People want today's broadband market to be competitive and reasonably priced. They also want tomorrow's networks to be faster than today's. The big challenge for regulators is that these goals are in conflict.
Some countries try to deal with the first problem — lack of competition — by forcing incumbents to share a portion of their networks with competitors at regulated rates. This approach is known as local loop unbundling, and a number of people have argued that the United States should take a similar approach.
Here's how this would work. Suppose that your only option for internet access right now is Time Warner Cable. Under an unbundling approach, TWC would be required to lease out the part of its network closest to you to other companies, at rates set by the government.
That would allow you to choose other companies to provide you with internet service, the same way you could choose Earthlink or AOL for internet service during the days of dial-up internet. Your data would travel a short distance over TWC's network before transferring to your ISP of choice — just as data once flowed over the phone network to reach your dial-up ISP. In theory, because multiple companies would now be competing for your business, you should enjoy better service and lower prices.
But the European experience suggests that this approach can have a big downside: reduced investment in network upgrades. Companies are less inclined to invest in networks that they're going to be forced to share with competitors. But there's also a more subtle problem: the ability to piggy-back on the incumbent's line reduces the incentive for anyone else to invest in building new networks.
The result: Europe has a lot of competition, but it's mostly competition to provide service over the same set of ancient telephone lines. Many countries with strong unbundling rules, including France and Italy, have lagged behind when it comes to upgrading those physical connections.
Three paths to high-speed networks
There are three basic ways to address this problem. One is with taxpayer subsidies. The countries with the fastest internet service, including Sweden,Japan and South Korea, got them by providing public support for broadband networks. If taxpayers help foot the bill, then it's possible to get faster networks and the kind of competition made possible by line-sharing.
In the United States, a few cities have gone a step further and built networks themselves. These cities — including Chattanooga, Tennessee, and Lafayette, Louisiana — now have some of the fastest networks in the country. But these networks didn't come cheap, and most American cities aren't interested in following their lead.
A second approach is to offer incumbents time-limited exemptions from line-sharing rules in exchange for technology investments. This is the approach that has been taken in Germany and the United Kingdom, two large European countries with relatively good internet service.
A third approach — the one adopted in the United States — is to dispense with line-sharing requirements altogether. The US used to have a line-sharing regime, but the Federal Communications Commission largely abandoned it in 2005, a policy the agency hoped would encourage more investment in network upgrades.
This strategy has been a modest success. Cable and telephone incumbents have invested significant sums in faster networks. Verizon has been particularly aggressive, building its FiOS fiber optic network out to nearly 20 million American households. Other major telephone and cable companies have also made significant upgrades.
Even better, a few cities have benefited from the entry of a third competitor. Consumers in Kansas City, for example, can sign up for Google's fiber optic network for internet service in addition to the traditional options of Comcast and AT&T.
Yet the rate of investment has been somewhat disappointing. After almost a decade with no unbundling rules, only about a quarter of households have been upgraded to state-of-the-art fiber optic networks. At this rate, it will take another generation for fiber to reach the whole country.
Do we actually need super-fast networks?
Universal fiber will likely be needed if America is ever to achieve broadband speeds of 1 Gbps per second — 20 to 50 times faster than a typical American internet connection today. Such speeds are already available in a few places, including Kansas City and Tokyo, and many broadband advocates see it as an important benchmark.
But it's not obvious that people actually need these super-fast speeds. Most households in the United States have access to cable television networks — something that isn't true elsewhere in the world. While cable-based networks are not as fast as networks based on fiber optics, they can easily deliver speeds of 100 Mbps, which is fast enough for almost all existing internet applications, including high-definition video streaming. Perhaps gigabit connections are a solution in search of a problem.
But the counterpoint to this is that we might not find out what applications are possible with ubiquitous gigabit network until after we build them. True, some cities have gigabit networks already, but there may not be enough customers in these cities to support a new generation of high-bandwidth internet applications. Hence, if progress in network speeds stalls, we might not even realize what we're missing.