Who Needs Streaming Video?

Streaming video is what we -  a generation of TV viewers – grew up with. You watch what is broadcast when it’s broadcast as it’s broadcast. You blink; you miss something.  Something disrupts the signal; a fragment of what you’re watching is gone forever. Doesn’t matter whether the content is as fresh as today’s sports event or as stale as a Bogart movie; it comes streaming at you. That’s what broadcast TV technology delivered and that’s what we consumed.

It’s absurd, however, to assume that streaming video is what we want.  This absurd assumption is nevertheless behind much of the planning – especially by traditional telecom carriers – for the next generation of Internet.

We consumers have lapped up everything we can get our hands on that de-streams video.  VCRs – remember them? – turns out that what we really want is to watch programs when we want to watch them – not when the broadcaster chooses to blast them at us. And, by the way, we’d like to pause when we go to the bathroom, please, and fast-forward past what we don’t care about.  And VCRs were really hard to use.

Then along came TiVo and a generation of copycat DVRs which still aren’t nearly as good. Now we can watch one end of a show while we record the other.

I like to watch baseball, basketball, and football; I don’t like to watch games whose outcome I know. I do want to watch the most current game and usually roughly at the time it’s played. But I don’t watch live; my DVR does that for me. I start watching when I’m ready, after dinner and after the news, perhaps.  Now I have a nice buffer built up.  No commercials for me unless they’re really interesting. No inane patter during halftime. Instant replay when I want it rather than when the producer thinks I want it. Last thing I want to do is have this streamed at me the way it was in the old days.

We set TiVo up to collect Sherlock Holmes, Sean Connery (for Mary), and Bogart.  It accumulated a nice backlog of shows we want to watch.  Did we know or care when the shows were actually broadcast? No; couldn’t care less.

Gradually, video content we want is becoming available on the Internet.  In Stowe we don’t have enough bandwidth to stream TV quality video in real time although we can watch flash fine once a buffer’s built up.  Does this matter?  Not much.  It’s a lot more immediate to download and then watch than it is to wait for the movie to come from NetFlix (which isn’t all that bad either).

The most important features of video on demand are quality, selection, convenience, and availability – not necessarily in that order.  These criteria apply whether we get our video from a Blockbuster store, in the mail, over the Internet, or from our cable provider.  The only advantage that streaming as a method of delivery gives is that we can start watching immediately after making a selection instead of waiting some period of time. But the price of streaming is that we can’t pause or skip forward – unless, of course, we record in which case we’re not watching in real time and the streaming delivery is being wasted because we’ve elected to use TiVo as a buffer.

Occasionally there is a news story so compelling that not watching in real time – given a choice – would be almost inconceivable. Watching the terrible collapse of the World Trade Center on 9/11 is something I’ll never forget. Even then, of course, I didn’t know and didn’t care whether the process of broadcasting to where I was watching introduced a few seconds of delay. But granted that there are rare occasions when streaming capability is important to have.

The chart below is from an otherwise excellent Swedish study. It assumes that, in a few years, we’ll need enough capacity to deliver FOUR streaming videos in very high definition to every home and that networks MUST be designed for that load including of course, the next mile connection (3P+ is their view of what a service provider should be able to provide five years from now).

I’d appreciate reader comments on what four events you think you may need to or want to watch in absolute real time in the highest definition five years from now.

Why does this matter?  Why shouldn’t a network be built with these lofty goals?

First, because a network like this is unnecessarily expensive to build. Second, because this network makes an artificial distinction between the bandwidth used coming in for TV only and the relatively small amount of bandwidth usable symmetrically for “Internet access”.

What will really be happening is that we will find most if not all the video we want on the Internet. Anything archived’ll be there including all old movies.  I think real time events like sports and news’ll be there as well; it’s starting to happen now. So almost everything we want squeezes through the relatively tiny amount of bandwidth identified here in the right hand column as “Internet Access” while the vast bulk of our capacity is reserved for these four non-interactive streaming events that someone thinks we will want to watch simultaneously.

The right design is to have all of the available bandwidth usable for any application. Much less total bandwidth is then required.  The occasional requirement for streaming can be met from the pool of available bandwidth. At our choice, available bandwidth’ll be used for upload, download, video, gaming, using Google’s Office apps remotely, listening to or sending music, or doing a whole lot of things we haven’t thought of yet.

So why is the Internet of future being arranged to suit streaming which is actually an inherited limitation of broadcast TV?

1. Lack of imagination.
2. (very real) the middlepeople who own the contractual rights to much of the content we want to watch will not agree to having that content released except in streaming mode for fear (well-founded) that they will lose their ability to monetize by charging directly or selling ads.
3. (also very real) the carriers who want to build these over-engineered networks know that they cannot charge as much for pure delivery of bits as they can for “differentiated services”.

Just to emphasize how important #2 is: when a network like Burlington Telecom’s municipal fiber to the home is built, the builders of that network cannot deliver the content currently on the cable networks unless they agree to deliver that content in an emulation of cable TV. They do not have the choice of making all of the content available as video on demand or making it accessible through the Internet access capability they give their customers.

Comments

[Nico Flores]

Intersting post. A few comments...

Perhaps a distinction is needed in your post between one-to-one and one-to-many streaming (multicast / broadcasting). There are technical ways of making the latter economical (although the commercial/regulatory issues are considerable), but I take your post is mainly about the former.

About that, I agree with you that much of the content that today is watched 'live' will in the future be watched on-demand. But I disagree that 'live', one-to-many, broadcast-type consumption will disappear or decrease: that is an open question and has a lot to do with consumer behaviour that can't be predicted.

You say elsewhere that TV channels are just inefficient ways of delivering shows. They are that, but not just that. Part of watching TV is about being part of a social ritual that has to do with liveness and shared experiences, and the demand for that won't go away because of a technical change.

What I think will happen is that 'live' will remain, but its formats will change. It will be mostly truly live (news, live concerts, sport) and you will rarely see long-form recorded content (e.g. movies) on live channels.

Also, one-to-millions live distribution is controlled by gatekeepers today and will remain so for quite some time (for my reasons see this http://tinyurl.com/28yx5o). This will be key to promoting content available on-demand, for which distribution is cheap and open to anyone.

[Craig Plunkett]

As far as what I need in real, or near-real time, HD video of my beloved and frustrating New York Rangers, some kind of news feed, some traffic info, ( that really needs to be delivered to my vehicle in real time ) and maybe weather. That's it, the rest can be timeshifted and place shifted. I do have to comment on Adrian's and Patrizia's comments. P2P bittorrent traffic is currently constrained by my provider, Cablevision. As soon as I crank up Azureus, poof, there goes my bandwidth. So there needs to be some assurance that you get the speed that you are buying, without regard to the application that runs on it. Until that is unfettered, asynchronous IPTV won't fly. And as far as Patrizia's comment about interactivity, I completely disagree. People don't want interactivity with the TV other than switching content. Interactive TV and movies have flopped time and time again. There's a reason TV watchers are called Couch Potatoes. When people want interactivity with their TV, they play video games. Now if there was a different kind of interactivity other than switching camera angles and choosing alternate plots and endings, you might have something there.

[Tom Evslin]

Adrian:

You are correct and a step (post) ahead of me. I agree with you that people will watch video on the Internet in ways that are fundamentally different from broadcast TV and you give good examples. Since both of these utilize p2p, they require upstream bandwidth as well as downstream but don't require real time error-free delivery.

[Tom Evslin]

Patrizia:

As usual you're a step ahead of me. The interactivity is VERY important (upcoming post) which is just one of many reasons why bandwidth shouldn't be locked up for dedicated uninterruptible down-streaming.

[Patrizia Broghammer]

Tom, you miss the important feature of the Internet: Interactivity!

That is what makes IPTV alluring.
Communication, do you remember? It's all what we are screaming and shouting since long.
Internet is the network of networks, it can put together people who are millions miles away, it gives voice to ones who wouldn't have one...

You forget that the most requested feature in EVERY entertainment IS being a part of it...
If you download and you look at it you have a may be cheaper, may be more comfortable, may be more used copy of the actual TV.
But we want something DIFFERENT, don't we?

[Adrian Cockcroft]

You don't mention bittorrent and joost. What I'm seeing is that many younger users are comfortable with bittorrent as a way to gather the TV and movies they want to watch. They have to pick things in advance since it often takes overnight to download the show. With bittorrent the data arrives in a randomized order, so it all has to arrive before you can start watching it. This is ok for geeks, but not for grandmas.

I've been playing with the Joost beta. It "feels" TiVo-like, actually works better than the TiVo user interface if anything, since it can use keyboard/mouse input for searching and navigation. It looks great as a user interface, the video quality is similar to bittorrent, i.e. good enough, but programs are available in seconds not overnight. Its got the same technology scalability/cost model as bittorrent, but is far easier to use. Its being monetized by a few very short adverts (5-10 seconds at present), much less intrusive than on TiVo, and it doesn't let you skip them. I think its going to be big. I can see a very wide audience using joost, and they seem to be landing the content deals that will make it worth watching.

[Bill]

If only Burlington Telecom ruled the world...