Streams Won’t Pay for Themselves


To stream, or not to stream? That is the video delivery question. The answer is definitely not streaming, if you belive Dr. Andrew Odlyzko, a mathematics professor at the University of Minnesota and author of the recent paper The Delusions of Net Neutrality (PDF). Odlyzko argues that video is a low-margin business (a point my colleague Stacey nicely recaps over at GigaOM) and that faster-than-real-time progressive downloads are preferable to real-time streaming.

Odlyzko believes that service providers do not need to build out fancy new video technologies or infrastructure, so long as people can wait a second for their content:

The vast bulk of video that is consumed by people today, and is likely to be consumed in the future, does not require real-time transmission. Most of it is movies and video clips that are pre-recorded, and thus can easily tolerate a certain amount of bu?ering. Even many apparently real-time transmissions are actually delayed.

Video can be delivered more efficiently and less expensively using downloads over streaming, says Oldzyko, and these downloads can be more accommodating to the viewing habits of online audiences and just as secure as streams.

Oldzyko thinks the belief that we need real-time streaming is a holdover from broadcast and phone networks. Holding a live conversation definitely requires no latency, and broadcasters traditionally didn’t have a buffering option. Neither of these are relevant concerns for enjoying video content online.


Dan Rayburn

I think your example in your response points out the problem of the paper and the lack of understanding of the technology. You can’t use video conferencing as an example of talking about “real-time streaming requirements” when video conferencing is not streaming. Two completely different technologies.

As for examples of public sources of where the “different options for video delivery have been
extensively discussed” all you have to do is ask customers. They discuss the pros and cons of the technology all the time. They speak out it in public about it, write cases studies on it and use it to determine what platforms they choose. Just do a search on the phrase “progressive downloads” on and you will find a lot of references.

Microsoft, Apple and Real have been talking about downloads versus streaming for ten years not to mention all of the universities that speak to the subject in papers and classes. This is not a new debate or issue.

Andrew Odlyzko

It’s great that Dan Rayburn took a look at my paper,
but a pity he did not read it carefully. He grossly
misrepresents what I wrote. As one simple example,
the paper does acknowledge prominently that many
transmissions (such as videoconferencing) have
actual real-time streaming requirements.

And I am definitely not claiming originality in
advocating faster-than-real-time downloads.
I wrote about this a decade ago, but others had
done so a decade earlier. And many other had
rediscovered this technique in the meantime.

The point of the paper is that most of the
publicity, especially at the policy level, is about
the need for complicated and expensive networks
in order to deliver video in real-time streaming
mode. I provide references to several such claims
by prominent industry sources. Dan says that the
different options for video delivery have been
extensively discussed. Can he provide some
references to public sources, especially in media
aimed at the general public or at policy makers,
and not just at specialists? I was not able to find
any, except in very technical publications. And
that is the point of the paper (which is to be
presented at the Telecommunications Policy
Research Conference later this month), to show
that the widely cited and widely believed claims
about the need for special network technologies
and traffic management policies are not justified.
That many services successfully deliver streaming
video today, over today’s networks, without any
special measures, only strengthens the case.

Dan Rayburn

We’re seeing a lot of topics lately on this streaming subject but people like Dr. Andrew are missing so many of the key points and have no real-world experience. You can’t make general statements about technologies like streaming and progressive download and apply it to all video content.

There are many, many applications that simply do not work in a progressive download model. Think of the online auction business. You can’t bid on items if the video feed is not live. There are many use cases like this including live earnings calls, new product launches, radio streaming not to mention all of the enterprise applications that need near real-time delivery using streaming protocols.

Plus, if delivering video is such a low-margin business, where is the data to show the savings? No where in Dr. Andrew’s paper does he say what the savings is. The major CDNs don’t charge more for streaming than downloading and if their infrastructure costs were that much great for streaming, they would have to charge more. While some CDNs may incur more costs for streaming, that is quickly disappearing, so the cost point is moot.

In addition, his paper speaks as if no one who does streaming has ever heard of downloading before or even knows it is an option. He says “let us consider the strange situation in which this issue [download versus streaming] is not discussed publicly, and the advocates of each of the two types of video transmission mostly seem unaware there is a real alternative to their preferred solution.”

This topic has been discussed for years, and the pros and cons of each delivery option is highlighted by the major video platforms like Microsoft and Adobe. Not to mention that the vast majority of content owners use BOTH means of delivery depending on the type of content they have, the length and the type of device they are delivering it to.

He continues to say that “advocates of streaming seem generally to be unaware that there is any alternative. On the other hand, advocates of faster-than-real-time file transfers are aware of streaming, but generally regard it as a bizarre aberration. How this mutual misunderstanding could have persisted for years without a public debate is a real mystery.”

The real mystery is who Dr. Andrew is talking to? Online video via downloading or streaming has been around at least fourteen years now. This is not new technology, not cutting-edge and delivery services are a commodity.

Dr. Andrew’s paper if full of phrase that make no sense and are completely inaccurate from a technical level and clearly he has not used the platforms before. He says that Microsoft platform has “already built in a 15 to 30 second delay to live video streams”. Many webcasters who have used the Windows Media platform knows that not the case and that all live encoders have a delay but typically is is under 15 seconds.

He goes on to say that “in practice, there is no need for even a 1-second delay. As is done in YouTube and other services, one can start playing the movie right away, from the buffer, as that buffer gets filled.” Really? The biggest complaint with YouTube is that it takes a long time for clips to start even when the clip is only 30 seconds in length.

He also uses a lot of examples of groups of people he has spoken to in order to confirm his ideas but later in the paper say that “many of them had been working on pro jects in wireless communication, in sensor and ad-hoc networks.” Delivering video over wireless networks is completely different than delivering video to the PC. He’s asking the wrong group of people. He says that live streaming “requires complicated and expensive gear from system providers” and is “not for network users”. Live and on-demand bits all go through the same routers and networking gear.

He also says that “the myth, that movies should be delivered in streaming mode, are very widely held.” Well lets see. The movie services we have had to date like Movielink, CinemaNow, Amazon and iTunes amongst others have all been downloads, NOT streaming.

His paper would never pass muster with anyone who is even remotely familiar with the online video industry and you could spend hours proving, with facts, how much of his “assumptions” or technical descriptions on how video works are completely inaccurate.

Comments are closed.