The technology of the Internet is not static. Although its 'end-to-end' arc
hitecture has made this 'connection-less' communications system readily 'ex
tensible', and highly encouraging to innovation both in hardware and softwa
re applications, there are strong pressures for engineering changes. Some o
f these are wanted to support novel transport services (e.g. voice telephon
y, real-time video); others would address drawbacks that appeared with the
opening of the Internet to public and commercial traffic-e.g. the difficult
ies of blocking delivery of offensive content, suppressing malicious action
s (e.g. 'denial of service' attacks), pricing bandwidth usage to reduce con
gestion. The expected gains from making improvements in the core of the net
work should be weighed against the loss of the social and economic benefits
that derive from the end-to-end architectural design. Even where technolog
ical 'fixes' can be placed at the networks' edges, the option remains to se
arch for alternative, institutional mechanisms of governing conduct in cybe
rspace.