These are my informal comments on RM-8775, the proposal by America's Carriers Telecommunications Association for the FCC to regulate the "acceptable uses" (i.e., applications) to which the Internet may be used. More specifically, to restrict or prohibit the sale of software that can be used to provide telephony over the Internet, ostensibly to "save" the Internet from congestion.
I am a computer communications engineer with nearly two decades of professional experience in the telecommunications industry. I spent 13 years at Bell Labs and Bellcore. I am presently employed by Qualcomm in San Diego, California, a rapidly growing developer and manufacturer of advanced digital wireless communication systems. My comments here are my own.
This proposal is extremely difficult to take seriously. I first heard about it at a Los Angeles meeting of the Internet Engineering Task Force (IETF), the group most directly responsible for establishing the networking protocols and operational practices that characterize the Internet. Many of the original Internet designers and operators are still active members of the IETF. I personally have been an active IETF member for about 10 years. No other group is as acutely aware of (and challenged by) the Internet's severe growing pains, so one would expect us to be sympathetic to ACTA's professed concern with Internet congestion.
Yet the proposal was universally greeted with derision by every fellow IETF member with whom I spoke. Before it was confirmed, many (including myself) suspected it was an early April Fool's joke.
I'd like to paraphrase two of the more astute (and printable) comments from my IETF colleagues:
"ACTA's claim that they're concerned about Internet congestion caused by voice telephony rings about as true as the former East German government claiming that they built the Berlin Wall out of sincere concern for West Berlin becoming overpopulated." (This was from a senior executive of one of the largest US Internet carriers.)I couldn't have said it better myself. If ACTA's member telephone companies were as sincerely concerned about Internet congestion as they claim to be, they could make it easier to expand capacity by reducing the exorbitant rates they charge for the point-to-point links on which the Internet is largely built. Those providing local telephone service could deploy ISDN and Frame Relay services at more reasonable prices. Or they could contribute their technical employees' time to the many ongoing IETF activities devoted to improving the efficiency and scalability of Internet protocols and operations, as do all serious companies in the Internet marketplace.
"I guess the buggy whip manufacturers are at it again, trying to stop the automobile."
The fact that ACTA apparently does none of these things is overwhelming evidence that their real goal here is to destroy whatever potential the Internet has as competition, however meager. One is reminded of the insanely jealous pre-divestiture Bell System that crushed anyone and anything that dared stand in its way. Given all that has happened since 1984, and coming on the heels of the recent passage of the Telecommunications Deregulation and Competition Act of 1995, ACTA's attitude is amusingly quaint, if not downright antediluvian. Or just plain silly.
ACTA should wake up and smell the coffee. The telecomm world has changed considerably, and for the better. If ACTA's members feel threatened by the Internet, they are welcome to enter the market themselves. Trying to destroying legitimate competition while piously claiming "public interest" won't work anymore.
The barely concealed anti-competitive tone of ACTA's petition should be cause enough for the FCC to dismiss it. However there are also several myths and misstatements of fact in ACTA's petition that I feel must be answered.
Many ISPs do in fact charge for usage beyond some set maximum time included in the basic monthly fee. Rates vary, but a common rate for excess monthly connect time $1-$2/hr. Bear in mind that these rates may well be incurred by the users on both ends of an Internet telephone connection -- unlike conventional telephony, where usually only the caller pays.
Furthermore, a fairly simple calculation will show that the lower per-minute rates of dialup Internet service as compared to conventional long distance telephony are fully consistent with the lower average data rates in use. Most Internet telephones use modern digital speech compression techniques that require only 10-15 kb/s to produce high quality speech. In contrast, standard telephone company practice, at least on domestic calls, is to encode speech in mu-law PCM at 64kb/s, a technology that has remained almost unchanged since the early 1960s. Furthermore, standard circuit-switched telephony is full duplex while most Internet telephones operate in half duplex, reducing average transmission rates even further.
If ACTA's members were to consider competing honestly with Internet telephony instead of trying to destroy it, they might find that they could adopt many of these same cost-saving techniques in their own networks. This might even let them reduce their prices to competitive levels.
If anything, the exact opposite is true. In the Internet's present state, the computer networking applications for which it was primarily designed will crowd out the voice users whenever there is insufficient capacity to satisfy both.
Packet switched computer networks like the Internet were built to meet the very different requirements of computer-to-computer communication, including communications on behalf of a human operator. Computer networking is inherently much less "real-time" than voice telephony. This is not to say that performance isn't important. It is. But as a rule, computer network performance analyses tend to emphasize averages over time (average packet loss, average delay, average response time etc).
Contributing to this phenomenon is the increasing autonomy of the average personal computer; unlike the "dumb terminals" of old that were useless without mainframe computers (and the network to connect them), many personal computers run "agents" on behalf of their users that tend to loosen the delay and availability demands placed on the network. A mail agent like Qualcomm's Eudora is a good example. It permits the user to do almost everything a purely "offline" mode, with occasional batch transfers of mail in the "background".
This stands in stark contrast to real-time voice telephony, where the worst case dominates the network design. Human conversation is inherently a highly intimate and interactive application with stringent human factors requirements. A century of research has established the tolerable limits on delay, distortion, loss, noise and other impairments to human voice conversation and they are fairly well understood.
Computers can be very patient and methodical; if a packet is lost, it can be resent. Human speakers are far less tolerant of such problems, even when momentary. Circuit switched telephone networks were specifically built to meet such requirements. Packet networks like the Internet were not, although there is significant ongoing research in this direction.
As a result, few users of Internet telephones would compare the quality favorably with conventional telephone service even under the best of conditions. And when congestion occurs the voice users are the most likely ones to give up in frustration and go away, leaving the network to the more patient computers and their automatic congestion-control protocols.
Most users of Internet telephones would probably find it downright comical that ACTA finds them so threatening. Any such threat is best dealt with by staying current with technology and reducing costs and prices. If it feels it cannot compete effectively due to over-regulation, the answer is to eliminate those regulations, not to bring them down on the competition.
The Commission should particularly reject any notion of enumerating "acceptable" Internet applications or uses. The Internet has found a remarkable range of novel and useful applications largely because it was built with the fewest possible assumptions about and restrictions on the type of traffic it would carry.
Not only would restricting the uses of the Internet be unwise, it would also be unenforceable. An Internet packet is simply a bundle of arbitrary bits that have real meaning only to the parties that send and receive them. Should the parties to a conversation decide to conceal its nature, e.g., with encryption -- as they probably would if the application in question were "banned" by regulation -- there would simply be no way for the network to reliably distinguish them from "permitted" applications.
Attempts to control the publication of user software that implemented "banned" applications would be no more successful. The serious Constitutional issues aside (one federal judge has already ruled computer software to be Constitutionally protected speech), any serious attempt to enforce such an action would probably make the so-called "war on drugs" look like a rousing overnight success.
The Commission should have the wisdom to recognize that establishing and trying to enforce the destructive and completely unnecessary rules like those proposed by ACTA for the Internet would only waste resources that could be better spent on far more constructive ends.
Phil Karn San Diego, CA