ABOVE THE CROWD Dispatch * Monday, Dec. 22, 1997
http://www.news.com/Perspectives/gurley.html
J. William Gurley * Hummer Winblad Venture Partners
> What would you think if I told you that the Internet would soon have a
> competitor for broad-based data distribution? Could something as
> mind-numbingly huge as the Internet see competition in such a short time
> frame after its own burst to stardom?
If I had a million dollars every time someone predicted this kind of
thing... well, I'd be a millionnaire and would no longer have to read
peoples' predictions.
> Over the next 12 months, a new form of service will arise that will
> simultaneously upset the current market for Internet content, topple the
> current standards in the consumer electronics industry, save the storage
> industry from its worst slump in years, and qualify the FCC's HDTV
> bandwidth grant as the greatest charity event in the history of the world.
> The technology that could cause such profound change goes by the name of
> "data broadcasting," and although it's been around for years, its time to
> shine has finally come.
Okay, he's got my attention. Why data broadcasting now?
> To understand the potential for data broadcasting, you must first accept
> the following two arguments: that bandwidth is still scarce and that local
> storage capacity is abundant. The place where bandwidth scarcity is most
> serious is still the home, and as a result, it is in the home where data
> broadcasting will have the greatest impact.
Okay, I accept both of these arguments.
> Appealing solutions for the "last-mile" problem seem as distant as they
> did a year ago. Infrastructure requirements, capital constraints, and
> bureaucratic hurdles have all combined to slow the march of progress. A
> current solution is to multiplex two 56-kbps lines for a cool 112 kbps of
> bandwidth.
112 kbps is cool?! Hello, does no one remember ISDN lines? And what about
cable lines??
> This solution, which will require the user to purchase two phone
> lines, can only be described appropriately as a kluge.
I thought the word kludge has a D in it.
> (Nonengineers may need to look this up.) Perhaps an even larger issue is
> the growing concern by some that the largest long distance companies are
> dangerously close to maxing out their backbone bandwidth.
This couldn't be true for MCI or AT&T -- they seem to have bandwidth to boot
right now. Everyone else, however... aw, who cares about everyone else?
See also:
http://www.teledotcom.com/1197/features/tdc1197sonet.html
> In other words, fiber needs may be
> outpacing our ability to lay down new trunks.
Not sure I believe this, but I'll give it the benefit of the doubt.
> The abundance of local storage capacity is easier to grasp. One need only
> look at the falling stock prices of Seagate, Quantum, or Western Digital.
Or visit the local Fry's.
> Hard drive manufactures, in a race against commoditization, have trained
> themselves to always look to the high end of the market. Unfortunately,
> they forgot to poll the market with respect to the need. Does a starter PC
> need a 2GB drive?
Yes!! In fact, more!!!
> Does a high-end PC need an 8GB drive?
Heck, I won't be happy until I can store Terabytes, man! I'm sick of
downloading the South Park "Spirit of Christmas" every time someone
wants to see it...
> The rise of the Internet has had two negative impacts on the demand for disk drive
> capacity. First, users have less need to "horde" software, as most freeware
> is readily available on the Web. Second, we spend more time surfing, which
> results in less time acquiring and using software. The bottom line is that
> our drives are getting bigger, yet our demand curve is flat to falling.
Mine isn't -- maybe because I like to keep movies, and pictures, and sound
clips, and neat documents and programs I find on the Web. Am I the exception?
Am I exceptional??
> Yet it's practically guaranteed that next year there will be bigger and faster
> drives at lower and lower prices.
Excellent. I say, keep em coming.
> So what is data broadcast? The phone network was designed to facilitate
> one-to-one interactive, real-time communication. The video broadcast
> networks that we use to watch TV (UHF, VHF, cable, and satellite) were
> designed to facilitate one-to-many transmissions (noninteractive). The
> Internet was designed to facilitate one-to-one interactive asynchronous (no
> time guarantee) communication.
So much for multicast. :(
> In the beginning, Internet usage was
> primarily focused on this one-to-one type of communication with
> applications such as email or a single user interacting with a Web page.
> Over time, we began to appreciate data that might appeal to multiple users
> simultaneously. Applications like audio and video streaming are sometimes
> limited by Internet traffic. Additionally, we have come to realize that
> multiple Internet users want to see the same event-based data at the same
> time, such as election results, stock market crashes, or major news events.
Amen to that, brother! Bring on the multicast... :)
> IP multicast is proposed as a way to alleviate the inefficiency of current
> streaming technologies. However, there is a much cheaper way to deliver
> bits that everyone wants to see: by transferring them over a network
> designed for broadcast, such as television, cable, or satellite.
Okay, I'm game.
> Here are the details of how this will work. We will first need technology
> (possibly an ISA card) in the home that will allow us to connect either the
> PC or a set-top box of some kind to a broadcast network. In the case of
> cable or satellite, this will be through a coax cable. (Intel has a product
> in this space but it has had only marginal success.) The first wave will
> likely use excess capacity in VHF channels, allowing us the ironic
> combination of attaching "rabbit ears" to our PC. Once we are hooked up,
> "data broadcasters" will push data past our homes on a continuous basis.
> Our devices will be programmed to "capture" the information that we might
> find most compelling. Hopefully, now we can see why we referred to
> bandwidth scarcity and storage abundance as prerequisites for success. This
> whole model assumes that storage is so cheap that it makes sense to blast
> bits past everyone's home on the chance that they might want to see it.
This is starting to sound dangerously close to video-on-demand.
> It is important to realize that this solution will never replace the
> Internet because it is not interactive.
You got that right.
> Therefore, things like "interactive
> trading" and electronic commerce cannot possibly take place over this
> broadcast network. However, commodity content like stock prices, news
> reports, sports scores, and weather clearly belong on a broadcast network.
Yeah, but they're called CNNfn, CNN, ESPN, and the Weather Channel,
respectively.
> This information is plentiful and is available for free, and there is no
> reason not to put this information in the air where everyone can get to it
> quickly. No more waiting for the phone line to connect; no more waiting for
> clumsy graphics to cram through a 28.8-kbps modem. Just point, click, and
> enjoy. What most people fail to realize here is just how cheap it is to
> broadcast data that everyone wants to see. If you think about it, there are
> something like 175 DirecTV channels soaring past your head every minute.
I wonder if this is why I feel dizzy.
> Just one of those channels could represent 32 megabits of data each second.
> That represents a throughput rate of 240 megabytes per minute. That's a ton
> of data, and that's just one channel.
Yes, but where is the economic model that makes this data broadcasting viable?
> The world will likely be surprised by the amount and breadth of data that
> is delivered over broadcast networks. Moreover, the amazing difference in
> bandwidth speeds on broadcast might open up markets that we didn't expect
> for years. For instance, the <i>Wall Street Journal</i> could arrive at
> 4:15 a.m. every morning as a 20MB multimedia application, complete with
> videos and 360-day stock charts on all equities.
But I *like* my morning *paper*.
> Your next update to your
> operating system or browser could be pushed over in a flash.
Okay, I gotta admit, this is nice in the abstract. Of course, the
thought of Windows 98 taking over my home computer overnight (and as a
result, nothing working in the morning) horrifies me.
> Lastly, and
> most importantly, that market for purely digital audio and video could
> finally begin to explode. That's right, we could begin to receive large
> files of audio, video, and interactive content (such as the newspaper
> mentioned above) that would be enjoyed locally.
Mmm, baby, bring it on. (drool, drool)
> The reason this should have a big effect on the consumer electronics
> industry is that manufacturers seem overly obsessed with fixed-length
> storage devices such as CDs and DVDs.
Yeah, no kidding. What's up with that?
> This must certainly be tied to a
> long-term belief that physical media will be purchased in stores with
> content preloaded on these discs. This need not be the case.
Please please please some commercial entity free me from having to maintain
http://www.cs.caltech.edu/~adam/local/cds.html
> Broadcast networks will solve the bandwidth problem, and large rewritable storage
> devices, such as hard drives, can act as music and video repositories.
Replace my four bookcases of CDs with a single hard drive? Where do I sign up??
> "Nonsense," you say. "Hard drives aren't that big!" Think again. IBM
> recently announced a 3.5-inch disk drive that stores 16.8GB. Uncompressed
> CD-quality stereo sound occupies 10MB per minute. Liquid Audio, a leader in
> high-end audio encoding and delivery, can compress that down to about 1MB a
> minute. This means that IBM's new drive could hold 16,800 minutes of music,
> or more than 300 full-length albums.
I'd still need 6 of them. Not to mention my video collection...
But won't the content industry be (rightly) concerned with large-scale piracy?
Digital watermarking isn't mature enough yet, is it?
> Physical drives of this size should have a huge impact on the design of
> consumer electronic devices. The device we know of as a CD player will
> eventually be replaced by a cabinet with a hard drive connected to a
> broadcast network.
Cool idea.
> A CD jukebox that holds 300 physical CDs will be no
> match for a drive-based music device.
You got that right.
> The hard drive model will have
> unlimited and immediate access to all songs and have a virtually unlimited
> number of features, as a computer program can serve as the user interface.
> Want to hear every instrumental ballad in your collection that is less than
> four minutes in length and was recorded before 1940? No problem. Would you
> like that in reverse chronological order or sorted by title? The jukebox
> will be stuck swapping disks inside a chamber the size of your dishwasher.
> Why have all those mechanical parts when they simply won't be necessary?
Sweet.
> Some people will challenge this hard drive concept on the basis that we
> will still desire some form of portable media. High-storage media from
> vendors such as Iomega or PCMCIA-based flash cards will easily address
> this. 1GB drives from Iomega and SyQuest are clearly fast enough to support
> audio, and these drives could currently store 1,000 minutes of encoded
> music. Looking forward, flash cards might make an even more appropriate
> portable solution as they can withstand more movement and shock than a
> spinning mechanical device. The thing to remember is that we are
> permanently crossing an audio threshold. If this pure bit-based model
> doesn't emerge in the next 12 months, it will most certainly emerge in the
> next 24.
I've been hearing this for years. But still, it would be nice...
> In case you still think this is all technological rhetoric, you might be
> surprised to learn that the next version of WebTV's Internet terminal
> actually has a 1GB drive, specifically included to take advantage of data
> broadcast.
Look, ma, Network Computer no more!
> Microsoft has established a relationship with Wavephore, which
> will help distribute data over the extra scan lines known as the vertical
> blanking interval. This VBI space is available on every channel and may
> prove quite valuable over the next few years. PBS has already licensed its
> VBI bandwidth, and you can expect others to do the same. It will be
> interesting to see if broadcast networks such as ABC can control the VBI
> space at their affiliates or whether these local stations might lease space
> out to third parties. Of course, keep in mind that VBI is really just a
> temporary solution as it can only support a few hundred kilobits per
> second. The real change will come from dedicated data channels that support
> 32 mbps.
Bring it on, baby!
> It is interesting to note that broadcast bandwidth is not as readily
> available as Internet bandwidth. Anybody can broadcast on the Internet.
> Local TV stations, cable companies, and satellite broadcast companies
> control almost all the available data broadcast conduits. As a result, we
> are likely to see quite a bit of jockeying as content providers try to
> shore up space. This will be a particular challenge to Internet-based
> content aggregators that have no real claim on a broadcast feed. On the
> other hand, this market could be a real boon to cable providers that could
> potentially support both Internet access and data broadcast over the same
> network. (See related column, "The data channel: Cable's ace in the hole")
That would be at,
http://www.news.com/Perspectives/wg/96/wg10_7_96.html
> Satellite providers have the most available data broadcast real estate but
> suffer from the fact that satellites alone will never support interactive
> services due to latency challenges.
This sentence really downplays the potential for hybrid approaches.
> Who are the big winners? There are actually several. First, storage makers
> will be thrilled with this new multimillion-unit market for hard drives as
> audio (and eventually video) storage devices. Second, companies that have
> been in this market for years, such as Wavephore and the Data Broadcasting
> Corporation, may finally realize the success they have always foreseen.
> (CBS recently entered into a business agreement with DBC.) Also, the
> consumer electronics industry will welcome a new technology that has the
> potential to totally replace the current installed base of CD listening
> devices.
You know who might also win? cellphone/pager products that can tap into
these streams and suck out cool data realtime...
> However, the biggest victor of all may be the spate of current television
> broadcasters who were granted free licenses of spectrum intended for HDTV.
> Recognizing the growing market for data broadcasting, these vendors will
> unquestionably reallocate this free gift, valued by some at over $70
> billion, toward more realistic and near-term uses like data broadcasting.
It's a nice dream. We'll see how close the prediction comes.
Leaving in the Web resources:
"Big pipes, bigger problems"
http://www.teledotcom.com/1197/features/tdc1197sonet.html
Intel's Intercast
http://www.intercast.com/
"IBM to unveil disk technology"
http://www.news.com/News/Item/0%2C4%2C16190%2C00.html
"The data channel: Cable's ace in the hole"
http://www.news.com/Perspectives/wg/96/wg10_7_96.html
"Data Broadcasting sets new focus"
http://www.zdnet.com/intweek/printhigh/112497/biz1124.html
"FCC gives license to HDTV"
http://www.news.com/News/Item/0%2C4%2C9404%2C00.html
"It's us against them...and them...and them"
http://www.teledotcom.com/0297/features/tdc0297cable.html
"Digital TV: It's all about the data"
http://www.teledotcom.com/1197/headend/tdc1197headend_dtv.html
"HDTV: The 'D' is for data"
http://www.teledotcom.com/0996/headend/tdc0996headend_hdtv.html
----
adam@cs.caltech.edu
If this is coffee, please bring me some tea; but if this is tea, please
bring me some coffee.
-- Abraham Lincoln