| Broadcast Video vs. Internet Video |
One of the more riveting bits of speculation wrapped around the recent Google/You-Tube merger is the one about a Google which someday sells packages of video content to consumers. Just like a cable operator. Or a satellite company. Or a telephone company. Google as a multichannel video provider is a head-spinner. On the regulatory front, you have to wonder: Will Google relax its death-grip on network neutrality, if it too needs to send "broadcast quality" video as a part of a subscription package? And then there's the question of high definition TV content. If audio snacks on bandwidth, and video eats it, HD devours it. Is the public Internet plumbed to deliver big vats of big HD video, without collapsing? Can it handle even the ordinary needs of broadcast video? Enter the technology portion of the conversation. Broadcasting a digital video stream, either over the air or over a series of wires, differs considerably from sending a video stream through the Internet. This week's translation is about those differences. If you're having a déjà vu moment, rest assured that you did already see the foreshadowing for this. Kind of. Remember that National show in 2003, when Microsoft chairman Bill Gates grilled Comcast CEO Brian Roberts about how much of his digital spectrum he'd make available in IP (Internet Protocol)? Add time and Internet momentum, and here we are. Gates posed a long-term question: While you cable people are going about making your networks all-digital, why not make them all-IP, too? Three years later, the answer is the same: Why? As with most things, both techniques -- traditional broadcast, and IP -- have strengths and weaknesses. Broadcast video, like what you get from your local TV stations, cable operator, or satellite provider, has two identifying characteristics: Signal direction (one way only) and audience (no upper limit, no need to know your electronic address). "One-way" means there's no mechanism for a receiver (your TV) to tell the sender (your video provider) that a packet was trampled along the way. That's why digital broadcasts apply extra bits to what's called "forward error correction," or FEC. FEC works in the background to make sure your digital picture (standard definition or HD) stays solid. FEC applies to all three types of broadcasts (cable/satellite/broadcaster). It was built for efficiency. (Yes, cable boxes are two-way, but they weren't always. And broadcast video, because of that FEC, only needs a one-way ride.) By contrast, Internet Protocol -- which underlies the public Internet, and thus Internet video delivery -- is two-way. Its audience is point-to-point and individualized. It was built for a different kind of efficiency. Two-way also means there's a path for a receiver to tell a sender that some bits got smooshed en route. In tech talk, this exchange goes by "ack-nack," for "acknowledgement" ("got it!") and "negative acknowledgement" ("re-send please.") Because every packet goes through the "ack-nack" process, IP doesn't use very much forward error correction. It doesn't need it. It follows that because IP was designed for point-to-point, individualized deliveries, it is not so great at anything that needs to be sent simultaneously to lots and lots of receivers. Think of all the ack-nacks. Plus, any sender -- Google or otherwise -- essentially needs to know who wants its video. Bandwidth notwithstanding, they can't just blast it out to an unknown "everyone," like a broadcast. Translation: Digital broadcast techniques used by cable, broadcast and satellite providers excel at efficiently sending popular, commonly watched, "one to many" content. They are not so inherently good at "long tail" or highly individualized material. (That's a job for VOD.) The inverse is true of IP. It's not so good at hitting the masses with broadcast material, but it shines for lesser-watched shows with smaller audiences. Of course, the IP camp is cooking up ways to remedy its broadcast shortcomings. You'll hear people talk about a technique called "multicast" (translated in the August 30, 2004 edition). It's like putting the flag up on your mailbox, to indicate that you have mail, except instead you're indicating to the Internet that you'd like to receive a particular piece of Internet video. So far, it's regularly described with terms like "clunky" and "inefficient." So, Mr. Gates, the "when all-IP" question is still plausibly answered with a "why." Why do one to the exclusion of the other, when you can already do both? Things best delivered on IP will be sent on IP. Things better delivered on broadcast, like high definition or live events, will be delivered on broadcast. This column originally appeared in the Technology section of Multichannel News.
