April 12, 2004

Trinh: Beer Via E-mail, And Other IP Challenges

by Guest Contributor

Mike Trinh has a computer science and software engineering background, works as a policy analyst on emerging technology issues, and is a 2L at Georgetown. He blogs at Lawgorithm in his spare time. -Ed.

One of my friends had this great idea for a “broadband beer distribution” network. The theory was that you could use the Information Superhighway to deliver beer to you at your computer via the Internet, saving time and reducing the threat of drunk driving. At the height of the Internet boom, an idea, ridiculous as it was, could have managed some venture capital. At this early point of implementation, it consisted of other people making the run to the fridge for a cold brew, as it was simply a joke by yet another fun-loving guy trying to “harness the power of the internet” to solve the beer run dilemma.

However, the underlying concept of a broadband beer distribution network captures one of the fundamental network design concepts of the Internet Protocol (“IP”): data abstraction. While beer pushes the envelope of the definition, data abstraction, generally, is a design concept that means that networks are designed to handle all kinds of information, treating them all as functionally equivalent.

Using data abstraction, a computer can send any kind of data, knowing that the underlying data network will be able to handle it. Steve Bellovin captures this advanced concept quite succinctly: “anything on [Internet Protocol packets], [Internet Protocol packets] on top of anything.” Quite simply, any data can be abstracted into an IP packet, and an IP packet can be layered onto any networking medium (though beer is not quite “data” as we currently understand it). The current trend towards digital “convergence” (voice over IP, web over cellular network, etc) is simply the logical conclusion of a migration from single-purpose networks (cable TV, telephone networks) to general purpose networks built on concepts of data abstraction. Technical policy must adapt and change into robust policy frameworks that embrace the flexibility and growth capabilities that Internet Protocol networks achieve through data abstraction.

Data abstraction (and layering) is one of the keys to the growth of the Internet, because IP can work with unpredictable future data formats. The World Wide Web, P2P music sharing systems, and massively multiplayer immersive 3D gaming environments are all simply protocols layered on top of the basic Internet Protocol, which through data abstraction, treats all these radically different protocols as simply bits of data. Each of these prototype communications technologies did not have to overcome the massive entrance cost of building a new network—they simply could use the existing Internet Protocol.

Similarly, the Internet Protocol itself can be layered on top of any network (e.g., phone, cable modem, mobile phone, satellite). Before the Internet, the architecture of most data networks was oriented to a single use. Cable television networks carried just cable television, and were the only networks that carried cable television. Traditional phone networks carried only traditional phone traffic. But abstraction and IP networks changed that by introducing general purpose data networks. The result of the data abstraction and protocol layering is that new data networks can carry any type of traffic and were far more flexible than traditional specialized networks. Cable TV networks now carry instant messages and e-mail. Cellular networks now support file transfers. Singe-purpose networks are quickly the going the way of an ISDN line or a bag phone: a cute anachronism of limited relevance in a modern communications environment.

Data abstraction is both the technical beauty of the Internet and a fundamental ugliness in communications regulation. Current regulatory schemes and legislation are still rooted in an assumption of single-purpose data networks, and data abstraction challenges the assumptions inherent in these policies. Data abstraction blurs policy distinctions between types of networks and types of traffic. If existing policy is not to become derelict, regulatory assumptions about single-purpose networks must reflect the dynamic capabilities of modern networks.

Regulatory approaches must adopt to recognize that a particular type of data may travel over any number of data networks, instead of assuming that a particular type of data must travel over a specific network. For example, the Cable Communications Privacy Act (“CCPA”) regulates the dissemination of viewing habits and other TV records by regulating the collection and dissemination of data by cable TV operators only. The CCPA requires that subscribers to cable networks give their written consent before personally identifiable information is collected—but this requirement only extends to cable TV network operators. This model for legislation assumes that only cable TV operators have access to TV subscriber data, which was a fair assumption in 1984, the year the CCPA was passed. This underlying assumption may not be true. As people begin to watch shows over non-cable networks, such as over satellite networks or the Internet, cable TV viewing habits, data intended for protection under the CCPA, may now leave the confines of regulated actors under the CCPA—circumstances that the CCPA does not anticipate.

As networks evolve from single-purpose designs, telecommunications policy must evolve to deal with these general purpose networks carrying commingled data. Policy choices that made sense for single-use networks must be rethought for a general purpose network, where regulation will reach all data traveling over that network. The pending emergence of Voice over IP (VoIP) technology is a prime example of the problems that converging networking places on traditional regulations. Briefly, VoIP allows a data network, such as a cable TV network, to carry telephone services traditionally traveling only on the traditional phone network. VoIP is an anomaly that does not fit well into the existing regulatory schemes: it is not a network solely used for telephone service like a traditional phone network, but a service that is ‘layered’ to run over IP and any number of underlying networks that carry other data. Should Voice over IP be subject to the policy and regulations of traditional phone networks (e.g., universal service, access taxes)? The FCC is currently studying issues of VoIP, and will also address difficult issues of law enforcement access to VoIP under CALEA. CALEA, generally, requires telecommunications service providers to meet government promulgated standards for government surveillance capabilities. VoIP systems blur the CALEA statutory distinction between telecommunications service and information service. Thus, extending prior policies in from the telephone network to VoIP present extremely difficult questions that challenge existing policy frameworks.

The digital convergence that data abstraction has enabled will bring with it numerous challenges to existing data regulation. Congress and the FCC will have to examine existing frameworks, designed for single-purpose networks, for their continuing relevance in a world of general purpose networks and data abstraction. What results will shape the path of the Internet, and the technologies that replace it.

April 12, 2004 12:00 AM | TrackBack
Sitting in Review
Armen (e-mail) #
PG (e-mail) #
Craig Konnoth (e-mail) #
About Us
Senior Status