The Brookings Institution held a forum this morning on "Fostering Internet Competition"—at which, oddly, many panelists seemed resigned to the idea that one layer or another of the Internet would not be competitive: The question, as they saw it, was how to regulate the monopoly player at one layer to foster competition at the next layer up.

For Loyola University law professor Spencer Waller, it is online social platforms like Facebook and Twitter and Google that raise the specter of monopoly, and the question is how to regulate them so as to ensure competition and innovation in services built atop these platforms.  Harvard's Susan Crawford thought the application layer could probably take care of itself, provided the monopolistic physical infrastructure—the means of providing broadband connectivity to end users—was properly regulated. Only one panelist, media theorist Doug Rushkoff, seemed interested in the possibility of fostering competition all the way down—he was, rather astonishingly, the first to utter the phrase "mesh networking" at the very end of the question and answer period. The others—as revealed by frequent analogies to electric grids and interstate highways—seemed stuck in a model that failed to take seriously what is probably the most important fact about Internet policy: Technology moves faster than politics.

The beguilingly broad word "infrastructure" may be partly at fault here. Roads, sewers, railway lines, electric grids, broadcast spectrum, broadband pipes, the TCP/IP protocol, and Facebook are all "infrastructure" in some sense. They're also wildly different in many ways—and loose analogies that conflate them are lethal to sound policy thinking. Whether a particular infrastructure provider constitutes a "monopoly" or even a "natural monopoly," after all, is powerfully determined by technological context. Telephone service is only a "natural monopoly" until someone invents cell phones—and as an important and prescient Cato anthology The Half-Life of Policy Rationales pointed out, technological progress often alters that context so radically that it undermines the justification for policies implemented in response to the problems of the old context.  This obvious point suggests a simple rule of thumb: Even if you have a clear cut problem that seems amenable to a regulatory solution, only act if you're sure the context in which that problem is embedded will change a good deal more slowly than the political process moves, because a regulatory scheme that no longer fits the facts on the ground may well be difficult to dislodge even when it's doing harm. Or, in a nutshell: Make sure there's more inertia in your infrastructure than in your regulatory structure.

Roads and bridges and electric grids are technologies with a lot of inertia. They're big, clunky physical objects that, once built, are apt to remain in use for 50 or 100 years. Even as particular physical pieces of each network are torn down and replaced, the essential nature of the technology remains constant: We drive on wheeled vehicles over concrete; power is delivered to homes over buried or suspended wires. When these infrastructures look like natural monopolies, reasonable people can debate what kind of regulatory structure is appropriate, but a policy well adapted to the facts of the technology is likely to remain relatively well adapted, because the facts change slowly. The roads and power grid in the town where I was born were mostly there before I was born.

This is not what Internet technology looks like. When I was in high school, as ordinary Americans were just starting to get wind that something called "the Internet" existed, almost everyone who connected from home connected over ordinary phone lines to a dial-up service—and many wondered which of these behemoths might ultimately dominate the market: CompuServe, Prodigy, GEnie, Delphi, or America Online? (Remember them?)  By the time I finished college, home users were largely connecting via  cable television pipes—and for many Americans, that remains the lone wired broadband option even now. But as growing numbers of Americans connect primarily through mobile devices, it's hardly their only option or getting on the Internet—and as 4g wireless broadband networks roll out nationally, with a variety of other wireless broadband technologies waiting in the wings—it becomes increasingly possible for the average user to ditch wired broadband entirely, even for applications like high-quality streaming video. (Crawford, rather oddly, referred in passing to wireless broadband as a "natural monopoly"—by which I think she meant there are fewer national carriers than she'd like, but I can't be certain.)

To be sure, wireless broadband is unlikely to match the top speeds of the fastest wired FiOS lines anytime soon—and Baja Fresh isn't a perfect substitute for Chipotle. But perfect substitutability has never been necessary to provide competitive pressure and avoid the harms of monopoly. The wireless alternative just has to be a good enough substitute for enough customers that the wireline provider can't afford to act like they're the only game in town.

Notwithstanding all this, it is no doubt true that there are currently many Americans for whom broadband is a monopoly service available from their local cable operator, with locally available wireless Internet too slow to constitute a realistic replacement. But even if (purely for the sake of argument) you've got the perfect legislative response to the current facts on the ground, the relevant policy question is whether those facts are likely to remain constant over the time it takes to implement that legislative response—and, because regulatory structures have their own inertia, two and three and five years later. It seems obvious they are not.

The assumption of a persistent monopolist in online platforms seems even more obviously confused. Facebook is now supposed to be the invulnerable social networking monopolist. A few years ago it was MySpace, which took the crown from Friendster. In the world of search, we're all beholden to the imperial will of Altavista... wait, sorry, I meant Yahoo!... wait, sorry, I meant Google. Is it still Google? How about now?

There is, to be sure, plenty government could do to foster greater competition at the lowest layer of the Internet stack. It is a little insane that, in a country where the overwhelming majority of households have cable or satellite TV (or have abandoned traditional TV entirely for services like Netflix and Hulu), federal policy keeps some of the most valuable spectrum locked up as a delivery mechanism for reruns of Friends in high def, something the FCC is slooowly moving to change. The agency could also be moving faster to encourage experiments with spectrum sharing and "white spaces." All of these exciting possibilities would have made for a fascinating discussion about how public policy could "foster Internet competition." So it was disappointing that most of the Brookings panelists seemed to assume the indefinite persistence of the status quo, and focused on how to make monopoly bearable. If we'd taken this approach a decade ago, we'd probably be getting the first final rulemaking out of the Subcommittee on Ask Jeeves any day now.