The Internet was born when that idea died. It was only a matter of time, after the "foreign attachments" rules fell, that all sorts of alien attachments would change how the telephone network was used. After a generation of modems, routers and RJ-11 jacks, we take for granted the benefits of this competition. We treat as commonplace the principle behind its rule.

The idea is this: Of course there are aspects of monopoly in telephone networks. People don't want three telephone poles outside their homes or multiple standards for telephone interconnection. But just because some parts of a network are monopolylike, it doesn't follow that the whole of the network must be monopolylike. Competition in phones, modems and long-distance access can rage, even if only one phone line enters the home. The principle is to isolate monopoly, not to extend it.

Everyone gets this argument when it comes to phones. Now a new battle is brewing about broadband access through cable. Does the same idea apply to cable? Bizarrely, some say no. More bizarrely, AT&T is again leading the anticompetitive charge.

The story is much the same. Cable was originally designed to stream video to televisions. After a bit of recoding, it can now stream video (and less) to computers and back. It does this over a cable modem that takes a cable wire in and feeds an Ethernet jack out and thereby links your computer to the Internet hundreds of times faster than your phone. It's a whole new world of Internet access. No one who touches cable ever goes back to phones.

But this cable system is being designed in an interesting way, and as Kevin Werbach of Release 1.0 puts it, this "architecture matters." Rather than linking you to a place where you might choose your ISP (out of the thousands that compete to deliver cheap, reliable service), the network will pick your ISP for you - either Road Runner or AtHome, which means (if the recent mergers go through) AT&T for short. Choice is being coded away; your ISP will come bundled with your cable modem.

Code is thus limiting competition. The network is being designed to restrict ISP choice, and thereby lock broadband customers to the cable operator's local broadband network. It is still the Internet: Users can still access any IP address in the world. But access is through AT&T's ISP, and local caching will mean content on the local cable network will flow faster. The cable network will become a premium IP network; AT&T will collect the premium.

There is no technical reason to limit choice like this, unless one counts campaign contributions as a technical reason. Cable networks can be designed to be open just as phone networks are. A range of designs could give customers a choice in ISPs. Canada is experimenting with one design, but there are others. Under any of them, ISPs like MindSpring and AOL (dossier) would buy access from AT&T and then compete to provide ISP service and support.

Cable's response to open-access arguments is blackmail: Cable interests says that unless they can control customers as well as access, they won't build broadband. But these threats are just that. Obviously, access fees should be high enough to reward cable for its investment. But no monopoly was needed to build the Internet, and one isn't needed to build broadband access. The idea that an industry would threaten this suggests a deep monopoly problem lurking inside.

The FCC argues that access policies should be national. Chairman William Kennard attacked an Oregon district court's ruling permitting local municipalities to require "open access."

The chairman is absolutely right. Access rules should be national. But he has also suggested that it is too soon to require any particular open access architecture for cable. Maybe. But there's a difference between specifying the code and establishing the principle. The policy should be clear now, because it will be much harder to rebuild the network later.

The FCC is floundering for lack of a mission. It has been embarrassed by Peter Huber-types for its slowness in embracing and pushing competition in communication. Here's a chance for redemption. This is an argument of principle. What is true for toothpaste is true for phones and true for cable: Monopolies should be limited. Competition should be the policy. And code that enables competition should be the rule.

Lawrence Lessig is Berkman Professor of Law at Harvard Law School.