Broadband: Make It Right Before You Make It Fast

by Andy Oram



Recently there's been a lot of buzz about promoting broadband. Six
years after the passage of the epochal Telecommunications Act, a lot
of people are suddenly agreeing that it has failed. Not that they say
it in so many words. But there's no denying that all the
Telecommunications Act's gifts to major corporations—the
deregulation, the removal of barriers to buy-outs and mergers that
have allowed a handful of companies to dominate U.S. media, the
blurring of boundaries between services—have not achieved their
stated main purpose, which was to spread high-speed service to all
Americans.



This concern has percolated upward to some chief advisors of the Bush
Administration. And thus the front page of the Wall Street Journal on
January 18 reveals that the administration is considering several
regulatory proposals, tax breaks, and even possible subsidies to
promote broadband. Democratic Senator Majority Leader Tom Daschle is
cited as being supportive. Not coincidentally, a January 15 article in
the Washington Post reports that "broadband is a new battle cry in
Washington."



Meanwhile, a lobbying group named TechNet—counting leaders from
Cisco, Intel, and many other Silicon Valley firms—calls on
government to reduce regulation in the communications industries.
Indeed, reducing regulation may promote broadband in some case (as
when communities have gouged a company trying to use their rights of
way to string cables), while hurting it in others (when the regulation
has made incumbent companies provide facilities to competitors).



So business and government are looking to the next step. Well, I have
a suggestion for them: don't try to push broadband on the public in
its current form. People need to trust it before they commit their
livelihood to it.



Pipes That Spring Too Many Leaks




The main problem faced by Internet service right now is not speed, but
reliability.



Anyone who is willing to pay for broadband is going to let a lot ride
on it. Broadband will be a lifeline for their job, their consulting,
or their company that's as important as an air tube to an oxygen pump.
Broadband users will be running their business over the line and
expecting instant access to the Internet for last-minute tactical
shifts.



In case you haven't guessed yet, I don't believe that entertainment
will drive broadband adoption. That's because broadband is expensive.
Wireless may reduce the costs of local access to a point of presence,
but wireless is not suitable for all for geographic and meteorological
conditions, and the costs of the long-distance link still have to be
paid. Gordon Cook, producer of the
COOK Report on Internet,
believes that data traffic will never support a telecom company that
is accustomed to previous revenues from voice traffic.



Reliability is the bane of current broadband. ADSL users, just as much
as dial-up users, complain about being disconnected without a
warning. Cable service is no match for plain old telephone service in
uptime, either.



The ultimate reliability lapse is the carrier that goes belly-up and
loses all the email and other material it stored for its customers.
Just such a scandal hit Excite@Home, the major cable Internet
provider, last year. Note that Excite@Home had contracts with some of
the largest service providers in the world—the major American
cable TV companies. These companies did nothing to rescue it, or its
customers. What users need is some kind of escrow or cache where
their data is buffered against a massive market failure.



Potential users are also frustrated when they find that their area is
not served by a high-bandwidth option—sometimes after contracting
with a cable or phone company for broadband.



Some of these problems may require extra provisioning, and therefore
still higher costs. A change in architecture is also called for. ADSL
and cable modems, both essentially hacks to lines that weren't
designed for those services, just won't cut the mustard in the long
run.



Computer programmers have an old saying that's meant to restrain
unduly clever hackers who try to improve their programs' performance
through coding short-cuts: "make it right before you make it fast."
The same advice could be given to the telecom industry.



The current pursuit of quick profit through minimal upgrades is
failing. And while the string-and-masking-tape ingenuity of the early
Internet years—the years of mom-and-pop ISPs—was a great
start, the big vendors are still doing essentially the same things in
a very different era. A new approach is required, and probably new
companies to carry it through.



What Would Be Reliable?




If ADSL and cable modems are insufficient, we must either return to
tried-and-true technologies like optical fiber or push ahead into
brave new initiatives like packet radio (wireless Internet).



Our age is one of decentralization, and that may well work when it
comes to owning fiber. While optical fiber is what every major
corporation depends on, in Canada it is being democratized. A
consortium called
CANARIE
encourages schools, universities, and other institutions to form
private fiber networks.



In contrast to the E-Rate (universal service fund for schools and
libraries) in the U.S., Canadian institutions are pursuing a strategy
that lets them own fiber and ultimately control their own connections.
They are experimenting with keeping traffic off the Internet, instead
passing it from one participating node to another in a series of hops
resembling peer-to-peer networks like Gnutella.



In Canada, anyone can string their own wire; unlike competitive
organizations in the United States, they're not at the mercy of the
governments or incumbent companies that own the poles and conduits.
(However, the Canadian government has recently removed caps on the
fees that can be charged for using these support structures. This may
increase the costs of stringing new fiber.)



Individual households may be able to join the democratic fiber
movement by forming co-ops. They could string a fiber from a telephone
company's central office to the neighborhood, and pay for short loops
to each house or put up wireless antennae. A combination of fiber and
wireless may finally bring broadband to the bandwidth-starved masses.



But these initiatives have little to offer large telecom corporations,
and those corporations are less likely to hop on the bandwagon than to
toss nails in the road. We must avoid policies that reinforce the
control of the incumbent players.



In the United States, for instance, the E-Rate succeeded in bringing
Internet access—or improving current access—to thousands of schools
and libraries. But it did so through a Faustian bargain: almost all
contracts went to incumbent telephone companies, and the law required
institutions to rent service rather than build their own. They are now
in a situation of permanent dependency.



Subsidies and deregulatory proposals now being batted around in
Congress—often aimed at promoting broadband in rural areas—carry the
same danger. It is the current winners, the incumbent phone companies,
who would reap the benefits.



Make It Smart Too




While I agree with the policy analysts that widespread broadband would
spark a wealth of socially useful applications, I also think we can do
better with the network we have now.



Route control is getting more sophisticated, allowing routers to avoid
lines that are temporarily loaded down. Caching of many sorts has been
used for years to speed up Web access. Extending that concept, a
distributed filesystem (of which
OceanStore
is an intriguing, if heavyweight, example) could reduce bottlenecks on
other kinds of file sharing.



If courts had a clue that these sorts of filesystems are the part of
next stage in Internet infrastructure, they wouldn't be busy closing
them down. But users bear some of the blame too. I can understand why
they'd start using the improved file systems to pass around familiar
content, but they ought to be showing more responsibility. It's time
to take the next step and develop new content unencumbered by
copyright restrictions.



Quality of Service is a contentious subject among network engineers,
but it's already available at some ISPs, who offer different options
for throughput (high transfer rates) or latency (quick response time).
Further technical and financial developments may allow users to get
more out of the current lines without having to buy new ones.



Unsolicited email, along with new generations of spam such as instant
messages and cell phone messages, is a tremendous burden on Internet
connections. Given the ease of spoofing addresses and of sending
email between any two countries, I can't see any policy that
governments or ISPs can take that will stop spam. As the
MAPS
initiative shows, all such projects divide the Internet into a
legitimate segment and an illegitimate segment, which is a dangerous
precedent that could lead to censorship.



However, the burden spam places on the last hop from the ISP to the
user could be eased if they adopted IMAP, a mail protocol that lets
users view subject headers and delete mail before downloading. The
low-bandwidth Internet needs more such smart solutions across the
board.



Current Dangers




With all the hoopla over government support for broadband, it's
natural to fear the hand of another hidden Enron. The proposals
reported in the Wall Street Journal, luckily, sound pretty
even-handed.



The Bush administration is trying to avoid taking a side in what they
call the "food fight" around Dingell-Tauzin, a bill that would upset
the balance between stimulus and free market in the Telecom Act.
Dingell-Tauzin would give the incumbent Bell companies free rein to
offer broadband without opening their networks to competitors.



But there are other ways to weaken the commitment to competition.



Two recent inquiries from the FCC suggest they might match the
Dingell-Tauzin relief for the incumbents on a smaller level. Unable
to overturn the regulations in the Telecom Act—indeed, Chairman
Powell affirms that "the Commission is duty-bound to continue our
implementation and enforcement of these provisions"—the FCC's
strategy apparently is to nickel-and-dime them. They suggest picking
off certain lucrative markets where they would give the incumbents
free rein to expand and to raise prices without requiring further
commitment to competition.



Finally, despite the reluctance to anger large companies like AT&T and
support Dingell-Tauzin, the Bush Administration may aim their gun at
the same target by trying to deregulate future Bell investments in new
high-speed lines.



One bright point stood out in the Wall Street Journal article: the
Administration may free up spectrum for packet wireless. A recent FCC
endorsement of ultra-wideband (UWB) is also promising.



Big telecom and cable companies have lost the public's trust on all
fronts: they offer neither the reliability one expects of established
firms (so far as broadband goes) nor the innovation that younger and
nimbler players would provide. A set of new policies that include
a release of wireless spectrum, a democratization of rules for
interconnecting with the phone network, and encouragement for
government institutions to build their own networks could provide the
impetus for making broadband the norm.




Is reliability more important than speed or throughput?