Last Tuesday’s presidential inauguration was one of those moments where I think all business except for vital functions like transit and public safety stopped all over the country as people tuned in to watch Obama’s swearing in. Another thing that stopped for a lot of people was the internet. Arguably this was one of the biggest, if not the biggest live streaming video events in the history of the event. It was also one of biggest tests for streaming video over the internet, and the results were decidedly mixed.
I was at work on Tuesday, where one of my responsibilities is providing instructional media support. As soon as I got in that morning I started getting requests from people all over our building to set them up to watch the inauguration. Now, the building I work in is poured concrete monstrosity that acts like a Faraday cage, successfully blocking reception of most broadcast signals. On top of that, there’s no cable TV in building. So I advised anyone who asked about getting a TV that they should consider viewing a live stream. Then I went to go set up a live stream in a large conference room with a video projector. At that moment I realized that maybe the live stream wasn’t going to work out so well, as it took many different attempts on several different sites before we could get anything to stream for more than a few seconds. That was around 30 minutes before the inauguration was set to begin.
When I returned to my office all attempts to get a stream there–whether from CNN, Ustream or even the CBC–resulted in failure. A few minutes after the ceremony began I received an email from our central IT network department, advising us that our multi-gigabit campus network had ground to a halt due to people watching the inauguration online. Looking at Twitter and the CNN live Facebook stream I saw that we were not alone, as folks all over the internet were finding it hard to get a reliable stream.
In the end it looks like about 7 million people were able to get live streams of the inauguration, according to Dan Rayburn whose estimates are based on talking to actual content distribution networks. By any standard that’s an impressive simultaneous viewership for the internet. But it’s less impressive compared to broadcast television, where 37.8 million people watched the inauguration.
More illustrative of the difference is the number of people who were denied the ability to watch the inauguration due to capacity limits. That is, another 37 million people could have tuned in to the inauguration on broadcast, cable or satellite TV while still leaving capacity for 37 million more. Whereas on the internet 7 million appears to be the upper limit — past that nobody else could watch.
That’s because streaming internet video is based upon a technology called unicast. With unicast every single viewer is getting a unique stream of data from a video server on the internet. So, if the video stream requires a bandwidth of 256kbps, then ten viewers requires 2.5 megabits, and a hundred viewers require 25 megabits. So, how much do you reckon 7 millions viewers might require? Using my thumbnail estimations, the number is in the region of 1700 gigabits. On top of that the video provider streaming that content has to pay the bandwidth for each and every one of those viewers.
Although there are certainly costs associated with reaching larger audiences in the broadcast realm in terms of larger transmitters or contracts with cable and satellite providers, by and large each additional viewer to a broadcast channel causes no additional cost to the broadcaster. If CNN and FOX News both are able to reach 50 million households, but only 10 million tune in to CNN but 12 million tune in to FOX, there’s ostensibly no additional cost to FOX to reach those extra 2 million households. But it would be much more expensive to reach 2 million more with internet streams.
The reason why this difference is important is because so many arguments in favor of loosening ownership rules for broadcast or justifying the Clear Channelesque gutting of our broadcast media is that the internet provides a rich, nearly infinite alternative tailored to many more diverse and narrow interests. Of course, I’d be fool to say that’s not entirely true. Indeed the internet is a medium that enables many more people to become small-scale narrowcasters at very low cost compared to broadcast TV or radio. Yet, it is still ill-equipped to offer a live, truly mass media experience on the scale and magnitude that television and radio can.
Furthermore, when the a large number of so-called internet broadcasters attempt to offer up such a live broadcast experience it risks bringing the whole enterprise to a halt, getting in the way of other streamed audio or video content. By comparison, 37 million people watching the inauguration on broadcast TV posed no barrier to watching cartoons or home shopping instead.
At root the internet was never intended to be a live broadcast medium. It’s foundational technology, packet-switching, is very efficient for moving data around a diverse, multi-nodal network, and was invented specifically to allow for little pieces of data to move around out-of-sequence and even by different paths, and still be able to be put together whole and in order. It also wasn’t designed to happen in linear real time. The problem with live audio and video is that they’re fundamentally linear — the order of each frame of video and each second of sound is critical, and any misordering renders the program unintelligible. An increase in the speed of the internet combined with clever protocols and programming has made streaming live content with minimal lost data a reality, but it’s still really a hack.
One of the reasons that the inauguration was able to be viewed by 7 million online in the first place is that major content distributors like Akamai have invested heavily in edge servers. With this approach the content you receive comes from a server that is located as geographically close to you as possible, so that the data passes through as few networks as possible, putting less strain on the whole internet. I’m certain that advances in edge networks combined with increased bandwidth capacity at all levels will make the internet more accommodating to live streaming media in the future. But that day is not here yet.
I see the problems with the inauguration streams as a reality check to remind us, and especially our new policymakers in DC, that the potential of the internet as a true media broadcast alternative has not yet arrived. That day will happen only when there’s sufficient investment in and deployment of bigger pipes at all levels of the internet. Yet, this is something that our broadband ISPs don’t necessarily want to see because rich internet streaming poses competition to their cable TV business. So while with one side of their mouth they’ll point to the diversity of the internet as rationale for deregulation and consolidation, with the other side of their mouth they’ll lobby against increasing content-agnostic expansion and network neutrality because of the competition it represents.
We need to invest in our broadband infrastructure, from backbone to last-mile, while keeping the information superhighway open to all traffic, without any prejudicial tolls. It must be a truly neutral network. Only then is there the hope that the internet might be a suitable alternative, supplement or replacement for broadcast media. And until then, we owe it to ourselves to value the broadcast media we have and not to let it be starved to death in the way that’s happened since 1996.
Leave a Reply