Month: January 2007

This neut debate is interesting, featuring Tod Cohen, George Gilder, Peter Huber, Lawrence Lessig, Paul McWilliams, and Andy Kessler at Gilder’s Telecosm Conference.

Lessig really stands out – he sounds like a twelve-year-old girl demanding an increase in her allowance, and the rest tell her to do her chores. That’s not much of an exaggeration. Gilder is also pretty incoherent, but the others have some interesting things to say.

And this talk between Ed Feigenbaum and Bob Kahn should open a few eyes. Kahn says he’s not interested in network neutrality (1:44 in), which to him is a destructive slogan. Networks are built on incremental improvements, and net neutrality stifles that growth. Here’s some quotes:

If it’s more efficient to do things inside the net than outside the net, I’m OK with that…Companies that make innovations should be able to control them…You really want to incentivize people to innovate…I’m totally opposed to mandating that nothing interesting can happen inside the net.

He says he doesn’t agree with his former assistant, Google’s Cerf, that innovation has to be pushed outside the net.

(Note: Bob Kahn is the guy who actually invented the Internet, not Al Gore, not Vint Cerf, not Jon Postel, not Bob Metcalfe, not Tim Berners-Lee. Bob Kahn invented the Internet and he’s against network neutrality. This will be on the test, so write it down.)

The title of this post is a slogan one of my fellow engineers at Texas Instruments had in his cubicle in the early 80s when we were developing the first network-based operating systems. Engineers understand that money doesn’t grow ion trees, the Tooth Fairy doesn’t leave it under your pillow, and that real stuff has to be paid for. It would be great if the laws of physics could be suspended and Our Internet could have all three virtues, but in the real world (the place where engineers are paid to make trade-offs) this isn’t going to happen. If you want your network to be fast and reliable, it’s going to be expensive; if you want it to be cheap, it has to be either slow or unreliable.

Two common networks make this trade in different ways: the telephone network is highly reliable and not very expensive, but it’s not fast. The packet network that we call the Internet is faster and more expensive than the telephone network, but it’s less reliable in the sense that it’s performance is more variable. Neither network is more virtuous than the other, they just make the trade-offs in ways that are appropriate to the ways the networks are used, to the kind of traffic they carry.

Our pals at Free Press, the organization that created Save the Internet and pays its employees, unveiled a new plan for the Internet at its annual conference this week. Free Press, by the way, is primarily concerned that deregulation of media ownership law a threat to our democracy, just as deregulation of Internet access networks is, you guessed it, a threat to our democracy.

Free Press’ plan for the Internet is couched as Declaration of Internet Freedoms and goes like this:

* Universal Affordable Access: Broadband Internet access should be universally available and affordable.

* An Open and Neutral Network: Access to broadband networks should be open to all producers and consumers of Internet content on fair and equal terms without discrimination.

* World Class Quality through Competition: America must build the world’s most advanced communications networks and maximize competition on those networks.

That sounds an awful lot like “Cheap, Good, and Fast: I want all three.”

I’m sorry, but that’s not how things work on the planet Earth. Upgrading the overall quality of our broadband networks is going to take a large infusion of cash, to the tune of several billion dollars. That money will be spent running fiber optic cables, buying routers to light the fiber, building network operations centers to spot and repair problems, upgrading routers inside homes and businesses, and purchasing long-haul fiber optic circuits. The Internet doesn’t generate enough carrier revenue to pay for this upgrade by itself, so the carriers will have to pay for it in no small part by selling the stuff that people are willing to pay for, television and phone service.

That’s the way we’ve always paid for infrastructure in the US, and it’s the way it’s paid for in other countries as well. You can get a raw connection rate of 100 megabits/second in Korea, but you can use any VoIP but the one provided by Korea Telecom. Similar rules pertain in the UK and in France. That’s a violation of network neutrality, but it takes place in the places Free Press touts as Model Nations.

The net result, pardon the pun, of the regulations proposed by Free Press on the American Internet access network can only be one thing: the network will never be substantially upgraded because there won’t be any money in it.

And that’s what happens when you go for Good, Fast AND Cheap: you got none of them.

Matt Sherman notices Stanford University law professor and copyright crusader Larry Lessig expressing doubts about some of the regulatory jihads he’s embraced recently. Lessig admits to being wrong about the Microsoft anti-trust case because it was rendered moot by the relative success of Linux. The professor is less certain he was wrong about net neutrality, but his mind seems to be opening just a bit.

Unfortunately, Lessig is still caught up in a false dichotomy between government-regulated networks vs. government-operated ones:

Those who oppose network-neutrality regulation should also oppose this regulation of last-mile broadband’s most important competitor. Municipal competition won’t kill commercial broadband any more than Linux has killed Windows. Yet it could change the business model of last-mile broadband, just as Linux has changed the business model of Microsoft. If there’s going to be a Linux-like miracle to counteract innovation-threatening broadband business models, then, at a minimum, miracles must not be a crime.

If the only competition AT&T and Verizon had was government-operated networks, their stock would outperform the markets to an extreme degree. In fact, their real competition can really come only from other for-profit companies smart enough to hire good people and develop sensible business plans. The one tip I’d give regulators is not to allow telcos to control too many different media. Licensed WiMax is a big threat to residential broadband over twisted-pair, so it’s best not to allow telcos to have a large piece of that action.

Ultimately, the free market will keep AT&T’s rapacious instincts in check better than any half-baked regulatory scheme, but only if it’s allowed to develop. For that reason, the provision of the AT&T/Bell South merger agreement about WiMax divestiture was good; the rest of it is garbage.

A week has passed since AT&T submitted to the will of Democrat FCC commissioners Copps and Adelstein by volunteering to “network neutrality” regulations and we’ve seen the entire range of possible reactions. Not from the pro-neutrality vs. the anti-neutrality forces, mind you, but from within the ranks of the regulationists themselves. Free Press’ chief telecom wonk Ben Scott touts the agreement and his own press blurbs on the Astroturf Save The Internet blog and Columbia law professor Tim Wu gushes about the “historic” nature of the agreement in the same venue.

But all is not happy in Neutland. Another law professor and neutrality hard-liner, Susan Crawford, says the agreement is the end of the Internet (copy paste http://scrawford.blogware.com/blog/_archives/2006/12/29/2604993.html into your browser, she won’t allow direct linking from here), and late-breaking neutarian Tom Evslin says the agreement is a sham. Others on the pro-regulation side say the agreement is hollow as the FCC’s majority doesn’t intend to enforce it.

So depending on which virtuous, regulation-happy soul you want to believe, the AT&T merger agreement is either the best thing that’s ever happened to the Internet, the worst thing, or nothing at all. That’s a lot to choose from, but don’t worry, I can explain everything in a very few words.

If you read Ben Scott’s quotes, it’s clear that he’s happy about two things:

1) The merger agreement contains a definition of “network neutrality”; and
2) It imposes telecom network regulations on at least part of the AT&T DSL network.

The first is really nothing, because the definition used in the agreement was simply lifted out of the Snowe-Dorgan Internet Freedom Preservation Act and has no special significance beyond the fact that the Democrat commissioners were familiar with it. Scott claims it puts the lie to the claim that “nobody can define network neutrality”, but that never was the issue, really. The intellectual problem with neutrality is that everyone can define it, generally in many different ways. Some say it’s about routing, some say it’s about delivery times, some say it’s about unbundling and some say it’s simply about pricing. As the agreement contains many of these elements, we’re still no closer to a universal understanding of network neutrality than we were when Professor Wu coined the term a few months ago.

And that, I believe, explains why Wu is so giddy over the agreement: it ensures that the term he coined will be part of the discourse on Internet regulation for years to come, extending his fifteen minutes of fame. That’s good for book sales, speaking gigs, and tenure. And more power to him for that, it’s no small feat to manufacture a regulatory principle out of thin air and have it stick to a major merger agreement.

The reason Ben Scott is so happy is because the agreement applies telecom regulations to the Internet. He’s a telecom regulator, you see, not an Internet guy, so he was facing unemployment if the trend toward de-regulated networks embodied by the historical Internet continued. But now he believes he’s made the case that the Internet is no different from an old-fashioned telephone network (in a regulatory sense) so he’s still relevant.

And this is the reason Evslin and Crawford are unhappy. They’re Internet people, not telecom people, and they understand enough about the Internet to see that it’s not at all like a phone network and can’t be treated that way. They’re all for regulating it in principle, but skeptical about the applicability of Scott’s wisdom to a fundamentally different medium. So they naturally point out that the telecom wonks who forged the agreement have missed the point so badly that they’ve actually made things worse for the Internet.

So this is the bottom line: the AT&T merger agreement applies traditional telecom law to the Internet, which is either a good thing or a bad thing depending on whether you think the Internet is phone network.

Just so we’re clear, networks of the past were designed for a single service: the telegraph network only carried telegrams, not phone calls; the telecom network wasn’t designed for television, and the CATV network, in its original form, didn’t carry data or voice. The Internet is a packet data network that fundamentally is capable of serving a variety of needs without much effort: e-mail, porn, video, and voice are all happy on it, more or less.

If we’re going to get regulatory with the Internet, we’re not going very far until we recognize this unique character. It’s not enough to be strict about how users of a single service are treated relative to each other, we have to understand how different services relate to each other. The telecom wonks’ approach is simply to ban “different services” in order to force-fit the Internet into the Procrustean Bed of the traditional regulatory model. The disgruntled neuts are sounder in their reaction to the agreement than the happy-faced ones, more in tune with the Internet’s character and less reactionary.

To be perfectly frank, Ben Scott and Tim Wu are rank opportunists who obviously care much less about Your Internet than about Their Own Careers. That’s not a bad thing, but it is something you should bear in mind lest you find their jubilation infectious.

UPDATE: Matt Sherman surveys yet another reaction.