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

<< [ Clueless in Seattle ]   |  Out of Sight, Out of Mind  |   [ TV Oaxaca ] >>

Weekly Column

Out of Sight, Out of Mind: Big Old Stupid Telephone Companies Are Throwing Away Their Only Real Asset

Status: [CLOSED]
By Robert X. Cringely
bob@cringely.com

The Internet Era, if that's what we are in today, is by definition also the era of packet switched networks. When Len Kleinrock was designing the first U.S. packet network at UCLA back in the 1960s, I doubt that he had in mind using the technology to replace the regular phone system, but that's certainly the kind of talk we hear today. Starting a decade ago with George Gilder and now getting louder and louder, we're hearing that the circuit-switched network is obsolete and packet switching is the way to go. And this idea has been embodied in a lot of corporate decisions, most recently this week when SBC announced that it will be spending $7 billion on a network upgrade -- all of it for packet switching. Now I don't often argue with George Gilder, but this time I will, because I think telephone companies the world over may be making a huge mistake by embracing packet switching so completely.

The competing technologies are packet switching and circuit switching. Packet switching is a dumb network flowing with all these packets of data that don't know what they are or who they work for. The beauty of packet switching is that if I need a lot of packets or a few, I can have just what I need, with all the of the intelligence of the network running at my house and at some server, but almost none of it at the phone company that connects us together. Circuit switching, on the other hand, is a network that knows what it is doing: It connects telephones to telephones. The intelligence is in the network, and the edge devices are stupid old telephones or computers pretending to be telephones. Call-waiting, call forwarding, call blocking, distinctive ringing, voicemail -- these are all applications of the network, not the telephone, so they work just as well on a receiver from the 1930s as they do on a brand new telephone today. Instead of swarming like flies, circuit-switched data is provisioned in impervious chunks. A voice circuit is 64 kilobits-per-second and a T1 (now DS1) trunk line is 24 voice circuits.

The arguments against circuit switching come down to these: 1) A smart network costs more to maintain and an aging smart network is even worse; 2) By dedicating 64 kbps to each call, bandwidth is wasted; 3) By limiting a voice circuit to 64 kbps broadband services aren't available as an integral part of the network (DSL, for example, is rather duct-taped to the phone network), and; 4) A circuit-switched network can't support neat new services without years of planning by the telephone company.

Those are a lot of arguments against circuit switching, but there is one huge argument in its favor -- the circuit-switched network is already installed and paid for. This was always the weapon wielded by the big telcos to keep competition away. "We have this $100 billion network that reaches into every home in America." To compete with the telephone company always meant spending $100 billion. But, guess what, it happened. The cable television systems in America reach nearly every home and so do the mobile phone networks, and in both cases if you add-up the total cost for those build-outs, it comes to roughly $100 billion each. After decades of legal monopoly and long lunch hours, the telephone companies suddenly have plenty of competition -- enough so that most of the profit has been driven out of the phone business. So their answer is to scrap the network they spent $100 billion building and start over. Not a good idea.

What the telephone companies have going for them is that they already offer universal service and their network is already paid for. But now they have lost confidence, have read too many articles by George Gilder, and have been taken out for too many expensive dinners by packet network vendors. So they no longer value their main asset, and in many cases appear willing to give it away or even abandon the network completely.

While to some this may seem the only logical move, to me it seems like lunacy because the telcos have no idea how to compete in an unregulated world. So giving up their only advantage is tantamount to committing corporate suicide.

There is a better way, but the telcos just can't see it. And for this I blame their lack of technical depth. There used to be Bell Labs and then Bellcore, but now the telephone industry has no lab of its own, no indigenous technological depth. They only know what their suppliers tell them, and their suppliers see more profit in packet switching.

The telcos need to pick their battles, which to me means that they need to concentrate on two main technologies -- one existing technology and one new one. The existing technology is plain old telephone service, which they should be able to keep competitive with VoIP by controlling costs. The other technology is entertainment -- music and video. Everyone from Microsoft to Intel to Apple sees digital content delivery as the next big business, and I just think that the telephone companies and their circuit-switched networks are in a prime position to dominate that industry.

Huh?

Other than the fact that it is already paid for, the advantage of a circuit-switched network is its massive total bandwidth. While 64 kbps per circuit seems pitiful in a world of DSL and cable modems, if viewed from the perspective of the total network, it really isn't, because all of those 64 kbps channels can operate at the same time. Broadband providers typically provision about 11 kbps per broadband subscriber, meaning the circuit switched network actually has six times the total capacity of a comparable packet network.

Imagine a scenario where every broadband user wanted to use their connection to watch video and they all wanted to watch different videos. What would happen? Why the network would fail, of course. Routers and servers alike would be overwhelmed, the network would slow to a crawl, and most people would see nothing at all. Now do the same thing on a circuit-switched network. If the server capacity was there, the network would run just fine. And server capacity at today's prices really isn't a big issue, either; it's just that we provision servers based on the bandwidth limitations of the packet network.

If I set myself up as a Competitive Local Exchange Carrier (CLEC) in California, and my goal was to offer digital entertainment, how many points of presence (PoPs) would I need to serve California's 30 million-plus couch potatoes? Hundreds? Thousands? Try seven, because as a CLEC I can count on the ILEC (Incumbent Local Exchange Carrier -- the old monopoly phone company) to provide FREE back-haul for all calls within one of California's seven Local Access and Transport Areas (LATAs). So with seven data centers and no thought at all about bandwidth, I can support all of California. There are approximately 200 LATAs in the U.S. so that means 200 data centers to cover the entire country, but remember that by "data center" I mean some racks down at the phone company building.

At this point a few thousand nerds are getting furious with me because I pointedly (and stupidly in their view) am pretending that 64 kbps is enough bandwidth to actually do something, which of course it isn't. How can I propose to deliver video or even audio with that kind of bandwidth?

Audio is possible today. Enhanced AAC encoding can send CD-quality audio down a 24 kbps connection. The sticking point is video. Conventional thinking says audio takes 10 times the bandwidth of text and video takes 10 times or more the bandwidth of audio, so why the heck am I talking about sending video over telephone lines?

There is a saying in Brazil: "He who doesn't have a dog hunts with a cat." The phone companies should stop trying to replace their cat with a dog and instead teach their cat how to help with the hunt.

Here is a $100 billion asset (the circuit switched network) that is paid for and in place. The phone companies are seriously thinking about just ripping the darned thing out, taking the write-off, and replacing it with another imperfect technology that will cost $100 billion more. That's a $200 billion decision or about $2,000 per U.S. household.

What could you do with $200 billion? Could you develop and install the technology to send DVD-quality video over a 64 kbps phone line? I think you could and for a lot less than $200 billion.

We're not talking here about making a network for the quick download of new Linux distributions. You can't make bandwidth out of nothing. We are talking about creating an application for transmitting video-on-demand over a circuit switched network. Circuit switching inherently gives you all the advantage of on-demand, like the ability to view a show when you like, to fast-forward, stop, or reverse it. Yes, those capabilities are available on other networks, too, but not if everyone does it at the same time and especially not is everyone is watching something different, which is actually pretty much impossible on those networks but they never mention that part.

So the real sticking point is "only" cramming video down that skinny wire.

Fortunately, we have long taken the wrong approach to video encoding, seeing it primarily as an exercise in transferring data rather than transferring images. Though we tolerate lossy images -- images that can't be faithfully recreated in their original form -- with codecs like MPEG-4 and H.264, we really haven't paid as much attention as we might to replicating the way we actually see. There is bandwidth savings, for example, in recognizing that we apparently get a lot more of our image from luminance levels than chrominance so we can sample the color at some lower resolution and get away with it, but that's not going to give us "Gone with the Wind" down the phone line.

What we need to emulate here is the eye, itself. Look at the optic nerve that connects the retina of your eye to the visual cortex of your brain. The optic nerve is composed of approximately one million stringy cells called ganglia that fire in parallel over a distance of two to three centimeters with the actual visual signal travelling at about 4,400 feet-per-second. Taking into account recovery time between signals, the optic nerve has a total bandwidth of approximately 100 kbps.

100 kbps!!!! That's all the bandwidth we have available to appreciate HDTV and IMAX? We see the world through a pinhole that small?

Yup. So sending DVD-quality video down a 64 kbps line shouldn't be impossible at all.

All an SBC or Verizon has to do is take maybe a tenth of that $7 billion and apply it to a crash program in understanding the physiology of eyesight and emulating it in silicon. After all, what is our retina but a video encoder, our optic nerve but a network, and our visual cortex but a video decoder? Come up with a new encoding/decoding process, a new storage process (stored in compressed form this video would be vastly more compact than anything today -- you could put 100-plus movies on a DVD), and build it all into an appliance packed full of DSPs and priced like a video game console.

Oh, and there's another little side benefit -- the end of blindness.

And it is doable, the algorithms have already been worked out and are running today in Matlab. But it will only happen if the phone companies stop whimpering and start thinking.

Comments from the Tribe

Status: [CLOSED] read all comments (0)