There Is No Cat - Good luck; you’re going to need it

Tuesday, January 21, 2003

Good luck; you’re going to need it

I’ve read David Weinberger’s Open Spectrum FAQ, as well as the paper he wrote with David Reed, Jock Gill, and Dwayne Hendricks, and I have to say, I think they’ve got stars in their eyes. It’s a seductive premise, this idea that spectrum is unlimited if it’s used properly and with new technologies. It may even be true. But to expect it to be implemented any time soon flies in the face of human nature and history.

One thing that all futurists seem to forget is that Infrastructure Is Hard. So many of the infrastructure-intenstive disruptive technologies that we’re so enamored of today took freaking forever to build out. The telephone network took over 100 years to reach virtually everyone. Cable television distribution has taken 55 years to get to the point its at now. Satellite distribution has taken 40 years; Telstar launched in 1962. Cell phones have taken at least 25 years to get to the point where even I have one. The Internet became an overnight success after being built out for 30 years. This stuff takes time.

The result of this extended build out is that there is a ton of legacy equipment out there that works under the assumption of spectrum scarcity. That’s your installed base. Radio has taken 100 years to get to where it is; practically speaking, though, we’re talking about approximately 75 years worth of equipment in place. I can turn on a radio created as early as the 1920s and it still works and receives today’s signals. As I write this, three feet to my right is a Zenith radio made 50 years ago, in 1953. It sounds magnificent, better than most radios made today. There is a sh¡tload of this stuff out there, and it works. That’s a powerful incentive not to upgrade.

The FCC has run into the legacy problem with its attempts to foist digital TV on the public. They want to sell a portion of the spectrum currently occupied by television broadcasters. Congress has been counting on it in their budget calculations. By 2006, they want to switch off the current stations. Customers have greeted this with complete indifference. Current television is good enough. Why the hell would I want to buy a new television when the three I have work perfectly fine and provide an acceptable picture? Why should people accept the forced obsolescence of perfectly functional equipment? There is no way in hell broadcasters will be able to switch off their analog signals by 2006. The public outcry would prevent it. Why? The impact of the installed base. 2016, maybe.

Inertia is the second most powerful force in the universe. (Entropy is the first, which makes me wonder why it’s the Second Law of Thermodynamics that talks about it....) A body at rest tends to stay at rest. The public seems perfectly happy where it is, and the cost of getting them to move is enormous. It’s not going to happen quickly.

Weinberger et. al. talk about software-defined radios as if they’re magic. Maybe they are. But I’m not so sure. I’ve played with any number of DSP-based processors that claim to reduce interference, from outboard units that work on the audio after the radio’s finished with it to expensive radios costing thousands of dollars that do all of their processing in the digital domain. Software is not a panacea.

It’s true that spread spectrum improves that. The FAQ mentions BLAST, an approach developed by Bell Labs that takes advantage of interference. I’m familiar with it; I posted pretty much all the press releases about it on the Bell Labs web site back when it was my job to do such things, and I made a habit of reading the releases and looking further into the things they mentioned when they interested me. It’s fascinating, a radical change. But as the New York Times article David Weinberger links to points out:

Because of economic problems, the wireless industry has been slow to adopt even 3G networks in the United States. So Blast is unlikely to become available soon.

Spectrum is a funny thing. Different segments of it work in different ways. Frequencies up to about 50 MHz have the ability to travel around the world, something I’ve certainly taken advantage of in 25+ years of listening to shortwave radio. At UHF and above, communication is pretty much limited to line-of-sight. Of course, that line may include the moon, or the artificial moons known as satellites, so they’re still useful over long distances. Go high enough in freuqency and you find that rain drops start to cause reception problems, because the wavelengths are so small that the rain drops are bigger and cause a significant disruption. 20 MHz used in the unlicensed 2.4 GHz band has very different characteristics than 20 MHz in the HF bands down below 30 MHz. If you had an application that required jumping around over a band of 20 MHz and you tried to do at at HF, at least half of your packets wouldn’t get through, because propagation on different segments of that band works very differently from other segments. Some parts work best at night; others during the day; still others are in-between. Some spectrum is relatively easy to work with, such as the frequencies around 700 MHz that the FCC is trying to reclaim from television for the cellular industry. Other spectrum requires expensive electronics and loss-preventing antennas and cabling to work well, such as the Ku-band satellite signals up above 10 GHz. This is a basic law of physics. The result is that you may want to use different areas of the spectrum for different projects for which the spectrum characteristics are useful. Shortwave is very useful for international broadcasting, for example, but not so much for computer networking. If you fail to take this into account, you wind up with chaos. We’ve only got one spectrum. If you’re going to piddle all over it, you better make sure you’ve got the floors covered with newspaper.

The spread spectrum technologies that have the two Davids so excited are fascinating. They have a lot of promise. But it’s going to take a long time, probably decades, before they have the kind of impact that’s being promised. So keep plugging away. It’s a long row to hoe. Don’t expect this battle to be won any time soon. This is a massive change. It will not happen quickly.

Posted at 2:07 AM

Comments

Thanks for the long and thoughtful response to our FAQ.

Your analysis fails, however, to take into account the existance of inflection points. The internet grew slowly for 20 years and then hit an inflection point and BAM. How many URLS in 1992? How many URLs in 2002? Non lnear transitions can happen. This is not to say they have too or that we are in one at the moment with respect to Open Spectrum.

Also legacy systems can die very sudden deaths. Anyone remeber X.400 email? It was a US Government mandate in 1992. It died a sudden death in 1993.

If Open spectrum can be shown to deliver the benefits we believe it will offer, then the power of the market will determine the speed of its adoption and diffussion - powered by end user purchases, not the US Gov, not VC funds, not Corporate R&D. This need not be a linear event.

Regards,

Jock Gill

Posted by Jock Gill at 10:45 AM, January 21, 2003 [Link]

Thanks for the commentary.

Your points are quite important. Legacy systems stand in the way of much better systems.

But the reason legacy radio systems "work" so well is because they are very inefficient compared to what we now know how to do in wireless and wired communications.

Instead of allowing competitive innovation to improve the system, the history of the FCC has been to protect the investments of the past. Had this form of government regulation been tried with automobiles, we would have had to have automobiles that were limited to behave like horses (limits on speed, exhaust converters that created piles of manure-like sludge on the streets, burning hay instead of gasoline, etc.)

You can’t have innovation without obsoleting systems that are already in place. Give the users (rather than spectrum "owners") the choice of what technologies to buy for getting content, communicating, and locating things, and so forth. It’s possible now to do that - that’s what the plasticity of software-defined radio networking lets you do.

The legacy systems that really "work well" can stand the heat.

Very few people still love vinyl and tube hi-fi, or float carburetors in cars. I may be sad to see an era pass, but the value of those things is now entirely sentimental.

I’m not sad that the Apple ][ has become a historical artifact either, despite the fact that I spent a lot on programs for it, and would have made a lot more money from programs I wrote for it if a Federal Computer Commission had decided to protect my incumbent interests from new technologies that were incompatible.

PS: I’m not specifically enamored of spread spectrum, nor of UWB, etc. I am frustrated that these useful techniques have been blocked from deployment, even where they have little or no negative impact on legacy systems (e.g. in rural locations, and at power levels that have no effect on legacy receivers).

Posted by David Reed at 11:54 AM, January 21, 2003 [Link]

Thanks, Jock and David, for your comments. Let me address your points.

Legacy systems and quick deaths. Yes, I remember X.400. I worked on the documentation for a LAN product for AT&T in the late 80s based on OSI, and as I recall, X.400 was the basis of the system’s e-mail. X.400 may have been a government mandate, but it never gained traction. It never had the kind of installed base that would have made it difficult to displace. RFC 822-based e-mail had been around for years at that point, and it ate it for lunch. So using it as an example of how legacy systems can die suddenly doesn’t map very well to the example of radio, which has a huge, widely adopted installed base.

Inflection points. The web also didn’t have to deal with a huge installed base. What was it replacing? Gopher? Command-line ftp? Xanadu? The web grew at a tremendous rate, sure, but there was nothing standing in its way.

Both of you deride the influence of government in spectrum allocation and promote the use of the market. I’ve got two words for that: AM Stereo. The FCC left the choice of a standard up to the market. We wound up with what, four incompatible systems? The one that was technically superior, Kahn, lost out to Motorola’s marketing dollars for their inferior, nausea-inducing C-QUAM. Receivers had to implement decoding for four standards. The result was that AM Stereo died on the vine thanks to the FCC’s reliance on market forces to make a choice. The market chose monophonic sound. I don’t see the market as the panacea that others do. Sometimes it’s appropriate. Sometimes it’s not. Market fundamentalism seems to me to be almost as big a mistake as the faith the Soviets had in the command economy.

I understand that replacing legacy systems necessarily obsoletes existing equipment. But I think you underestimate the extent of the impact that will have. Forcing users to change imposes huge costs on a set of legacy users that basically consists of everyone everywhere. The computer industry may be totally enamored of the possibilities of open spectrum, but the costs are borne by everyone else. I don’t see the computer industry lining up to cover the costs; that will be up to the mass of consumers, the governments that use radios for public service communications, the industries that rely on existing systems....

If you look at what’s happened in the ISM bands at 2.4 GHz and 5 GHz, you could say that it’s been an initial testbed for unlicensed and open spectrum. So far, it seems to be a huge success. I agree, David, that it would be good to expand this sort of use in areas where the impact would likely be minimal. I would like to see that happen. But I get the impression that you’re calling for a wholesale change to the way spectrum is handled, and I don’t think that’s practical or desireable yet. ISM is proof-of-concept. Next step would be controlled introduction. Some way down the road, if the controlled introduction is a success, then you broaden it further. If you look at this as a long campaign, it’s probably appropriate at this point to start pushing for more.

Theoretically, I like the idea of open spectrum. I particularly like the idea of returning the spectrum to its rightful owners, the people, rather than the spectrum "owners" (actually leaseholders; the spectrum is supposed to belong to all of us). But the obstacles are formidable, and I think the four of you who wrote the original article tend to minimize them. Long-term, minimal-impact spectrum use is a winner. But getting to that point is going to be painful and costly and take much longer than I think you suspect. The UK originally broadcast its television at VHF frequencies using a black-and-white 405-line system. They introduced a superior 625-line system, with color, at UHF frequencies in 1969. They weren’t able to turn off the 405-line transmitters until 1985, and even then there were people still watching the old system.

The FAQ and articles seem very sanguine about the prospects for open spectrum. But I think the reality is that it’s going to take a long time, and may never be fully implemented. It’s a great idea. My closet is full of great ideas.

Posted by ralph at 11:46 PM, January 21, 2003 [Link]