|Posted by tadelste on Feb 13, 2006 2:05 PM|
LXer; By Andy Oram, O'Reilly Media
In recently-aired plans by telephone companies, content providers who are willing to pay extra would get their content delivered at a higher bandwidth. While it's easy to wax indignant over telephone companies' presumptuousness in deciding what packets should travel at what times, it's harder to step back and take in the economic issues driving the proposed change. And there are technical questions about it as well.
The rhetoric is whipping through the air in Washington and racing
at a dizzying pace across the Internet, as highly publicized
on a "network neutrality" bill were held on February 7. Vint Cerf
testified that the issue would determine "the future of the Internet."
In opposing the bill, the incumbent Bell telephone companies suggest that
without their paternal care, the Internet doesn't have a future in
The target of the bill (introduced by Senator Ron Wyden, D-Oregon)
is recently-aired plans by telephone companies, which
I reviewed a few weeks ago,
to be decidedly non-neutral on how they deliver Internet
traffic. Content providers willing to pay extra would get their
content delivered at a higher bandwidth.
While it's easy to wax indignant over telephone companies'
presumptuousness in deciding what packets should travel at what times,
it's harder to step back and take in the economic issues driving the
proposed change. And there are technical questions about it as well.
All of which I'll try to cover in this article.
Pay before you play
The most telling moment in the Senate hearings, philosophically, came
during a presentation by J. Gregory Sidak of Georgetown University. He
said that ideally, each "product" going over an Internet connection
would pay its fair share of the cost of the connection. This principle
places all communication at the mercy of financial considerations.
Taken to its conclusion, the principle would eliminate any
communication that lacks a financial justification.
And that goes, of course, for communications that have no business
model yet, but could have one in the future. With apologies for
restating the familiar, I have to remind readers that the Internet and
the World Wide Web started as noncommercial initiatives.
written testimony (PDF)
pretty baldly indicates that the value of each communication (an
online chat, a file transfer, etc.) should be determined by the
company providing the connection. What legal or economic doctrine, I
ask, justifies that assertion?
Sidak also explains, through a principle known as "Ramsey pricing,"
that companies should use their control over highly desirable
content--such as TV broadcasts--in order to lower prices for
infrastructure. For example, everybody could get to pay a little less
for their telephone line is some pay a premium to get HDTV sporting
events over that line. Even better (one might think) is that everybody
could pay less for the line because Google and Yahoo! pay extra to
deliver ads. This is certainly nice, but the catch is that the
strategy gives the one company--the telco--control over the line, the
broadcast, and (indirectly) the choice of ads.
To understand the conflict over network neutrality, we should consider
the architecture of the Internet. Sometimes the builder determines the
architecture. For instance, when I made a major addition to my house
several years ago, I talked to a building contractor who said, "I
won't work with the design your architect has chosen." We made major
changes to the architecture, on his advice. The lesson is to choose
one's builder carefully. And given today's telephone companies, I will
ask later in this article--are we stuck with this builder?
Fiber without vision
To sell their two-tier Internet proposal, the telephone companies hold
out a juicy prize: an optical fiber network that would transmit
traffic through light waves with a hundred times greater bandwidth
than we now have, even with cable modems or DSL.
As the hearings started, the two sides faced each other off with
dramatic rhetorical differences: the network neutrality side tried to
take the moral high ground--talking of openness, ethics, and
vision--while the telephone company supporters asked pragmatically how
the building of the next generation of networks would be funded. I
thought the hearings became most useful and interesting when the
network neutrality side picked up the theme of the telephone companies
and addressed the economic aspects. The critics of the telephone
companies thereupon pointed out that:
Telephone networks have traditionally been built by billing the users
(not the content providers).
The Bells have repeatedly promised the same thing they're promising
now without actually building the network (see the new book,
The $200 Billion Broadband Scandal).
Content providers already fund the network through their own ISP and
telephone line fees.
I'm cocky enough to add here my own assessment of the chances for an
optical fiber network. The telephone company strategy of charging
content-providers will, I think, hold back fiber upgrades--not promote
The telephone companies will evaluate customers at each central office
and decide whether the demographics of that neighborhood can pay back
their investment through orders of video or other premium services;
roll-outs will be leisurely to say the least. I predict the companies
would also play games with their existing servers and pipes to ration
content. They will milk their existing lines to the hilt before
investing in optical fiber. That's what happens when a company is
driven by incremental business models rather than some kind of
Yeah, sure, the president of the United States Telecom Association,
Walter McCormick, talked in his
of the benefits his proposed network could hold for telemedicine and
telecommuting. But I doubt the pricing structure he wants would lead
to those valuable applications.
In theory, hospitals could pay for telemedicine by reducing costs and
using it to attract more clients. But making them anticipate and
account for bandwidth charges puts more barriers in the way of this
difficult culture change.
And does anybody believe businesses would pay extra to promote
telecommuting among their staff? They don't support telecommuting now,
except perhaps for a handful of star performers.
So the telephone company proposal leads squarely to better bandwidth
for action games and sports channels.
Why do I take such a cynical attitude toward the phone companies? I've
been studying them for years. They are conservative with their
investments. Certainly they've sunk a lot of money into their
infrastructure--few companies have invested more than they have. But
they've done it with the security of monopoly status. They invest only
on the basis of well-understood paybacks. You can't imagine a phone
company acting with the boldness of Microsoft jumping into the
consumer electronics business, or Google offering WiFi networks to
Technical issues: what are the telcos betting on?
Maybe you don't approve of my pre-judging telephone company plans.
Well, then, let's look at the technology. I'm not convinced it can
even deliver what the telephone companies want.
There's nothing new about using smart routing to direct different
types of traffic along different paths and get different quality of
service. ISPs and major Internet users have been doing it for a long
time. But they do it for systems that are directly connected, and
Internet users do it on generally friendly systems.
Plans to charge different rates on an Internet-wide basis are a horse
of a different color. These plans assume that different users can be
recognized by addresses or by some kind of tagging. They also assume
that a single router at the telephone company can reliably shape the
bandwidth to a large number of recipients. Both assumptions, I think,
Both addresses and other tags are easy to spoof. To ensure that they
are favoring traffic from the people who paid them to expedite it,
telephone companies would have to make some major architectural change
to the Internet, essentially creating their own new network. I
wouldn't mind their creating a new network based on any system they
like (Internet protocol designer David Clark has suggested doing so),
but they shouldn't call it Internet traffic.
I also don't believe telephone companies can predict user behavior
well enough to deliver the quality of service they promise. Quality
of service has been tried for years; it works only under strictly
controlled conditions, normally on local area networks.
Gary Bachula of the
Internet2 research project
summarized this finding in his
at the hearings. Routers just don't know what users are going to do,
or when. An unexpected spike in usage could topple the carefully
constructed tower of smart protocols that were supposed to guarantee
Industry issues: what else are the telcos betting on?
Preliminary appeals to the sanctity of the Internet suggest that
content providers will resist telephone company plans. But if the
telcos get to institute the plans, I predict the big Internet sites
will come around.
AOL has been in the entertainment business since its merger with Time
Warner; announcements by Google show that it wants to get in too. And
other large Internet companies are showing interest. I've looked at
their possible business model in an
article on a possible "webcasting" treaty.
This treaty is a mysterious proposal so unintuitive that it looks like
it was designed by a Geneva institute bureaucrat rather than by
anybody in the Internet industry, but it is actually backed by major
The proposed treaty on webcasting sets the legal status of Internet
portals to be like TV or radio broadcasters. It further gives them
rights over fixation (that is, making a copy of Internet content) and
reuse. It seems to be a way of guaranteeing a revenue stream for the
mere service of providing a host computer for streaming content. Note
that if Google became a monopoly, it could do what the phone companies
want to do: charge money to promote some search results over others.
(Other search engines have done it.)
Will a reliance on entertainment strengthen the Internet portals?
Well, news from the record industry's Grammy awards says that viewers
have declined for the over the past few years. This is just one
indication, along with declining movie attendance and CD purchases,
that content is not king (as
Andrew Odlyzko has been saying for years)
and that it will be hard to get rich by skimming off profits from
I can sympathize with the Internet portals' distress. We have entered
a period of intense competition among Microsoft, Google, Yahoo!, AOL,
and other companies with offerings that could quickly become
interchangeable. This is great for the public. We enjoy a wealth of
new services verging on a second dot-com boom. But it's exhausting and
anxiety-provoking for the companies' staff and management. It would
well lead to a pact between content providers and telephone companies
to slow the pace of innovation.
Alternatives to the Bell company proposal
None of us like to feel we have to sit there and take it when anybody
dishes out a bitter feast. So a lot of people raise hopes of bypassing
the Bells and bringing high bandwidth to the public a different way.
Some have gone so far as to declare the incumbent telephone companies
hopelessly outmoded and destined for the dust heap. Other people take
a proportionately moderate view and look forward to alternatives that
would raise the competition a little and lead to a healthier
That last attitude is the one I find most realistic, if not the most
inspiring. None of the proposed alternatives sound strong on their
own, but if we put them all together we may have a future we can live
WiMAX white knight
It's conceivable that an entirely new species of data/voice company
will arise to use
to conquer the last mile. WiMAX, which is a kind of marketing term for
networks based on the 802.16 IEEE standard, is similar to Wi-Fi but
offers a wider geographic range (city-wide coverage) and higher
bandwidth. It achieves these benefits partly by using more spectrum
(usually licensed, and therefore noise-free) and partly by using
protocols that allow more flexible and robust service. Already,
companies have erected cell towers to offer Internet and voice over
early versions of the protocols that have recently been standardized.
WiMAX is a welcome newcomer, but it probably won't provide the big
bandwidth fix for several reasons:
It will never be as high-bandwidth and reliable as fiber. You can't
expect 100 MB or a gigabit of data throughput with wireless, and there
are always security issues and the risk of interference.
The spectrum that's best for WiMAX is owned by cellular companies,
many of them the exact incumbent telephone companies WiMAX was hoped
to bypass. The space will not be a great culture in which young,
muscular companies will sprout, such as those that drove forward the
Internet. Because the spectrum is overpriced and companies spent huge
amounts of money just to license it, a healthy market in that spectrum
will not develop. In fact, companies have been sitting on the best
spectrum for years, doing nothing with it. They probably want to
develop it into a walled garden with even higher fences than those
recommended by the Bells for their fiber offerings.
Even if WiMAX develops into a real Internet option, we are not likely
to see the magic number of competitors--three--that supposedly pushes
forward innovation. The first competitor, incumbent telephone
companies, are conservative and slow-moving. The second competitor,
cable companies, are architecturally limited by old copper and a
Municipalities take the last mile
When we complain about bandwidth, we forget how lucky we are. The
really hard legs of the journey--the cross-continent and
intercontinental stretches--are well provisioned, at least in the
developed world. They got filled with fiber during the dot-com boom.
And there's enough competition left to keep costs reasonable while
ruling out the two-tiered nonsense pushed by the Bells. So we're free
to focus on the last mile.
Cities and towns are well-placed to pick up coverage of this last
mile. They have a history of utility development that extends
naturally to data networks. I've reported on their efforts for years,
as far back as
The best practice seems to be a strict separation between conduit and
service: the municipality provides the physical infrastructure
(perhaps through a private contractor) and opens it to all interested
But there are problems with the municipal solution too:
Governments are poor, and government investment is frowned on in the
current neoliberal climate.
It's quite possible to make bad technological or financial choices,
without a market in place to correct the course of action. This is, of
course, part of the neoliberal critique.
Still, the promise of municipal data networks is powerful enough to
scare telephone companies, which have spent millions to push bills
through state legislatures preventing the cities from exercising their
rights to create such networks. Pennsylvania was the most
well-publicized arena for this fight.
Hold your own
The wildest proposition, and therefore the one that might just work,
is for everyone to buy his or her own fiber. This is what large
institutions have been doing for years. In particularly, Canadian
universities are doing it, aided by laws that make it easy for private
institutions to connect to public networks.
Stringing fiber from a telephone company's central office to a local
building could cost in the tens of thousands of dollars. Sounds like a
lot of money for you in me. But if several households on the block
combine to buy a line, it could come out to a few thousand dollars per
home. Now consider that possessing your own line saves you the cost of
monthly Internet service--presumably you'd just pay a small fee for
interconnection--and gives you access to video and other advanced
services. Sounds cost-effective all of a sudden.
Who I'm mad at
Invoking my right as an American (the politicians and media told me it
was my right in the 1990s) to be an angry white male, I'll list here
all the forces that I feel have let us down in the high-bandwidth
Of course, I'm mad at the Bell telephone companies. The telcos have
tried to control everything they touch for over a century. They're
playing games with a critical social infrastructure now.
I'm mad at Congress, who designed laws around the interests of big
existing companies (admittedly, it's hard to design laws around the
interests of companies that don't exist). And at regulators who
glimpsed the emerging vision of wireless, Voice over IP, and
competitive opportunities but never quite followed through. Congress
and the FCC are struggling now to avert a flood, after spending years
weakening the levees.
So far, nothing new. But I'm also mad at my favorite companies, the
dot-com commerce sites. They took no responsibility for developing the
Internet connections that form their life blood, but just assumed the
lines would be there when needed. I recommended a more active approach
The reason Google made its offer to build WiFi networks is that it
recognized the growing value of local content. It saw cellular
companies building a business model on offering information in
customers' geographical locations, and it wanted to play in the same
space. Competition worked the way it's supposed to, driving bold new
Moving on, I'm mad at Internet2, a research project that has enjoyed a
network of mind-boggling capacity for over a decade and has come up
with nothing to show for it, as far as the public is concerned.
Internet messaging, Napster, BitTorrent, Flickr, search engines--none
of them emerged on Internet2. The big advances in Internet services,
driving forward Internet use among the public, were all thought up by
independent entrepreneurs. Internet2 was meant to stimulate interest
in high bandwidth, but hasn't even made a ding on the public noggin.
And finally, I'm mad at the public for taking the lazy route and
accepting the cheapest form of half-crippled Internet access instead
of a high-capacity bidirectional connection that could make us full
Let's not blame the telcos--or at least not stop with them. No one in
a position to care has cared enough.
Ultimately we may have to depend on the vestigial competition that
comes from the various alternatives I listed early. Different market
structures will come into being in different places. But what I'd
really like is a company with a vision--a willingness to leap into the
unknown future and bring us an unencumbered wealth of bandwidth.