A Practical Navigator for the Internet Economy



Some Notes from your Editor: In preparing this issue I have been doing much thinking about how to view Internet technology from both a business and regulatory model. Four articles have forced this focus. Three, one on Equinix Internet Business Centers, one on venture capitalists and one on ICANN are published herein. The fourth on the issue of how Internet Access to cable networks in Canada could become the focus of a new regulatory paradigm is by Francois Menard. It couldn't be finished in time. I anticipate being able to publish this as a part one of the January 2000 COOK Report within a few days after I return from Nepal on November 6th.

This article grew out of conversations with Francois Menard who has been looking very closely at the dispute between Videotron and Canadian ISPs over Open Access to Videotron's cable network in Canada. From the framework outlined in "Netheads versus Bellheads" (http://www.tmdenton.com/netheads3.htm), Menard has applied his understanding of the technology to make some persuasive arguments about the choices facing Canadian regulators. As I read his drafts I pressed him to elaborate on and generalize the assumptions that he had applied. The result has become a rough draft of a document that I believe carries forward the ideas expressed in Netheads versus Bellheads. The conclusions that are so well founded that they may become extremely compelling basis from which to rethink current regulatory approaches to telecommunications. I want to use this preface to summarize my own understanding of the issues involved and to alert my readers to some things to think about between now and my publication of his completed piece in mid November.

The Impact of the Internet on Technology Development and Regulation

We are looking squarely at a situation where the pace of technology development has outstripped the ability of politicians and regulators to deal with it. We need to remind ourselves that ‹ from the introduction of the telegraph 150 years ago through radio, telephony, and television ‹ telecommunications technologies have grown and prospered as vertically integrated businesses with heavy emphasis on infrastructures which to be need regulated as natural monopolies in order for them to grow and prosper and to serve thereby the public's needs. Such technologies often included proprietary twists designed to give one large corporation's vertical monopoly a competitive edge over another's infrastructure. Significant economic inefficiency was produced by each company having to build its own distribution infrastructure. This then led to a situation where companies could turn to regulators and plead for protection. They would argue that they could only afford to undertake the expense of an upgrade to their infrastructure if they were promised that they would not be required to share it with competitors.

During the past 25 years Moore's law and the TCP/IP protocol have built a very different foundation for telecommunications. Astonishing advances in integrated circuitry have created a situation where equipment needed to add requisite intelligence to a telecommunications network was no longer so terribly expensive that it could only be afforded by a vertically integrated monopoly. For example a PC having the power of a minicomputer of a decade ago now costs as much as a television set. With the release of the Apple G4 the computational power of a 1990 supercomputer is yours for $2500.

The second part of the revolutionary wave facing us is the impact of the TCP/IP protocol. IP gave us generic "envelopes" into which binary data could be dumped and sent via basic transport protocols across a network for processing at ever more intelligent endpoints on the desktops of users. This is the essence of the "stupid network" as argued by David Isenberg in 1997. Just deliver the bits. For the first time a horizontally oriented telecommunications infrastructure could be built via inexpensive technologies with companies invited to plug into each other tinker toy fashion.

Those operating under this world view merely offer others TCP/IP bandwidth into which they may plug their networks. Bandwidth providers can interconnect and, just so long as they use the same public domain interoperable protocols, they can focus on an effort of interconnecting separate and individually owned infrastructure. Because this infrastructure does not have to be vertically integrated, it can be plugged together in chunks like building blocks of Lego or the Tinker Toys of an earlier generation.

This infrastructure does not need a single central authority point since the reliability of the network is governed by the TCP/IP stack residing inside each user's machine talking to the TCP/IP protocol stack resident within the machine elsewhere within the network with which the user communicates. The value of the resulting inter network was enhanced not by any one company's vertical market share but rather by the size of the network measured by numbers of computers connected. The costs of interconnection and transportation continued to plummet as expensive telephony oriented transport equipment such as SONET and ATM could be replaced by generic and inexpensive Ethernet. Ethernet is now reaching speeds of ten gigabits per second and threatening to become a universal transport protocol from desk top to desk top across the wide area network of networks known as the Internet.

Suddenly a structure based on new technology and only partially overlaid on the public switched telephone network had developed. This technology operated on a fundamentally different business model and philosophy from that of the public telephony network. But by late 1999 every one who looked at the new century ahead realized that, in less than a decade, TCP/IP had enabled Internet technology and, had created an infrastructure which unless it is choked off by regulation, will soon overtake the PSTN in traffic and equal it in size.

Growing a Stupid Network to Match the Intelligent in Size

The Internet operated from a suite of shared open protocols designed to make it easy for any interested company to inter network and any applicable technology to inter operate. This new telecommunications model was designed to break down vertical barriers. By using a common protocol that could encapsulate vastly different communications media, the Internet enabled a seamless interweaving of data, voice and video on its physical transmission media. A continuous flow of statistical multiplexed connectionless traffic filled the wireline infrastructure with an efficiency not attainable by bursty, connection-oriented traffic that was either voice, video or data. The cost of delivering the bits to intelligent user endpoints at the edge of the network was headed downward in a way that gave the vertically integrated dinosaurs much heartburn. Worse yet, from the big corporate point of view, was the fact that the economies scale behind this new telecom model encouraged horizontal cooperation between companies that could serve as specialists for outsourcing services desired by end users. For example, EarthLink could employ one company to maintain its customers' email services and another to do its web hosting. The efficiencies of horizontal cooperation also made it possible for end users to compete successfully with vertically integrated empires in delivering their own content.

The vertically integrated telcos were supposed to be natural monopolies because of the vast size of their wireline networks, built by billions of dollars investment over several decades. It was assumed that such fully connected infrastructures could only be built over the course of decades. And yet with the flowering of the internet paradigm at least four new national fiber networks (Qwest, Level 3, Williams and IXC) have blossomed in the past five ears. A concentration on horizontal inter operable inter networked services has enabled a previously unimaginable increase in the total size of the telecommunications pie. Qwest has grown from revenues of nothing to 2.2 billion in 5 years and is now acquiring US West with revenues of 12 billion per year. Internet growth in bandwidth consumption continues to double every six months and is moving into entirely new areas of human commerce and information sharing .

This growth of the unregulated, horizontally inter-linked Internet has enabled an entire new blossoming of telecommunications technology where venture capitalists fund entrepreneurs to develop new technology which cannibalizes other new technology at maximum speed. Given the business model and operational organization of the Internet and Internet savvy companies, the continual innovation of Internet is desirable in a way that is incomprehensible to those in the older more conservative vertically integrated telco structures. The interview with Flat Iron Partners in this issue is designed to give readers a taste of how venture capital inter operates to fuel the growing Internet. Given an adequate understanding of the operational paradigm described herein VCs will easily come to understand that despite the pleas of Vint Cerf and John Patrick, giving money to ICANN is contrary to their own interests.

The Equinix interview that is this issues' lead article will help readers to understand how the success of the internet's horizontal value proposition created an environment where the first vertically integrated ISPs found that they could prosper only by outsourcing functionality to specialists. Such functionality runs the gamut from web hosting to email hosting, to modem pool provision, to Internet telephony settlement services. Equinix is bringing service providers of all kinds together into neutral business centers where customers are able to locate and sell their services to ISPs without supervision or regulatory interference of any kind. The purpose of the Equinix Exchange is to facilitate all levels of interconnection and to make changing one's internet connections as easy and inexpensive as possible. Equinix Exchanges serve as collection points for the same unbundled layer 3 services that Menard is positing as the foundation of a new regulatory paradigm.

The Equinix business model could not operate within a vertically oriented regulatory market. Furthermore, regulators attuned to such a tradition have a vertical frame of reference. To them everything in the horizontal plane that intersects their vertical world view looks like a point. They simply cannot see the horizontal possibilities. To them the intersections in the plane appear like a bunch of unrelated dots. On the other hand to those operating from the horizontal Internet perspective voice and data packets look absolutely the same and should be treated the same.

However, if one looks at the horizontal plane on which the ISP lives, each intersecting intelligent network service also is seen as an unrelated unconnected dot. Each service is merely something that can add value at the third or IP layer of the ISP's network more cost effectively with its new stupid network technology than it can add value as an overlay of a telco "intelligent network." From the ISP point of view, entire vertical telco infrastructures may now be contained in just a single Dense Wave Division Multiplexing lambda (color) set up by the fiber provider with a few clicks on a web site. It is also a bit ironic that, in terms of this paradigm, national fiber based infrastructure providers are thriving by providing dark fiber services, while ILECs feel that their local copper plants are their worst liability.

A Single Point of Failure

Unfortunately in the process of scaling the distributed technology of the Internet, the Domain Name System developed by Paul Mockepetris in 1986 was designed in such a way that it was subject to central control. With the rise of the web people accepted as fact that DNS was the means for a user friendly form of address and did not debate whether the existence of a single root for the hierarchical DNS to point to was risky in that it was a single point of control in an otherwise utterly decentralized system.

Those who saw the enormous economy of such a system also saw it as a threat to their vertically integrated empires. To defend themselves, they fashioned a strategy of using the ostensibly neutral Internet Society as a magnet around which to organize participants from the vertically integrated telecom industry and content -trademark industry in an effort to organize ICANN. Those tied most strongly to the old vertically integrated view of technology (IBM, MCI, ATT) came together in ICANN. In doing this they turned ICANN from an allegedly neutral coordination body for internet plumbing into a body which in October 1999 is poised on the edge of monopoly control of the internet's DNS. The existence of a single root system for resolving internet addresses made it possible for those with the most to lose from the success of the internet's continued growth to fashion ICANN as a regulatory system which would impose a vertically integrated monopoly of control on a system that was beginning to threaten an installed base worth hundreds of billions if not trillions of dollars. ICANN will use network technology to fashion itself as an archaic central authority. This is the lesson of the ICANN article to be found on page 13 of this issue.

We must understand first that economic gains in telecommunications have always come from new entry into the marketplace and not from improved regulation. As the is pointed out below, in the Internet we have a set of technologies that, if they are not disabled by the by the politicians and regulators, can enable affordable broadband to the home immediately.

One must recognize that two different platforms based on different business and financial models are set to compete for telecommunications at the local loop. One business model would appear to be a choice of service from a single vertically integrated giant telco on the one hand for cable, telephony, wireless (PCS), internet, content on demand all brought to us on a DSL enhanced copper or coax based local loop.

A competing business model may be found in Canada. There as he deploys the IP over fiber CANet III, Bill St. Arnaud has concluded that, if aggregated horizontal services were brought in to homes in urban areas netted as much as $100 per month per house hold, a fiber based loop company working as a neutral carrier for various services could being fiber to the home. No one is asking for unbundled layer 3 services (horizontal plug-ins like IP Telephony, web hosting and email) to be given away. Regulation should encourage the owners of the copper plant to stick to a schedule for replacing the copper with fiber to which any and all service providers should have equal access. In the US perhaps the universal service fund could be used as a means of weaning away the operators of the intelligent network from their bell headed tools. In other words owner of the fiber local loop should be treated as a common carrier.

Menard writes in a draft of the paper he is preparing: "The layered model of the Internet, which has been previously explained by the authors in Bellheads vs Netheads (www.tmtendon.com), makes it possible for common carrier obligations to be enforced at a higher layer than the physical one. Doing so can finally ensure a viable regulatory framework that allows facilities-less INTERNET competition with horizontally specialized services to compete with facilities-based competition with a monopoly on vertically integrated services."

How to sum up? Regulators, given their vertical mindset, seem to have let the Internet progress as far as it has almost by inadvertence. Given the state of technology, Menard believes that now they will have to do one of two things. Either tell Internet users they made a big mistake and that since they must will maintain their vertical mindset, then users had better prepare for a much more centralized marketplace with several orders of magnitude fewer providers or that regulators must embrace the concept of level 3 unbundling. This will force facilities based players like Videotron to let ISPs pay appropriately for the privilege of using their infrastructure. The ISPs will be free to compete with Videotron by using their knowledge of the capabilities of the technology of the internet in more creative ways than has Videotron.

As Menard is finding (we intend to present his findings in our next issue) the onward rush of horizontally based internet technologies is running finally headlong into the entrenched interests of the vertically integrated telcos which are propped up by many decades of entrenched regulatory mindset and political payoffs. The battle for the future of the Internet may well lie in the industry's ability to understand its assets and educate both regulators and politicians into an understanding of the consequences of moving forward without a clue of the nature of the fundamental incompatibilities of these different world views.

CONTENTS of December 1999 COOK Report

Equinix Creating Neutral Exchange Points -- Adelson Traces Trend Toward Horizontal Outsourcing -- Equinix Business Model Offers Incubator for Coordination and Interconnection Between Providers of ISP Services pp. 1- 8, 24


Venture Capital and the Internet -- Jerry Colonna of Flat Iron Partners Describes the Process -- Predicts a Stabile Growth Path & Spread of VCs to Europe pp. 9 - 12


ICANN - NSI Domain Name Accord Creates Joint Monopoly: Registrants Deprived of Most Rights -- Symbiotic Relationship Expected to Free NSI of Liability for Acting as Agent of ICANN Which Can't Be Sued pp. 13 - 21, 24


Executive Summary -- Regulator's Dilemma: Cable versus Internet -- Some Vertical versus Horizontal Organizing Paradigms for Equinix, VCs and ICANN pp. 21-24