In late 2004, the idea of city-wide Wi-Fi networks was electric. These metro-scale Wi-Fi networks would cross the chasm of the digital divide by bringing affordable broadband to low-income parts of major cities, and broadband of any kind to marginal neighborhoods, small towns, and largely rural counties. In mid-2008, the juice has drained out; yesterday, the last of the three major independent city-wide Wi-Fi network builders, MetroFi, said they're pulling the plug. EarthLink, another of the large providers, had already given notice of their exit in August 2007, and filed suit this week to remove the equipment for their flagship Philadelphia network. (Kite, a provider mostly in the Southwest, abandoned their Wi-Fi networks starting in early 2008.)

MetroFi predates the muni-Fi movement, having being founded in the early part of the century when broadband penetration via cable and DSL was still modest in many parts of the U.S., prices were high, and current and future speeds were low and expected low. Wi-Fi could compete admirably against these wired networks, it was thought, and against the weak first wave of third-generation (3G) cellular, on price, speed, and availability.

The incumbents don't stand still, and Wi-Fi, designed for interiors, didn't scale well. While it turns out to be possible to build a large-scale seamless Wi-Fi network that delivers from 1 to 4 Mbps of service outdoors to a laptop, and indoors through a $100 to $200 signal booster, it also proved true that a provider needed two to three times the number of Wi-Fi nodes across a city to achieve those speeds than was estimated when networks were largely bid out in 2005 and 2006. If you budgeted for 20 to 25 nodes per square mile and need nearly 50 of these multi-thousand-dollar transceivers, it's hard to imagine how that affects the bottom line. USI Wireless, which biult Minneapolis's network, appears to be the only firm that got the numbers and engineering to add up for them so far.

Meanwhile, AT&T, Sprint Nextel, and Verizon Wireless all moved their 3G networks into the next generation, with speeds they peg at 600 Kbps to 1.4 Mbps for average downstream rates, with higher peak rates. That's what they say, but I just tested a new Sprint USB modem in my office--nestled inside a building--and crossed 2 Mbps for sustained downloads quite easily. True, the 3G services have 5 GB monthly combined upstream and downstream limits (except Sprint), and cost $60 per month with a two-year contract. But they're there, and they seemingly deliver. WiMax and the widely committed Long Term Evolution (LTE) standard for cellular network evolution will bring even higher speeds over the next 2 to 4 years, rivaling today's fast wired broadband.

Wi-Fi was ahead of its time for city-wide deployment: In 2004, it seemed like the best choice because everyone had or could easily get an adapter; no one owned the spectrum; and the cost of broadband was high, unevenly available, and not clear how much better it would get. It's a lot different in 2008, which relegates municipal Wi-Fi to more of a niche market, perhaps, oddly enough, as a good outdoor-only technology for purely municipal purposes.