Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ Y2K, That is the Question ]   |  You Can't Get There From Here  |   [ Cooking the Books ] >>

Weekly Column

You Can't Get There From Here: We Are Years Away From Watching Movies Over the Net

Status: [CLOSED]
By Robert X. Cringely

There was a front page story this week in the Los Angeles Times about wheeler-dealers cruising the Cannes Film Festival trying to buy Internet distribution rights to movies. The underlying assumptions were that Internet distribution of movies is just a couple years away, and that buying the rights now would keep the price low while producers are still generally unaware of the real value of what they'd be selling. There is little doubt that sending movies over the Net is possible, but will it really be economically viable in two years? I don't think so.

There is a general misunderstanding about the state of the Internet, how it really functions and what is really possible to do with it. This all has to do with connections, bandwidth, and network management. Let's start by defining some terms.

Dark fiber. This is a very popular concept., the idea is that there is far more fiberoptic cable strung around the world than the world actually needs. Some of this fiber is "lit" and some of it is "dark." If we were to only light the dark fiber, there would be so much bandwidth available that HDTV two-way videoconferencing would become the norm. Yes, there is a lot of dark fiber, but lighting that fiber is not cheap or easy, and is certainly not a simple matter of turning a switch. Lighting all that fiber would mean installing equipment costing tens of billions of dollars and incurring, from that moment on, a considerable maintenance expense. If it were just a matter of flicking a switch, that fiber would already be lit. To most of us dark fiber is no more useful than the road behind us, or the money we used to have.

The last mile. This refers to the actual connection between your house or mine and the telephone or cable TV network. For almost all of us, the last mile is not dark fiber but copper wire. AT&T, having thrown away as useless its last mile connections in the 1983 divestiture, has in the last year invested almost $100 billion in cable TV companies trying to regain that last mile. Ownership of the last mile is what makes telephone and cable TV companies worth so much money. Find a new way to cover that last mile (until lately, AT&T was planning to go wireless to do this), and the value of cable and phone stocks will go in the toilet.

Cable modems. These boxes from @Home or Roadrunner connect to the cable TV network and bring ultrafast Internet connections at low cost. It's true that the cost is low, but is the connection speed really as fast as people think?


There are two significant limitations on the speed of Internet connections through cable modems. The first limitation is the number of cable modems installed on a given segment of cable. This is a party line, a shared connection, and the more users connected to each string of cable, the less bandwidth available to each user. This is precisely analogous to Ethernet, where we moved to hub topologies and then to switches precisely to isolate each user onto his or her own 10 megabit-per-second network connection. It is possible to do the same kind of segmentation on the cable TV system, but most operators aren't doing it because it costs too much money. This means cable modems will only get slower over time. The second limitation of cable modems is the actual connection between the cable company and the Internet. You might get six or eight megabits-per-second between your house and the cable head-end, but what good is that if the actual connection to the Internet is a 1.544 megabit-per-second T-1 line that's shared with 10,000 other users? There's a lot that can be done with caching to improve apparent throughput, but this is never going to help streaming video. It is simple economics that there is no way a cable operator can sell you for $10 per month a fulltime Internet connection that costs $1,000 per month to provision. Something has to give.

ADSL. Here is the telephone company's version of a cable connection. The specification says that anyone within 18,000 linear feet of their telephone central office ought to be able to download data at speeds up to 1.5 megabits-per-second, which is what's required for a VHS-quality MPEG-1 video stream. But the truth is somewhat different. My office in San Mateo is 11,000 feet from the central office and the fastest DSL connection I can buy is 784 kilobits-per-second. This is based on actual testing with several different vendors. And even if you are lucky enough to get the promised 1.5 megabits, there is the same problem of shared access to the Net, itself. My DSL provider is Northpoint, and my ISP is Brainstorm Networks. The connection between Northpoint's rack in the central office and Brainstorm's network operations center is a T-1 line that I have to share with every other Brainstorm user connected to my CO. If I'm using the whole T-1 to watch "Wayne's World," what do those other people do for connectivity? Good question.

MPEG-4. This is the emerging new standard in video data compression that may finally enable something like acceptable videos over the net. Originally aimed at low-bandwidth modem connections, MPEG-4 has become a solution for connections at almost any speed, or will be once the specification is finalized and MPEG-4 products actually start to appear. Until then, don't expect to watch movies on the Internet unless you have at least a 1.5 megabit-per-second connection. MPEG-4 is not here yet.

Quality of service. This is the dream scheme of Internet access. Right now, no matter how big your Internet connection is, there is no way to enforce that bandwidth end-to-end with no bottlenecks in-between. Emerging standards like the RSVP protocol and ATM offer the possibility of declaring "Wayne's World" packets to be higher priority packets and to therefore get them preferential treatment. But no quality of service guarantee is worth anything if one or more of the network segments between here and there is smaller than your Internet pipe, which happens more often than your ISP would like you to know.

Jitter, latency, packet loss, rebuffering. Here is the grim reality of sending movies over a network that works much better for e-mail. When I was designing digital video decoder boxes for Fujitsu a few years ago, the idea was that there would be a giant nCube video server sitting right there in the central office, one hop away. Today's reality says the video server is at least a dozen Internet hops away and each of those hops is an opportunity to add latency, lose packets, or generally get things out of sequence. And the truth is that right now, there are powerful forces working to keep the network as crazy as it is or even worse. This is because Cisco Systems and its competitors make a lot of money from that complexity. The more hops there are, the more routers there are, and the more revenue to Cisco. This is not going to change in the near future.

Of course, the Internet is always getting faster and better, and there is likely to be a time when it finally is practical to watch movies over the Net. For that to happen, we'll need improvements in all the areas mentioned above, and there will have to be additional time for technologies and products to stabilize. How long will this take to happen, given current market trends? My best guess says six years.

Keep this in mind while buying those Internet video rights in Cannes. And remember to use sunblock.

Comments from the Tribe

Status: [CLOSED] read all comments (0)