial broadcasters other than the BBC in LUI Part 10 early next week.)

The use of P2P actually makes the problem much bigger for the ISP. Historically, the BBC's web traffic, although significant, has been manageable via direct peering relationships between the ISPs and the BBC. Replacing this with P2P looks (to me at least) like a two fingered salute to the businesses that have to transport the BBCs product.

Even using the lowest results in the analysis, the iPlayer promises to double the traffic on the UK internet between now and 2011. On top of that the iPlayer opens the door to other broadcasters, which could mean that instead of doubling the volume of traffic, the iPlayer launch could drive an increase by tenfold or more.

I'm going to be watching the iPlayer's use of bandwidth very closely over the coming months. As I have done with Joost, Babelgum and 4oD, I will be running traffic source analysis and looking at where the Kontiki client gets its traffic from. Channel 4 also uses Kontiki, but using their service, I found that the scarcity of peers meant that much of the traffic was client server from the seed caches instead of actually using P2P.

I will be keenly examining the peer hit rates as that will determine the BBCs costbase. I will also be looking at where these peers are and whether BBC/Kontiki keeps traffic within the service provider's network or whether (like other P2P I have tested), in-country traffic source management is random. I will be publishing the findings here at periodic intervals.

If I can get the client from the website, the first set of data will be published here by lunchtime tomorrow...

UPDATE: no client = no data = no update. Sorry folks...

I got to the site by 7.40am, regsitered but have yet to receive the invite. I wouldn't say that the message board is on fire yet (10 ir so people grumbling about the same thing), but there are people who stayed up until midnight to register who are in the same boat.

They let Mashable in though, so if you are looking for a sneak peak that's the place to go. If you want a different perspective on possible adoption rates, I also found this.

IWR were able to run an initial test and reported that a 30 minute programme was 108MB, which suggests an encoding rate of 480kbps. It is not known what the download speed was, which may be different from the encoding rate to allow for buffering. The picture defaulted to 400 x 200 screen size, which sounds small.

More on this when I get my prized invite...

Labels: , , , , , , , ,


Monday, 23 July 2007


LUI Part 6 - Television over the Internet

This is part 6 in the Leeds Unbundled ISP (LUI) series that Keith McMahon and I are producing. The aim is to deliver a view on the commercial prospects of a hypothetical ISP, serving a niche community (Leeds in our example).

Before we can properly present the numbers though, we need to describe what those numbers are modelling. We have already looked at backhaul, staffing and our short and medium term product set. Today we look at the biggest variable in the future of our made up ISP: video.

IPTV is better for viewers than broadcast because it is truly on-demand. It gives viewers timeshift capabilities for BBC, ITV, C4, Five and Sky so that they can watch what they want on TV around the rest of their lives. So what is the variable?

While we are confident that video services like YouTube will continue to grow, we are not sure whether mainstream TV will successfully move online because of economic, marketing and technology challenges. IPTV is competing with established digital platforms (satellite, cable and freeview) that already penetrate 18m homes (more than have broadband). Getting mainstream TV online means replacing these distribution networks with the internet.

Consider the scale difference between the two extremes of service adoption: YouTube consumption is a few minutes at a time, a few times a week. TV is 25 hours per household per week. YouTube is currently 200kbps, IPTV as a vehicle of HD means 10Mbps.

There is very little that LUI can do to make any money from YouTube, but conversely, once we have our gigabit backhaul links in place, we are not too concerned about the cost of carrying its traffic. If they cranked up the resolution to the levels used by Veoh (700kbps), we might be a bit more concerned but as it stands, we are happy enough to carry the traffic.

What cannot be allowed to happen is for us to end up in a situation where we are a simple transport network for everyone else's broadcast-replacement services. Our commercial model, and that of every other ISP in the world, is based on carrying relatively small files (peak traffic over total users equals around 35kbps). TV viewing moving over to the internet and adopting HD resolutions will make this closer to 5.5Mbps (159 x the current dimensioning).

This needs to be paid for and the value is in the content: people buy music, video and TV. They don't buy bits and bytes. This means that we need payment for our bits and bytes bundled with the payment for the music, video and TV services. This means adopting a Fed-Ex model for superfast delivery of premium, newly released content and ad supported models for the rest.

Will this happen? Maybe, but only if the economics are right - we know that IPTV offers better functionality than broadcast because the internet uplink opens new doors for interactivity. With the public becoming disillusioned with telephone based interactivity on TV we think that the internet can rescue what had until recently been a popular genre of content.

Furthermore local loop speeds of 20M or so means HD at 1080p is practical and can be made available on demand. It's all technically possible but all these developments will only be attractive at a price point that is competitive with broadcast.

So where does that leave LUI?

Nowhere, right now at least. We just have a basic access service with some customers on CPS. Next up dev wise is the photo blog (due Q4 2007) and starting work on the softswitch (due 2008 perhaps). Enhancements to the photo blog and community stuff are mid 2008 launches.

We looked at buying in a wholesale IPTV service, even before we unbundled the access service. When Tiscali bought HomeChoice, we heard some suggestions that the HomeChoice platform would be wholesaled alongside Tiscali's LLU platform. While the attractions were obvious, the differential advantage was not, so we rejected that option.

More recently, we have looked at Iliad's platform and the service that Fastweb offers with a view to buying that in lock, stock and barrel. These are not currently deployed in the UK and although we could overcome the competitive issue to some degree, we just felt a little underwhelmed by the idea of taking something that had been done before.

LUI wants to do something a little differently and has to exploit our core concept: our localness. For LUI, IPTV has to be build around the community, but we also have to remember that it is still essentially a distribution network for mainstream TV that replaces the satellite dish, cable or freeview aerial in the home.

LUI's problem is that our customers can get all that from other operators, notably Sky & HomeChoice, so where we need to be different is in the EPG. We need to offer social networking in the EPG that exploits our localness and the social groupings within our customer base.

We are a network company and a small one at that, we need someone bigger to bring us the content. That means the content won't be exclusive so we have to add value to it another way. Hence the EPG and social network mashup.

That means things like the ability to recommend a programme to your circle of friends and comment on what you have seen, perhaps with an SMS gateway tacked on for alerts. When you turn on your EPG you see the linear TV option and the timeshift scroll back for the mainstream channels plus a Friends Recommend Channel.

We also have the Leeds Community Channel which will be developed before our IPTV service, but which needs evolve onto the TV when IPTV does arrive. The Community Channel is where local interest groupings (schools, community education, sports teams etc) can post virtual private videos to their members - much like the Iliad offer. All this is built on top of a core of content based on today's free local press.

We are already lobbying to force the publicly funded BBC content to be made available via public API so that anyone with a delivery solution can use it to deliver BBC content. The others are different because they are commercial entities, so why will ITV, C4, Five and Sky let us carry their stuff, sometimes in direct competition with their offerings?

Money. Pure and simple. They can get more from our subscribers if they deliver content via our network than they can via other means.

How? Because we know the customer's postcode and we can deliver that when we place the request for content. We also deliver the ip address of course, but they would get that anyway. With the postcode, they can then use Geo Mapping databases to paint a very good picture of who the consumer is, so they can use a) demographic and b) personalised advertising. They can't get this postcode without the ISP.

We could also consider sharing any special interest profiles that the user may create on our social network but this raises some ethical issues I suspect, not to mention the technical challenges.

But all of this needs to be pulled together: content, advertisers, client software, DRM and CDN. We are looking for one party to bring this to us. LUI's plan is to work with them and vice versa to prototype the future of IPTV.

The prototype is based on Joost, or Babelgum or Veoh (we haven't stitched it all together yet, it's just a plan). Something that runs on either an AppleTV-like STB or on the TV itself. Their job is to aggregate the content and provide us with an efficient distribution using P2P and local caching. They also handle all the advertising including the targeting and pay us a revenue share.

In order for this to work, Joost (or whoever our chosen partner is) must bring a deal with the major broadcasters. Joost does the deals with the content owners for us because they can and we can't.

Our BBC, ITV, C4, Five and Sky content comes through the deals that Joost has with them. Joost can pay better ad revenue than the producer can get by themselves from broadcast because we give them the postcode. As a result, they can target ads much more effectively. We get a rev share because we are adding value to their proposition.

Furthermore, we are solving one of Joost's problems - the EPG and social networking, which are currently lacking in their product - and we are leaving them to concentrate on their role as the IP TV operating system. We carry Joost's traffic and help them develop their intelligent localised P2P routing.

We provide the EPG (or at least our software partner behind the photo blog / community web stuff do that for us) and that has a two way API into Joost (or whoever). The EPG is our value add, our brand, our directory of content and the portal through which users can get to the array of services that we offer. Of course, they can go onto the open internet but with our gateway offering them RSS-based access to the world, we reckon that we can hold a fair proportion of the screen-time on our own services. Which is great for our ad revenues.

Of course this is all made up. LUI doesn't exist and there are holes in the plan and some very rough edges. With any luck though, this might give you a few ideas...

Part 7, back on Telebusillis is going to look at Hardware, which Keith will publish later this week!

Labels: , , , , , , , ,


Monday, 11 June 2007



I have written a lot about Joost. Now it's Babelgum's turn to go under the microscope, following its announcement last week that the beta test period is now officially open to all.

I won't waste your time rehashing the Babelgum story. You can find a good description here and here. Instead, I will waste your time giving you my view of the service and particularly how it compares to Joost.

My first impression of Babelgum was, unfortunately for them, of poor picture quality. One of the first segments I watched was "Get Outta Town: Sydney", think Lonely-Planet TV and you will be close enough. The presenter, Joseph Motiki, talked so quickly that you literally couldn't see his lips move.

This was common throughout and destroys the experience. Have a look at the screen capture below. It is from David Beckham's greatest goals when he scored from the half-way line and made his name. They also have Maradona's amongst others, including the Hand of God moment. Don't think for a second that you will be able to see what the referee and the rest of the world missed in 1986. In fact the picture quality on Babelgum is probably worse than it was on TV 20 years ago when the pictures came back from Mexico via satellite.

I have had to reduce the capture size by 3x to fit it here, but even that doesn't hide the blockiness of the images. No, this isn't one of those spot-the-ball competitions. It's there, somewhere. And the goalkeeper really does have two legs, even if one of them is completely indiscernible from the ad-hoardings.

My traces seem to indicate that Babelgum bandwidth usage is slightly lower at around 600-650kbps than Joost's 700kbps. While the bandwidth on all services will undoubtedly be cranked up improving resolution as time goes by, it is interesting to note that Babelgum packets arrive as TCP. Joost packets are UDP. Babelgum's traffic profile is very spiky (see first chart below), while Joost is much less so. As an aside, you can see in the Joost chart the higher bandwidth used by ads (initial profile) compared to media (second slice).



There is really very little going on in the way of P2P on the Babelgum service. It is built in, but it needs the network to be seeded. The same was true of Joost when I took the above sample on 3rd April, although their peer hit ratios were better six weeks later when I went back to look at them again. I'll give Babelgum a chance here and reassess this point in a month or so.

In terms of content quality, there is more choice on Joost. Sometimes more is less and this might be a case here. The programming on Joost is very yoof - very hip. Just not very watchable unless you like burps, farts or loud music. The Babelgum archive has some interesting stuff that I have actually watched... Like the Turtle Treaty piece on how the inhabitants of remote islands revere, and yet end up eating, the turtles that they share their lives with. Not what I expected, but welcome nonetheless.

Another area in which Babelgum seems to have the edge is in the realisation that this is TV as well as Video, and that there is a difference between the two. There is clearly the broadcast angle built into Babelgum - look at the AP news channel for example.

The Babelgum interface is really clean and I think that also works better than Joost at the moment. I think maybe though the Joost one is more scalable somehow as the Babelgum GUI might restrict the number of channels you can track due to the amount of info on each channel that they carry.

There is also a different business approach. The Babelgum one is open, almost to a fault, describing the revenue share on ads up front for example. It seems like they are going to go straight for the second tier of no-name producers and give them an indie-platform. Joost are much more into the "household names" by the looks of things.

So which one wins? I would have to say that I prefer Babelgum, but that I think that Joost will win long term. The niche that Babelgum seems to inhabit with its programming and navigation is nice but that is an area that Joost can fix probably easier than anything else. Where Joost has the upper hand is in its impetus which is taking the company into the offices of the very biggest names in the industry. Joost is not really a service yet, it's a vision. It will either be successful breaking the status quo or it will be acquired to make sure that it does not.

Babelgum just does not seem like it can emerge (out of Italy) and break into the "not-invented-here" markets. Joost is making a big play for the US because it understands that the US is a thought leader and standards setter in the internet space. In the end, this is where I think both are heading - into the position as "the operating system" for internet video. Here they take on YouTube amongst others... Is there room for multiple O/Ss?

Labels: , ,



This page is powered by Blogger. Isn't yours?

 Subscribe in a reader