The End of Streaming Media

by Dale Dougherty

Low bandwidth and poor quality continue to limit the successful distribution of audio and video on the Web. There may be a better way, however, to distribute multimedia content online, by scheduling downloads of high-quality content for appointment viewing or listening.

Today, there are two basic methods for obtaining a media file. One is to download the file and wait for the download to finish before you can access its contents. Using ftp is one way to download a large file, but so is Napster, creating your own copy of a music file. Napster and its kin could be used to download even larger video files, but the size and the ensuing wait are so much greater that most people feel that video downloading will not be practical until broadband is widely deployed.

The second method is streaming. Streaming requires a specialized client and server such that the client is able to begin decoding and playing an audio or video file as it is received, eliminating the wait for a download to finish. From a user's point of view, this seems a lot like using a radio.

The problem with streaming is the compromises in quality that are made to stream an audio or video file. In one case, a streaming audio file might be compared to AM quality whereas an MP3 download might be compared to CD quality. In video, the compromises are even greater, as the frame size, frame rate, and resolution are typically reduced.

Nonetheless, streaming media helps to solve the problem of making multimedia content immediately available. The analogy for streaming media seems to be broadcast transmission, radio or TV. Thus, it is assumed that a combination of increased bandwidth and improved streaming media will eventually solve the problem.

I wonder if there's not a different approach possible. If we are willing to throw out the requirement that a song or video must be available immediately, we can make the analogy to digital recording devices such as TiVo or Replay. We specify that we want to get something once it is available and then it is stored until we use it. In other words, you negotiate bandwidth by taking more time for downloads, but you give the user some control by allowing him or her to specify a time when he or she actually intends to use the media file. Moreover, many regularly recurring programs can be scheduled for download without requiring user intervention for each download.

Instead of a streaming media client, we need a client that helps the user identify useful or interesting content and then schedules it for download. Think of a combination of Seti@home, PointCast, and Napster. There's a client that you can use to find out what's available, based on its schedule of availability, and then it negotiates the download times, notifying you when content is available. Most importantly, the client should allow you to set up a schedule for downloading the file based on when you actually intend to use it.

If I want to listen to an audio program, I really have all week to download it and listen to it. I might want it immediately, but I might also be willing to have it a day later. The program might have several days to download a file. And it could get rid of the file once I listened to the program.

I might want to have All Things Considered available at 8 p.m. each night, even if I don't always listen to it. Likewise, a conference webcast doesn't need to be viewed in real time to be valuable, but the conference supplier may not want to archive the content or offering continuously. You want to grab it while it's available and keep it until you are interested in viewing it.

A program like Meerkat could be this kind of scheduler, based on metadata provided by the producers of multimedia content. Meerkat, or some version of it, could feed data to an application smart enough to download, organize, and present the content that you have scheduled.

File size remains a problem for video, not so much for transmission in this model but for local storage. The client would have to manage disk space by setting limits. This is not unlike what TiVo and Replay do today.

Overall, I think the value proposition is putting the consumer in control.

Dale Dougherty is the editor and publisher of MAKE, and general manager of the Maker Media division of O'Reilly Media, Inc.

Discuss this article in the O'Reilly Network Forum.

Return to the O'Reilly Network Hub.