Politics

Wi-Fi Planet Keynote: Wi-Fi vs. Telcos

In his keynote speech at the Wi-Fi Planet Conference & Expo, "Wi-Fi vs. Telcos", a famed Internet guru spelled out why he's looking to nimble wi-fi providers, not "telcos, cablecos, and cellcos" for innovation and the next killer app.

by Alex Goldman ISP-Planet Managing Editor [July 26, 2005]

David Isenberg earned permanent cred when, in his essay "The Rise of the Stupid Network" he told AT&T that they were building the wrong network, that the network of the future would be flexible, changeable, and programmable, unlike the proprietary, closed, and purpose built POTS network. He called the network of the future the "stupid network." It was 1997 and he worked for Bell Labs, who immediately tried to suppress the report (and failed). He now runs an independent, eponymous prosultantcy.

So at his keynote at the last and final Wi-Fi Planet Conference & Expo he went back to first principles and spoke about the future of the internet.

Dinosaurs in trouble
Isenberg said, "three years after the acute phase of the telecoms crisis, companies continue to struggle. There is no controlled or planned rollout of products and services. There is no 20 year or 30 year capital investment. Companies are rolling out small mobile devices that are replaceable and cheap."

The biggest companies, he said, are in trouble. "The dominant telcos, such as SBC, Verizon, and Qwest, are losing their base, sitting on obsolete networks. They bet the farm on SONET networks built with bond financing. Investors bought the bonds as if they were asset-backed securities but the assets are worthless."

Other companies are doing no better. "CLECs are up in smoke. They were created by a figment of Congressional imagination, as if competition could replace a state sanctioned monopoly. We're seeing the end of the rest of telecommunications."

"So they're getting into television," Isenberg laughed. "The head of the FCC recently called TV a vast wasteland, with CNN covering Disneyland and a pop singer. TiVO is TV's biggest innovation, and it's not an innovation any of these companies are responsible for. The Internet proves TV commercials are not worth paying for."

Isenberg said that the cable companies would never improve the vast wasteland. "The cable TV companies are risk averse, they dislike competition, and I believe that even cable modems are about to hit an invisible ceiling of corporate fear that will ensure they never deliver TVoIP."

Obviously, Isenberg voted for change. "The alternative to the cable companies is unleashing all of the human spirit. If we had a choice, we wouldn't settle for blind dates in hot tubs. The pipes would be full if there was a freedom to distribute, but will it happen?"

The cellular phone companies wouldn't do it either. "The cellcos are now struggling too. Their ARPU is static. They love the walled network and will never let their 'value proposition' get away from them. Their idea of a killer app is a downloadable ringtone. Now the cellcos are going to do TV," he giggled. "They're going to bring the vast wasteland to the tiny screen in our pocket."

"If they survive it will be a detriment to the larger economy."

Smart networks die when they cannot adapt
Then he mentioned his theory, the idea of the stupid network. "They are failing because there are better alternatives and because of their fatal addiction to the smart network.

He said that any network design should adhere to the end to end principle as described by Saltzer, Reed, and Clark. Isenberg summarized the principle in a single sentence, "if you can do something in the MIDDLE of the network or at the EDGE, do it at the EDGE."

Isenberg's use of the term "stupid network" is confusing to some people. In his essay (above) he describes trying to change the way voice was transmitted over the network and being prevented from doing so by network elements that were not programmable. A "stupid network" is one that is programmable. It can be programmed to make way for new applications. It's the difference between a computer program that uses constants and one that uses variables.

The meta goal here is simple, "any application over any network."