David Isenberg just gave an amazing, stirring address on the Stupid Network at Supernova. My notes:
Sure you can do Internet on the phone network — you can do Internet on smoke signals, too. It's yesterday's news. The best network is a stupid network, which supplies simple connections, but no "services." Instead, "services" are created by smart, network-enabled products, designed for any networked application. Bring them home and plug them in.
[He holds up a slim cable containing 864 fibers that can be run down your street or under it] Two of these fibers could handle the peak load of the entire United States. You can light this up at a gigabit, just for your home — that's the capacity of a telephone office of a city of 100,000 people. In two or three years, you can have an entire telephone company's worth of bandwidth in your house for $2,000.
The phone companies value artificial scarcity. The most malleable of all laws (Moore's Law, Gilder's Law) is accounting law — depreciation (as we saw with Enron). Bean counters assume the net will be replaced in five years — but with the rate of growth in Gilder's Law, it's like replacing the paperboy's bicycle with a rocket-ship. The paper-boy can't deliver papers on a rocket-ship. [me: yay! obsolete paper-boys!]
Engineering effort doesn't scale like Moore or Gilder — one engineer can only do one engineer's worth of work. If we increase the amount of engineering required for our rocket-ship net, we'll run out of engineers. So keep it simple, stupid. All the smarts in the network should be at the ends, in PCs or devices, not in routers or other network pieces.
Internetworking shifts control and value-creation from the network owner to the end-user. A conventional telephone call touches every node in every network, and every node's owner can add features — call waiting, etc. The Internet's job is to ignore network-specific differences, like call waiting. Call-waiting is defined at the end-points between both parties on the conversation.
Networks that add cool features break the stupidity principle.
The Internet makes telephony into just another application. Traditionally, you need telephone wires, poles, network and service. You pay for the service, though, not all the hardware. The telephone company does business this way, it's the only way they know.
In a stupid network, telephony is just an application. The telcos know how to string wires and put up poles, but not how to make money on 'em. That's why all the winning apps weren't built by telcos: email, ecommerce, the Web, blogging, etc.
Most of the important future communications applications haven't been discovered yet. This is the green-screen, command-line era of telephony.
Inn the telco world, they charge money for providing this voice application and spend the money to support the network and physical plant.
In the stupid netowrk, the physical layer is designed for anything digital. The network layer is Internet protocol. The applications are anything: data, video, voice, whatever.
MSFT may have a monopoly, but it doesn't have the poles-and-wires monopolistic advantage that the telcos have. The potential for a marketplace in stupidnet applications exists.
So in the stupidnet world, who pays for the physical layer: poles, wires and so on? The wires are usually an expense subsidized by the voice service. When voice is free, who will keep putting poles up?
The telco won't make the transition. They're too addicted to their business. The cable-companies may have a better shot, but they're addicted to video entertainment business. They don't want to put in a net that will let anyone get any video signal they want from anywhere. Municipalities: there are 125 cities in the US that are actively investigating their own fiber nets. Utilities have wire and pipes in our homes. New kinds of companies may do it. Customers and corporations own their own networks.
Stupidnet has its own values: First Amendment, decentralization, not any-color-you-like-so-long-as-its-black.
Remember: Goliath lost! It takes smart people to build the stupidnet!
