deny the fundamental principle of “open access” that has animated telephone deregulation
merely because lawyers can disagree about how it should be implemented.
This is especially true for the FCC, because the FCC mandates that DSL offer
broadband under what is described as an “open access” model. How it is possible that there is no
concept of “open access” in the context of cable, but a concept of open access in the context of
DSL, frankly baffles us.1 8Certainly if the providers of DSL refused customers the choice of
ISPs, and then cited the Bureau’s findings as a defense to its actions, no court would recognize
the lack of a definition as any excuse.
In our view, “open access” is simply a short hand for a set of competitive
objectives. The objectives sought in the DSL context are perfectly adequate to apply in this
context, at least as a starting point. For the relevant question that the agency should address is
how to assure that customers have an easy choice among relevant competitors, so as to preserve
competition in the broadband market. The DSL requirements assure that.
The Commission can impose open access conditions on the AT&T/MediaOne
merger without replicating the complex regulatory scheme necessary to implement sections 251
and 252 of the 1996 Telecommunications Act. Interconnection to a cable modem network, even
by multiple ISPs, involves nothing more than a standard Internet connection between an ISP and
a router. It does not require collocation of equipment, nor would open access conditions require
AT&T/MediaOne to honor requests for interconnection at special locations within its network.
18 Indeed, AT&T has argued vigorously in favor of imposing open access requirements on local telephone providers.
See Reply Comments of AT&T Corp. (CC Docket No. 98-147), filed October 16, 1998, at 37: “the most important
action the Commission can take to speed deployment of advanced telecommunications services is to vigorously
implement and enforce the market-opening obligations that Section 251 imposes on incumbent LECs.” Why
deployment is encouraged by open access in one context, but closed access in another, is unclear to us.
So long as unaffiliated ISPs are allowed to interconnect at the same place—and at the same
price — as unaffiliated ISPs, the End-to-End principle will not be compromised.
The Bureau report seems to suggest that it is enough if access is open in one
broadband context, and not in all. But in our view, a principle is respected if respected generally,
not occasionally. And the benefits of a principle come from the expectation that it will be
respected. Further, it makes no sense economically to argue that competition in a small subset of
the broadband market is an adequate substitute for competition in the entire broadband market.
This is particularly true if (as the Bureau report itself suggests) the characteristics of the media
Finding (3), (4) and (5): The core of the Bureau’s arguments, however, rest on
findings 3 through 5. These findings, we submit, are internally inconsistent. On the one hand, the
Bureau seems to assume that broadband cable will voluntarily adopt some form of open access.
Finding (4) states that “market forces will compel cable companies to negotiate access
agreements with unaffiliated ISPs.” Thus the future, in the Bureau’s view, will be much like DSL
is (because of regulatory requirements) today. The naïve assumption that AT&T will voluntarily
open the market to competition flies in the face of AT&T’s established policy, compounded by
the consolidation that is occurring in the broadband market. The Bureau does not explain exactly
what “market forces” will compel AT&T to open this market. How exactly will customers of a
certified natural monopoly exercise the power to “vote with their wallets?” The only plausible
disciplining effect the market might have on AT&T’s closed access policy is to slow the rate of
subscription to cable modem service, because the bundled service AT&T provides is less
attractive than an open alternative. But there is no reason to believe that AT&T, lacking effective
competitors in the broadband business in any given city, will recognize or respond to this market
threat. Further, if the Bureau’s hope is that AT&T will be forced into open access because
consumers will delay their switch to broadband in boycott of its closed access policy, it is a
supreme piece of irony to suggest that it is the threat of regulationthat will delay the deployment
If, however, the future is not a future of open access, and if cable becomes
dominant, then finding (5) suggests that “regulation might then be necessary.”19Thus the Bureau
threatens regulation if access is not open, after stating, in finding (3) that “the threat of regulation
ultimately slows deployment of broadband.” These three findings cannot go together. The
Bureau cannot consistently maintain it is not threatening regulation and then threaten regulation.
The Bureau creates more uncertainty than it removes, by conditioning this threat of regulation of
extremely vague notions of how extensive cable broadband must be, and how much open access
Indeed, if the Bureau does in fact decide to regulate this industry because access
does not magically become open, we will end up with morerather than less regulation, because
the bureau will have to regulate not just access to the wires, but a whole host of industries that
could have been competitive but that ended up being bundled to the network itself. We will find
ourselves, in short, in a new era of regulation reminiscent of the old days of the Bell System.
The way to reduce uncertainty, and promote broadband adoption, would be for the
FCC to simply state a clear policy — that cable must be architected to facilitate open access to
cable customers. How quickly, and how precisely, are questions the agency can defer for now.
Just as the FTC has required online merchants to deal with privacy, or face regulation, so too
could the FCC require access providers with significant market power to provide open access, or
face regulation if they don’t. The policy — open access — should be clear, even if cable
companies control how it is implemented in the first instance.
The need for this policy is pressing. The Bureau’s evidence that cable will
negotiate open access contracts is both slight, and beside the point. The Bureau points to
negotiations with America Online, which it suggests is evidence that cable will provide
independent access generally. But the principle of open access, and the value preserved by End-
to-End architecture, is not that the largest and most powerful have the right to access. The
principle of open access, and the design of End-to-End, is that anyonewith a better mousetrap
gets access to the market.
Technological Limitations to Open Access
The Bureau repeats technological arguments about why open access is not
feasible in the context of broadband cable. These arguments are misleading at best.
First, the Bureau repeats cables claim that there is something technologically
impossible about giving ISPs access to the cable lines. Cable, it is claimed, is a “shared
medium,” while DSL is dedicated.
This is a fundamentally misleading argument. The Internet itself is a shared
medium. Its performance, as the report notes, “var[ies] depending on the number of actual
subscribers using the Internet connection at the same time.”2 0The only difference between DSL
and cable is the place where one enters the shared pool. It is true that cable is architected to share
19See Broadband Today, Staff Report To William E. Kennard, Chairman Federal Communications Commission On
Industry Monitoring Sessions Convened By Cable Services Bureau 35 (October 1999).
20Id. at 19.
bandwidth among local users, whereas DSL does not. But whether that difference results in a
difference in performance is simply a function of how many users the cable company decides to
connect, and not upon whether the users it connects have different ISPs. Give a certain profile of
usage, cable broadband can guarantee an effective equivalent of unshared access simply by not
overselling the access they attach at any single node. More to the point, the cable companies can
control usage whether or not they also own the ISPs, merely by limiting the number and size of
network subscriptions. So the shared medium argument does not justify bundling of ISP service
with access to the network.
Second, the Bureau argues that security on a cable node is less effective than on a
DSL connection, since data from other computers passes by all computers on a network node (as
is the case, for example, with an Ethernet network). This argument too is misleading. There is a
difference in the security approaches necessary to implement broadband cable securely, since
users on a particular node are all exposed to the same network traffic. But cable companies are
already developing technologies to eliminate that security risk. There is no reason to believe that
a properly implemented cable system would be any less secure than a comparable DSL system.
And again, there is no reason to believe that cable control over ISPs is necessary to achieve this
Third, the Bureau makes much of AT&T’s expectations that it would be permitted
to run a closed network. The report sites the colorful mixed metaphor of one analyst, that an
open access requirement “puts a shotgun slug through two inches of Excel spreadsheets that
[cable companies] use to generate their rate-of-return calculations.”2 1The argument is apparently
21Id. at 34.
a point about economic justice or fairness — that it would be unjust or unfair to change the rules
If AT&T had these expectations, they were unreasonable. In an age that has
reaped extraordinary benefits from the regulations that deregulated the telephone monopoly, and
that is beginning to reap the benefits from similar regulations deregulating other local
monopolies — for example, electricity — no reasonable business could believe it likely that it
could sustain an old-world regulatory structure that protected monopoly. And if AT&T did build
its models on that assumption, doubtless Excel is quite capable of recalculating the returns on a
different set of assumptions. That, after all, is what an electronic spreadsheet is for.
Incentives to Build Broadband Connections
The Bureau repeats the threat of cable companies, that they won’t invest as
quickly if they are forced to open access. In effect, the argument is that we must grant cable
companies not just a monopoly over the wires, but a right to expand that monopoly into
competitive markets, in order to give them an incentive to implement broadband access. This
argument is simply wrong as a matter of economics. It is possible to grant whatever incentives
are needed by setting the appropriate price for control of the wires themselves. Allowing the
cable companies to gain that incentive by monopolizing a competitive market offers no
guarantee of giving the appropriate incentive, and (as discussed above) poses significant risks to
competition and innovation.
We also suggest that the cable companies protest too much. We have heard many
times the argument that an industry won’t ever develop — or will collapse — if it isn’t given
preferential treatment by the government. Most of those arguments turn out to be illusory. In the
late 1970s, Hollywood argued to Congress that the movie business would not exist in ten years
unless VCRs were banned. We wisely decided not to ban VCRs, and Hollywood is doing better
than ever. More recently, respected legal scholars argued as late as 1995 that no one would ever
put any valuable content on the Internet unless Congress passed special copyright protections for
Internet works. The amazing variety of useful material on the Internet today, despite Congress’
failure to give special perks to copyright owners, belies the argument. It may well be that cable
companies will provide broadband Internet access whether or not we give them special
incentives to do so. And if we are to grant such incentives, they certainly shouldn’t take the form
of the power to destroy a vibrant ISP market.
Further, the speed of investment in broadband is not the only economic and social
value at stake. There is as well the environment for innovation which is affected by the
competitive environment of the Internet. If the cost of a faster deployment of broadband is a
reduction in that competitive environment, then it is not clear the benefit is worth the cost.
Again, the extraordinary returns that AT&T enjoyed before the 1984 consent decree may well
have sped its investment in its conception of what a communications network should be; it
doesn’t follow that there was a net benefit to society from that increased incentive to invest.
As an alternative to its argument that the government should do nothing now, the
Bureau argues that if things turn out for the worse — if cable does in fact implement a closed
system as they say they intend, and if cable becomes an important aspect of the broadband
market — then the government can pursue open access to cable after the fact, through, one
presumes, antitrust litigation.
This is an extraordinary argument. Whether one believes the government is
justified in its suit against Microsoft or not, one cannot avoid the conclusion that the existing
systems for dealing with monopoly problems in the networked economy ex post are extremely
inefficient. Among the costs of using antitrust litigation to design markets are precisely the costs
of uncertainty that the Bureau discusses in relation to cable. To say there is no reason to use a
seatbelt because there is always the care of an emergency room is to miss the extraordinary costs
of any ex post remedy. There is little evidence that the government is in a position to intervene to
undo excess monopoly power in an efficient and expeditious manner.
Moreover, the costs of dislodging an existing monopoly power are always
significant, and always higher ex post. This is particularly true in this context, where if we must
regulate ex post we will face integrated, bundled broadband providers that will have to be broken
up, and ways will have to be found to recreate the competition the FCC will have allowed to
Conclusion: The Appropriate Presumptions in the Internet Context
The Bureau is right about one important fact: We know very little about how this
market functions. Ten years ago, no one would have predicted how the network would matter to
the creation of the Internet; as late as 1995, Microsoft itself confessed it had missed the
significance of the Internet. We are faced in the Internet with a phenomenon we don’t fully
understand, but which has produced an extraordinary economic boom.
In the face of such uncertainty, the question we should ask is what presumptions
should we make about how this market is functioning. In our view, these assumptions should
reflect the design principles of the Internet. The Internet has been the fastest growing network,
crucial to our economy, because it has enabled an extraordinarily innovative competition. It has
enabled this competition in part because of its design. It has been architected, through the End-
to-End design, to enable this competition.
This principle of the initial Internet should guide the government in evaluating
changes to the Internet’s architecture, or acquisitions that threaten to change this effective
architecture. The presumption should be against changes that would interfere with this End-to-
End design. The aim should be to keep the footprint of monopoly power as small as it can be, so
as to minimize the threats to innovation.
These principles should guide the FCC in the context of mergers affecting
ownership of significant aspects of the Internet. If a merger threatens an architecture which is
inconsistent with the Internet’s basic design, and if that merger affects a significant portion of a
relevant Internet market, then the burden should be on the party making that merger to justify
this deviation from the Internet’s default design. The presumption should be against deviating
As with any principle, these presumptions should apply unless there is clear
evidence that displacing them in a particular case would be benign. The burden should not be
upon those who would defend the existing design. The existing design has done quite enough to
defend itself. If there is good reason to allow AT&T to change the cable network into a version
of the old telephone network, then it should bear a heavy burden in justifying this return to past.
In our view, it has not come close to meeting that burden.