Debate: Smart Vs. Dumb Networks

I've just finished sitting through a very bizarre (for lack of a better word) session at the NGN Conference where two panelists directly debated the notion of smart (central, AT&T-like) networks vs. dumb (deregulated, open public Internet-style) networks. Here are...

November 3, 2004

3 Min Read
Network Computing logo

I've just finished sitting through a very bizarre (for lack of a better word) session at the NGN Conference where two panelists directly debated the notion of smart (central, AT&T-like) networks vs. dumb (deregulated, open public Internet-style) networks. Here are a few highlights from what turned out to be a very spirited debate.

Representing Stupid Networks: David Isenberg, Principle Prosultant (sm) with

Definition: Intelligent networks:

  • Application-specific

  • Application-aware

  • Run by experts

  • Centrally administered

  • Centrally planned

  • Resource optimized (QoS, load balancing)


  • Ridged

  • Expensive

  • Inflexible

Quote: Nicolas Negroponte: "Some bits are worth more than other bits."

We should optimize what's plentiful. We should build a network of the future with what scales--a big, fat-piped stupid infrastructure.

Is there enough raw material out there? Oh yes. Just look at these two laws:

Gilder's law (transmission capability)
Moore's Law (transistor density)

Definition: Stupid Networks

  • Are general purpose networks (not application aware)

  • Separate networks from applications

  • Are innovation-friendly

  • Not centrally planned

  • Have unreliable physical and link layers (but reliable where it matters)

  • Have explicit costs (buy as many 9s as you need)

90s wins: e-mail, e-commerce, Web, audio on demand, instant messaging, blogging, telephony, all were created at the edge of the stupid network.

Intelligent Network Failures: AT&T true voice, three years and $800 million to boost 3db to make voice sound better. And it broke applications (speech recognition).

Stupid Network Success: Skype: twice the audio bandwidth, runs over unmanaged public network, uses an echo canceller at the edge of the network.

The stupid network is built on the end-to-end principle: If you're doing something in the middle of the Net or at the edge, do it at the edge.

Quote: "Googin's paradox: The best network is perfectly extensible, and perfectly capital repellant."

The politics of the stupid network:

  • Freedom of speech

  • Participation

  • Pluralism

  • Diversity of opinion, believe actively

  • Competitive markets

  • Freedom to success or fail on merit

But not:

  • Big companies

  • Big government

  • Regulation of/by/for insiders

  • Monopoly/duopoly

  • Locked in slow growing markets

  • Freedom to pick any color phone you want as long as its black.

Representing Smart Network: Tom Nolle, President, CIMI Corp.

There are three reasons why dumb networks are a dumb idea:

  1. There aren't any dumb networks

  2. There aren't enough buyers for dumb networks even if we had them

  3. There isn't enough profit

You cannot have a market that only consists of consumers and not producers. Our mission is not to satisfy consumer demands at the expense of the provider.

The average router today has more intelligence than a 70s mainframe.

Consider Knights Constant: Ck=InIu

If you have a dumb service, you need smart people to consume it. If you have a smart service, it can't be served by dumb people.

90 percent of business sites have no technical support for their networks whatsoever.

Economics really do matter.

Quote: "We thought technology triumphed over dollars, and it doesn't. All the people who believe that still need to get paid."

ROI on dumb networks would be less than 18 percent, a profit margin suitable only for public utilities. In a world of declining margins, the last man standing is the guy with the lowest rate of return.

Quote: "There is no chance whatsoever in my lifetime that we'll re-regulate. If we can't change fundamental policies and regulations, what chance do we have to repeal basic economics? We'd have to re-regulate to retain capital credibility for the carrier industry in the dumb network scenario."

Worldwide, we've left utility status for carriers in favor of competition, which depends on opportunity.

What this is about is bucks not bits.

Quote: "We're talking about making tech open to all, not just to all technical yuppies. To do that, we need to have a producer whom we're willing to pay."

The debate we should have here is: How are we going to preserve the useful properties of the open Internet, the small number of sensible compelling things; how are we going to preserve those things while we create a network that's both consumable and profitable.

Quote: "So consider the truth: We know that populace-free networking doesn't work. Do we vote for smart networks and make those work or do we stick our heads in the Marxist sands of a lawless Internet?"

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights