TCS Daily

Being There

By Arnold Kling - September 10, 2002 12:00 AM

Editor's note: This article articulates a vision for the architecture of electronic communications. A follow-up article will discuss the economic challenges in reaching that vision.

It is easier to describe there - a relatively simple vision for the future of telecommunications - than it is to describe here - our comparatively messy present. The vision for there is by no means original with me. My understanding of it owes much to Reed, Frankston, and friends.

When we get to there, the electronic communications environment will consist of two types of components. Not being an engineer, I think of these components as:

  1. The Packet Express; and
  2. Thingies

The Packet Express is the network that delivers Internet packets. When I was acquiring my first clues about the Internet, almost a decade ago, I read Ed Krol's The Whole Internet Catalog. He described the Internet as sort of like the Pony Express. Data gets put into envelopes, called packets. The envelopes have addresses, called headers, giving instructions for where they should be delivered. Think of these envelopes as being given to a rider, who takes them to the next rider, and the next, until finally they are delivered to their destinations, where the recipients open the envelopes.

The Packet Express consists of communication lines (including radio spectrum) and routers. Until recently, there has been considerable uncertainty about the optimal architecture for the Packet Express, especially for the connection between the Internet backbone and individual users (sometimes called "the last mile"). There still are arguments among partisans of DSL, cable modems, and fiber-to-the-home about the relative merits of those technologies for the last mile.

However, it appears to me that Packet Express is going to consist mostly of the fiber-based Internet backbone and wireless relay stations. The wireless solution addresses the fact that people increasingly value mobile communications, and it avoids the high cost of modifying physical infrastructure buried in the nation's streets and front yard.

Thingies are devices that receive digital envelopes from Packet Express and convert the messages into human-usable formats. Thingies will replace telephones (including cell phones), televisions, personal computers, stereos, car radios, and other legacy electronic devices.

We might imagine a scenario in which each legacy device is replaced with an equivalent Thingy. A Thingy whose input-output mechanism consists primarily of a microphone and a speaker would replace a phone. A Thingy consisting primarily of a monitor, speakers, and a remote control would replace a TV. A Thingy with an interface consisting primarily of a monitor and a keyboard would replace a personal computer.

Alternatively, we might see hybrid Thingies. To date, there have been only unsuccessful efforts, such as the Woeful And Pathetic (WAP) interface for using telephones to browse the Web. But these early failures do not mean that the hybrid concept is untenable, or that we are limited to just the form factors that are currently popular.

The simplicity and flexibility of this architecture comes from the fact that Thingies all process the same sort of packets. Packet Express does not need to know whether it is delivering television programs or phone calls in its envelopes. The Thingies that open the envelopes can figure it out.

The ultimate realization of this architecture will mean that Packet Express is ubiquitous, and any Thingy can connect. Wherever you might be, if you have a phone-Thingy, you can make a phone call. If you have a TV-Thingy, you can watch a movie or TV program.

The Economics of the Simple Architecture

No great challenge is posed by the economics of Thingies. People will buy their Thingies, or they may rent Thingies when they are away from home.

Manufacturers of Thingies will need to charge more than the cost of production, in order to recover high expenditures for research and development. However, to obtain a reasonable outcome, the existing patent system interacting with market competition probably will suffice.

Packet Express presents a challenge, because firms must either have duplicate infrastructure or be able to share infrastructure. There are some existing mechanisms for dealing with this. Suppose that Packet Express consists of the Internet backbone plus wireless relay stations. The companies that own different parts of the backbone have "peering arrangements" that allow traffic to flow between them. Similarly, companies that own some wireless relay stations could have "roaming arrangements" that make it possible for consumers to connect from anywhere.

As a consumer, I will pay a monthly subscription fee that allows me to connect all of my Thingies to Packet Express. How will my subscription fee be calculated? By the number of people in my household? By the number of Thingies I own? By the number of packets I send and receive? Economic theory does not say much about which mechanism is best.

The subscription fees should satisfy two criteria. From an average cost perspective, the fees in the aggregate should cover the cost of operating and maintaining the network. From a marginal cost perspective, the fees should provide some incentive for the individual not to put excessive pressure on the network to expand capacity. However, usage fees should not provide an arbitrary disincentive to use the network.

Charging by the packet probably is overly punitive at the margin. My use of the network presses on capacity only if I try to send packets at times when the network is congested. Some form of "peak-load surcharge" can address this issue. For example, with my local electric company, I am signed up for a program that charges me lower rates all year round, but allows the company to selectively shut off my air conditioning for 15 minutes at a time when demand is high.

Similarly, Packet Express access providers might charge a lower monthly fee to people who agree that when congestion is high the packets that they send and receive can be delayed a few seconds more than normal.

My guess is that pricing for Packet Express will wind up looking something like cell phone subscription plans. For $10 a month, you will get up to a certain volume of packets. For $20 a month, you will get up to a higher volume. A plan may include one volume of "peak-time" packets and a larger volume of "off-peak" packets.



TCS Daily Archives