Four or five years ago—memory is failing me here—Phoenix, Arizona, became the first US major metropolitan area to get broadband cable-based Internet access, after small tests in dozens of areas around the country. Phoenix was a logical choice for this service—the city was also among the first to get digital cable, DSL, and similar services—because of its flat terrain and unbroken geography. And at 8:00 A.M. on the first day broadband service became available, I had an appointment with Cox Cable to wire my home to the Internet with cable. Back then, this process required a few hours of work. At one point, I had four people working in my house, with another few working on lines up the street. Needless to say, installation has simplified dramatically since then.

When I moved back to Boston 2 years ago, I was surprised to see how far broadband access had spread. In the tiny town near Boston where I grew up, customers already had three line-based broadband choices—RCN and MediaOne for cable Internet and Bell Atlantic for DSL—as well as satellite-based options. We tested Bell Atlantic (now Verizon) first but found the "pseudo-broadband" of Asymmetric Digital Subscriber Line (ADSL) to be less than satisfactory. Finally, we settled on MediaOne's RoadRunner service—soon to be sold to AT&T Broadband.

In the 5 years since broadband first reached America, the situation has changed dramatically. First, the Internet is slowing with the increase in broadband accounts. Near Boston, for example, my average online speed is roughly half what it was in Phoenix, probably because this area of the country is settled more densely than the desert Southwest. And the broadband players seem to be changing at an alarming rate: AT&T Broadband, for example, likely will be sold this year to Comcast or, at the very least, will spin off into its own company. Most alarming is the hit-or-miss technical proficiency at broadband access companies. The Verizon people I initially spoke with had never heard of DSL, even though we had received mailings advertising the service. And the person who came to my home to set up the service had no understanding of the system or how to test whether the service was working correctly (we had the service installed after we bought the home but before we moved in, so no PC was available to test it).

But I shouldn't complain—at least I can get broadband service. For many people, this service is just a dream. A recent Federal Communications Commission (FCC) press release provides some interesting statistics about broadband growth in the United States. The release is designed to show that broadband is growing dramatically, but dramatic growth is easy to show when you're starting from such a small base. The reality is that most Americans don't even have a broadband option; in those areas that do have such options, the majority of people aren't taking advantage of the service.

The FCC says that broadband connections to homes and businesses increased 63 percent during the second half of 2000. More than 7 million high-speed lines are now installed in this country; 5.2 million of these lines are for residential and small business customers. More than half of these lines offer speeds of 200Kbps or faster in both directions—a speed that the FCC says is "advanced" (we regularly approached 1Mbps in Phoenix; ah, well). Cable broadband, with 3.6 million lines, is still the number one option because of lower costs and higher speeds. DSL has increased to 2 million lines after a rough start with consumers that almost doomed the technology.

But recent signs show that broadband technology isn't the wild success story that the FCC suggests. For example, the most promising broadband wireless-access technology, Ricochet, folded last week—shut down by parent company Metricom, which then filed for bankruptcy. The Ricochet service offered 128Kbps wireless Internet access in major metropolitan areas such as New York, Boston, and Chicago. But Metricom didn't offer the service between those cities, leaving commuters (especially those on trains) and residences in the lurch. The Ricochet service was also expensive, and many users reported sporadic access.

Another problem with broadband is that service is largely confined to residences. Having broadband access at home is great—but then I work from home. And what about access while traveling? I've stayed in many hotels that have broadband Internet access, but I've stayed in many more that don't. Broadband needs to be more widely available, at businesses big and small and in areas travelers frequent (e.g., hotels, airports, trains, planes).

So although the PC industry sells tens of millions of new systems every year, less than 7 percent of these machines connect to the Internet through a high-speed connection. The vast majority of people in the United States—and the world—still access the Internet through a dial-up modem—technology that has been at a virtual standstill for years. And folks, this .NET stuff that we're so excited about is never going to happen on a dial-up modem. Microsoft spends a lot of time working on its online technology to overcome the low quality and often-disconnected nature of dial-up access, but that work is really just retarding the sophistication of the services we could, and should, be getting already. Unfortunately, compared with dial-up service, broadband is expensive to roll out and expensive for end users to purchase. Until this situation changes, we're going to be stuck in a bizarre, semiconnected world.

Surveying the current state of broadband, I find Ricochet's death to be a particularly bad sign because I always thought that this type of wireless "mid-band" Internet access would eventually be the norm. And 128Kbps is perfect for email and simple Web surfing. But broadband technology—and its wireless equivalents—are still just a promise to most users, and probably will be for the foreseeable future. It might take a federal initiative with the scope of the Interstate Highway program of the 1950s—something I don't see happening—to bring the Internet out of the dark ages of dial-up access.