It’s the Architecture, Mr. Chairman.

Lawrence Lessig

The Internet has a constitution. Its
architecture is this constitution — the way the
net is coded, its design, the principles that
govern its control. Like any constitution, this
architecture embeds certain values. These
values have consequences. In the case of the
Internet, they have produced the greatest
space of innovation that we have seen this
century.

We are about to see the first great violation of
this constitution. FCC Chairman William
Kennard is about to approve it. With the
merger of AT&T and MediaOne, the FCC will
permit a consolidation that will in effect alter a
fundamental feature of the Internet’s design.
And with this alteration, the FCC will allow a
threat to the innovation of the Internet.

The value at stake is a design principle called
“end-to-end.” First described by network
architects Jerome Saltzer, David Reed, and
David Clark, this principle counsels that a
network build complexity at its ends (where
the users are), and keep the middle simple and
clear. As Jerome Saltzer puts it, “don’t force
any service, feature, or restriction on the
customer; his application knows best what
features it needs, and whether or not to
provide those features itself.” It is the user
that gets to choose how the network will be
used. Not the network owner.

End-to-end was first adopted for technical
reasons. But it has had a competitive
consequence. By remaining neutral among the
ways that the internet might be used, the net
has inspired an extraordinary range of
applications that otherwise would not have
existed. For example, had the internet been
optimized for internet telephony, contrary to
end-to-end, as Reed/Saltzer/Clark have
observed, then the protocols for the world wide
web could not have been implemented. A
commitment to neutrality made that
innovation possible.

This commitment is about to change. When the
merger of AT&T and MediaOne is complete,
AT&T and its affiliates will oversee a
broadband cable network that will cover some
80% of the residential broadband market. That
network will be part of the Internet, but it will
not be architected according to the principle of
end-to-end. Instead, the network will give
AT&T and its affiliates the power to
discriminate in the service it will provide, by
limiting users to the ISP of AT&T’s choice. It
will be the owner of the network, not the user,
or the innovator, that gets to decide how the
network gets used.

The threat of this discrimination is great.
Users will not be free to use the Internet for
the full range of internet services; the cable
system will pick which internet services the
user gets. Already users are not allowed more
than 10 minutes of streaming video at a time;
already they are not permitted to set up their
own servers. Thus, already the cable
companies are doing something the network
has not before seen: choosing how the net will

be served, rather than allowing the ends to
choose for themselves.

This change will disrupt the environment for
innovation. It will threaten innovation in this
newest market on the Internet: When the
owners of the network, rather than the users,
get to choose how the network will be used,
innovations that threaten the owners of the
network won’t be allowed.

We’ve seen all this before. When Paul Baran,
an early architect of what we would come to
call a packet-switching design, presented a
RAND Corporation proposal for what was in
essence the internet — in 1964 — AT&T
stopped it. As AT&T’s Jack Osterman said of a
plan ‘First it can’t possibly work, and if it did,
damned if we are going to allow the creation of
a competitor to ourselves.”

Innovators will understand this, as they decide
whether to invest in applications that threaten
the cable monopoly, or applications that
challenge the telephone companies control over
telephony. Why waste time when the owner of
the network can simply disallow your
competitive design?

The FCC, however, seems resolved to do
nothing about this. “No regulation” is better
than defending end-to-end. The market will
take care of itself, the FCC assures us.

But this is an extraordinary misunderstanding
of the FCC’s role, and an extraordinary
forgetting of history. It was the FCC that

imposed rules that opened the telephone
network up; it was an open and general-
purpose telephone network that permitted the
experimentation that created the internet.
AT&T didn’t voluntarily choose to create an
open network; the government made them do
it. And because the government did, the
innovation that is the Internet was made
possible.

No one can say for certain what effect violating
this principle of end-to-end will have. No doubt
there are many contexts where changes to the
architecture are already threatening end-to-
end. But when tinkering with a basic
architectural feature of a network that has
produced the greatest boom in recent history,
we should at least be cautious, and should
understand our roots. The burden should be on
those who would change the constituting
values of this space to prove that their change
will not harm innovation. AT&T has not met
that burden.

It’s the architecture, Mr. Chairman. If it ain’t
broke, don’t break it.