The Hard Questions in Broadband Policy
By Andy Oram
During a period of life most people
try to forget, I learned from my high
school teachers the key to academic
success: how to score well on
standardized tests. "Answer the easy
questions first," they said, "then go
back and answer the hard ones if you
have time."
This is not a bad strategy for policy makers, either. It is the
route taken by Congress, the Federal Communications
Commission, and advocates for Internet service providers in
opening up new possibilities in broadband. They decide such
general questions as "Should all providers have access to
cable networks?" and leave the thorny issues of oversight,
cost, and equitability for later.
But maturity has taught this former high school student some
tough lessons. There is no intellectual training comparable to
20 years of showing technical documents to computer
engineers who rip them to shreds, plus five years of showing
policy papers to law professors who rip them to shreds. I've
found I can't hide from the hard questions.
So in this article I will focus on the hard questions that I see
as remaining to be answered in broadband. And I'll start from
the top, with the questions that are most difficult—because
these are the ones that generate the most points for the
right answers.
1. How do we provide truly universal access
to symmetric broadband?
I'm going to whisk right by cable modems and ADSL. (They
come further down the list of priorities.) Limited in their reach
and puny in upstream bandwidth, they were never meant to
be more than stopgaps. They don't meet the promise held
out by our society's leaders: to bring the entire public
high-speed connections that allow them to get education,
government information, telecommuting, reality TV, and
medical consultations everywhere, all the time.
Wireless is great in some geographic areas, but is hampered
by obstructions and weather in others. It's generally more
subject to crowding than wires, and somewhat less reliable.
Satellite promises universal accessibility but hasn't been
tested with large numbers of subscribers. Right now, it seems
like nothing will really meet our needs but fiber to the home.
It's always been assumed that fiber to the home is
prohibitively expensive to build. But that canard is now
challenged by Miles Fidelman, the founder and president of the
Center for Civic Networking, who advises local governments
on Internet technology and policy. He points out that fiber by
itself is cheaper than copper, because it's essentially made of
sand. Fiber is also lighter, which allows you to put more of it
on a pole without danger of toppling it.
Besides the cost of digging up the ground—which is being
done now anyway when new developments are built—the
reason fiber used to be expensive was the equipment used at
the endpoints. But the cost of this equipment has dropped to
the point where fiber is now completely competitive, coming to
between $1000 and $2000 per subscriber. Worldwide Packets
is one manufacturer already offering such equipment.
So, according to Fidelman, there is no economic reason to use
copper instead of fiber when building new developments. As
for established homes, several experts have suggested fiber
co-ops. In this model, customers band together to bring fiber
to a neighborhood. Each home has to pay for the short line
from the home to the pole. But by pooling resources,
neighbors can afford a line from their poles back to the central
office.
The radical notion of customers owning their own
infrastructure—kind of the ultimate in peer-to-peer
networking—has made headway in Canada, according to
François Menard, a product developer with years of experience
in telecom and now a fiber network project manager at the
consulting engineering firm IMS Experts-Conseils. Where the
E-Rate in the United States hamstrings schools into leasing
conventional service from phone companies, many schools in
Canada are investing the same money into stringing long fiber
cables to form their own private networks. Government
buildings are starting to do the same.
In Quebec, schools are spaced so that one can go from one
to another in no more than 75 kilometers. This makes it easy
to bypass the commercial Internet backbone and routing
system and to route traffic by the crude but effective
mechanism of hopping from one host to the next. This also
means that the users, not the Internet provider, define what
kinds of services are permitted.
"The regulatory regime in Canada is really favorable to building
infrastructure," says Menard. Anyone can become what is
termed a "nondominant carrier" by registering a
three-sentence letter. They can then attach to almost any
facilities.
The Canadian telecom commission created this favorable
situation without really meaning to: A 1995 decision gave
broad rights to budding carriers because the government was
desperate to break the dominant phone company's hold on
the telephone network. They didn't think about the Internet
at that point, but the radical decentralization created by the
bill now provides room for burgeoning Internet competition.
Now Australia is looking to Canada as a model for how to
promote broadband competition. Menard believes the way
forward is to allow practically anyone to register as a carrier,
give them the right to build facilities (such as by expropriating
public rights of way), and keep municipalities from granting
franchises to create monopoly carriers. In the U.S., by
contrast, competing phone carriers have great difficulty
stringing their own fiber, and instead are forced to buy it from
an existing carrier at high cost.
Fidelman and Menard, like many public-interest commentators,
believe the private phone monopolies are too happy with their
position to build the new infrastructure; their obsolete copper
is subsidized in a dozen ways by current regulatory regimes.
Governments or community organizations have to pick up the
slack. Even in the U.S., many local governments are trying to
build municipal fiber networks, or to offer service on existing
networks built by municipal electric companies. Ironically, they
often run up against lawsuits by phone companies.
These companies, having left small towns in the lurch and
declaring they can't afford to offer residents broadband
access, now try to stop the cities by claiming that municipal
networks are unfair competition! The hypocrisy of this
position is highlighted by the practice in most cities of offering
their networks in a nondiscriminatory manner to all ISPs.
Certainly, there are minor issues worth debating (such as
whether tax-free bonds should be used to fund networks that
compete with private ones) but the principal of municipal
broadband is in the best tradition of American self-reliance.
Whether or not local governments build municipal networks,
Fidelman recommends they take other actions to allow the
development of broadband. These include updating building
codes if necessary, and requiring developers to lay the conduit
for fiber when digging up neighborhoods.
Bruce Kushnick, who is the executive director of the New
Networks Institute, a public-interest group doing telecom
research, pointed out in his book The Unauthorized Biography
of the Baby Bells & Info-Scandal that local phone companies
made promises throughout the 1980s and 1990s to build
fiber to the home for millions of Americans. Utility
commissions across the country reduced regulation and
allowed the Bells to collect billions in fees to build out fiber.
Of course, it turned out that fiber to the home was incredibly
expensive at that point, and there were few applications to
make it worthwhile. So the Bells never built the network.
(They kept the money, though.) Kushnick thinks that, if the
fiber had been laid, a wealth of new businesses would have
sprung up to offer services and we wouldn't be experiencing
the Internet downturn we have now.
2. Can broadband ever be affordably priced?
The plague of failures among companies offering ADSL
indicates that something is wrong with current pricing. Some
people hasten to round up the usual suspects—incumbent
telephone companies. Kushnick points to numerous
complaints filed in various states by ISPs documenting that
incumbents offer worse service to competing ISPs than they
offer to the company affiliated with the incumbent. Overpricing
is also alleged, as in a ruling by the Kentucky Public Service
Commission that BellSouth discriminated against Iglou
Internet Services. The complaint was one frequently echoed
around the country: BellSouth charged high rates for
purchasing single lines and reserved reasonable, wholesale
rates for extremely large purchases that would be available
only to a very large service provider. In a market where small
ISPs line up customers a handful at a time, this pricing
excludes competition.
But other observers express a more thoroughgoing
pessimism. It's true, they say, that ADSL from the incumbent
phone companies (and cable modem access from cable
companies) is priced so low that there is no room for
competition. But perhaps, they say, it's not due to
overcharging. Instead, incumbents are cross-subsidizing their
own services.
For an incumbent phone company, phone bills from the mass
of captive phone users could help pay for ADSL. For a cable
company, Internet service is almost always bundled with
television and other services, so determining the actual costs
is impossible for an outsider. Some companies apparently
absorb the cost of Internet service in order to hold on to
customers who might otherwise take their television business
elsewhere. The partnerships between cable companies and
ISPs (Excite@Home and Road Runner) show that the cable
companies are explicitly subsidizing Internet access through
their content offerings.
And even if an ISP managed to get a cheap line to the
customer, it would still have to reserve bandwidth for that
customer on the line it buys to its network access point (one
of the major Internet routers). For instance, an ADSL line
carrying up to 1.4 Megabits to a customer has to be backed
up with the equivalent of a 1.4 Megabit T1 line on the other
end.
And the equipment required to connect to the phone
company or cable company system is extremely expensive.
According to Chris Savage, head of the Telecom/Internet
Practice at Cole, Raywid & Braverman, it will get worse if
phone companies continue doing things like SBC's Project
Pronto.
Today, most neighborhoods still have copper running from
the company's central offices to the home. The distance can
range up to several miles and can contain clunky equipment
that rules out the use of DSL. Project Pronto is stringing fiber
all the way to within a mile or so of each home (often less).
This is an excellent solution for the incumbent phone
company, but now an ISP wanting to offer service comparable
to the incumbent—or a competing phone company serving
ISPs—has a very unpleasant choice. It could reproduce what
the incumbent is doing and string its own fiber to each
neighborhood of 500 to 1000 homes. But since this is not at
all cost-effective; most competing carriers and ISPs will
instead be reduced to reselling the phone company's service,
or simply letting the phone company carry its traffic.
While most neighborhoods still have copper running from the
company's central offices and the home, Project Pronto is
stringing fiber past the central offices to within a mile or so of
each home (often less). Now, an ISP wanting to offer service
comparable to the phone company—or a competing phone
company serving ISPs—would have to string its own fiber to
each neighborhood of 500 to 1000 homes in order to get to
the clean copper needed to offer DSL services. This is not at
all cost-effective; most competing carriers and ISPs will
instead be reduced to reselling the phone company's service,
or simply letting the phone company carry its traffic.
Lawrence Hecht, creator of the Internet Public Policy Network,
which identifies experts that provide consultation to ISPs and
others on Internet policy, still holds out hope. He says, "There
are two popular business models for providing
high-bandwidth content: Get an exclusive relationship with
the content provider, or use the network access points to
cache content." The first option is the well-known strategy
pursued by cable companies, most notably AOL Time Warner.
It's limited to large conglomerates and holds the risk of
discriminating in terms of content. But the second option is
available to small ISPs through the strategy of banding
together and buying access to network access points.
"Vertical integration of the content and the pipe is not
necessary," says Hecht. "What's really necessary is to get
content near the edges of the network where you want it to
be delivered."
For small nonprofit and educational organizations, Hecht looks
for government support. It would be great, he suggests, if
companies involved in streaming media and caching donated
servers to nonprofit and educational use, while lobbying
governments to provide matching funds for these institutions
to develop content. For the companies, it would educate
customers about their services and promote wider use. "If
you're talking about democratizing the media," claims Hecht,
"you can't stick to text; you have to consider streaming audio
and video."
3. How do we promote competition on the
existing local telephone and cable networks?
Five years after the Telecom Act tried to open up competition
in U.S. phone service, it has emerged only for sizeable
businesses—and other countries have even less competition.
While incumbent Bells are supposed to foster competition
before they can offer long-distance service in their areas,
some are getting into long distance on the basis of pretty thin
evidence. Congress is threatening to cut the whole discussion
short and give the Bells everything they want.
And some critics of the incumbents say that long distance is
not a very juicy carrot anyway. According to Savage, the
incumbent's costs for providing a local connection between a
long distance company and an end user run about 0.2 cents
per minute, perhaps even less. But the incumbent charges 2
cents per minute (10 times that amount) to the long-distance
company in federally mandated access charges. So the local
companies are already profiting nicely without having to offer
long-distance service directly.
Cable in the U.S. is a horse of an entirely different color—what
Menard calls a "gray-zone," because the FCC and the courts
are still trying to decide what regulatory regime it falls under.
It was not the FCC, but the Federal Trade Commission that
insisted as part of its consent decree approving the AOL Time
Warner merger that competing ISPs be allowed onto their
cable network.
American ISPs are looking to models from Canada, where the
Canadian Radio-Television and Telecommunications
Commission ruled as far back as 1996 that nonprogramming
services over cable could be regulated as a common carrier.
Chris Taylor of the Canadian Cable Television Association
says, "Third-party access is a reality here in Canada. At this
point in time, it's a limited reality, but details about tariffs
(rates for ISPs to lease service from cable companies) are still
being worked out by the CRTC, and it's likely to grow. Some
cable companies were quite keen on third-party access from
the beginning. But all companies have accepted that it's the
reality and are working to make it successful."
Legal classifications, and even regulated tariffs, do not suffice
to create a level playing field. The devil is in the details. Cable
companies can manipulate the underlying architecture to
discriminate against competing ISPs in many ways:
While routing packets to the ISP (by checking the source
address or by other means), the cable company can
choose a lower quality of service (QoS).
The cable company can simply divert its own customers'
traffic to the Internet before it reaches the Point of
Interconnection where traffic for other ISPs have to go
to be routed to the proper ISPs.
The cable company can put competing ISPs on an
entirely different router, once again with a lower quality
of service.
These things are not a problem in Canada, Taylor claims. "By
law, the cable companies cannot offer a different quality of
service to third-party ISPs." The difficulties we've had in the
U.S. with local phone competition over the past five years
offer the lesson that true competition requires lots of
regulation, and lots of checking up, when one competitor is
using facilities provided by another.
And the longer we wait, the more people will sign up for the
cable company's service (which they are pushing
aggressively), and the more inertia will emerge against
changing ISPs. This is particularly true if customers who want
to switch have to buy a new cable modem to replace one that
only understands how to reach a single ISP.
4. How do we provide adequate resources on
shared cable networks?
A cable network is like an Ethernet LAN: when one person is
sending a packet, everybody else has to wait. This means that
high-bandwidth use on the cable network is a zero-sum
game, and can easily degenerate into a tragedy of the
commons. Furthermore, the small upstream bandwidth
requires companies to ban servers and even peer-to-peer
applications—like the barely alive Napster.
Many ISPs pride themselves on offering guaranteed quality of
service and fancy configurations such as VPNs. On cable
networks, most of these enhancements are prohibited by
cable companies in order to conserve shared bandwidth. Small
ISPs stay alive by differentiating themselves. But on a cable
network, they have limited options to do so. And the options
are at some remove from their primary job of moving traffic,
lying in such ancillary services as Web hosting, backups, and
customer service (assuming the customer problem is not on
the cable network).
Taylor warns, "ISPs have to be aware that as their use of the
shared network grows, and as the number of ISPs sharing it
grows, congestion will occur. And it takes time to segment the
network to relieve the congestion. But the cable companies'
own customers use the same network as all the other ISPs
and are subject to the same capacity constraints. So there's a
huge incentive for the cable company to make sure the
network is as efficient as possible."
Can multiple ISPs peacefully co-exist? Taylor says,
"Requirements are being imposed on cable modems to ensure
that the quality of service to that modem's user, as well as
usage across the whole network, is appropriate." Fidelman
suggests that the quality-of-service features on modern
routers might be used to apportion bandwidth to different
ISPs.
5. Who will pay for content?
Although infrastructure companies are hurting these days, the
really serious wounds have been sustained by content
providers. The value of banner ads is increasingly being
questioned, simply because so many users ignore them. The
science of banner ads is also being questioned. Agencies
rubbed their hands with glee when they realized that
click-throughs could be counted, but that's actually a pretty
crude measure of an ad's reach. Meanwhile, paying by the
click-through has a distorting side effect: it insulates
advertisers from their own bad judgment. They don't have to
worry about paying much for advertising on a site whose
readers aren't interested in them.
Perhaps you don't trust corporate sponsors; you could argue
that the Web would become smaller but better if content were
provided by educational institutions and nonprofits. These,
too, however, find it a strain to keep providing updated,
high-quality content. It takes highly trained staff just to make
sure links are visible in the right places and go to the right
pages, much less format and display new material in a timely
manner. Micropayment schemes are complicated, will take a
long time to put into place, and don't reflect the kind of casual
browsing experience most people look for. So by the time we
hook up all Americans to broadband, they may have nothing
worth looking at.
6. What will happen to wages and working
conditions?
We all talk about progress and innovation in the
communications industry. Most of the time we assume that
goes along with competition and privatization. But while those
often have benefits, they are also sometimes code words for
union-busting. It would not be fair to lower costs by making
workers put in 12-hour days (one of the main issues in last
year's Verizon strike) or slashing their wages in half. To sum
up, people are infrastructure too.
Andy andyo@oreilly.com, is an editor at O'Reilly & Associates
and moderator of the Cyber Rights mailing list for Computer
Professionals for Social Responsibility. This article represents
his views only.