Clearwire and Sprint Nextel agreed to allow roaming on their underway mobile WiMax networks: It's a huge shot in the arm for mobile WiMax, as neither of the two firms have enough geographic licenses to allow  allow full national coverage. Sprint Nextel planned to back up its WiMax network with its 3G EVDO coverage; Clearwire had no such plan, as it has no other spectrum or network holdings. The deal allows Clearwire to roam onto Sprint's 3G network, which is a boon for them. Both firms can avoid patchy networks now that would make business travelers and companies with national footprints less likely to sign up.

Now, the reason I pose the question as to whether municipal Wi-Fi networks could suffer as a result is that mobile WiMax is a superior technology to Wi-Fi for large-scale mobility and indoor coverage. Yes, this is a Wi-Fi-oriented site. Yes, I have spent much of the last six years writing largely about Wi-Fi. Yes, there are hundred of city-wide Wi-Fi networks in planning stages or being built.

And, true, mobile WiMax in a new, unproven technology with a lot of promise that has to demonstrate its effectiveness in the real world. Truly mobile adapters that fit in or are built into laptops, handhelds, and gadgets like cameras and gaming systems, have yet to come (although we should see laptop cards next year), and are critical for widescale adoption. (Intel has plans to embed Wi-Fi and mobile WiMax in a reference laptop design, of course.)

But there's no question that mobile WiMax is a fresh standard built for long-distance, mobile, outdoor use, and that doesn't use spectrum with tons of competing uses (Wi-Fi) or require support or interoperability with legacy technology (Wi-Fi, cell data). WiMax is the ideal technology for building city-wide wireless networks. It just has to prove itself, and the providers have to offer competitive pricing, too, when compared to those offered for metro-scale Wi-Fi networks, Wi-Fi hotspots, and cell data networks.

I've been saying for some time that Wi-Fi is a "best-worst" technology for metro-scale rollouts (it's the "best-best" for wireless LANs). It's the worst, because it's not designed to work in metro-scale environments, and accommodations for that involve decisions like having mesh nodes that blast the maximum allowable power out in every direction. Wi-Fi is well designed to cope with interference, but not in the complicated RF environments that already exist in every urban area in the developed world. It's a best effort technology, so that you can't guarantee results, because of interference, co-existence, and power limits.

Wi-Fi is the best, however, because it's here today, and well characterized. We know how Wi-Fi works. You can roll it out without licenses, it's flexible, and almost every laptop and an ever-increasing number of mobile devices sold today has a Wi-Fi adapter built in.

Municipalities considering today whether to write a request for proposals to encourage or authorize citywide Wi-Fi networks, especially with what appears to be a new requirement by service providers like MetroFi to sign up for multi-year contracts, may step back to think about WiMax. I've already been thinking that if you were starting today, the near-term availability of 802.11n and MIMO gear for metropolitan deployment--maybe 6 to 12 months away--should make you hesitate in finding a firm that's designing the network today.

Now you're saying, Glenn, Glenn, Glenn, WiMax will be controlled by two firms in this country, extending the oligopoly of the telecoms, wired and wireless, and that Wi-Fi can be deployed by anyone, at any time, whether a city, a non-profit, a neighborhood, or a large-scale service provider--or even a telecom. WiMax is totalitarian; Wi-Fi is socialist.

True. But there's a thin margin that Wi-Fi has over WiMax. Cost and availability of adapters will likely be an issue for as long as two years. Speed is another: Wi-Fi's boost into 802.11n should allow dramatically higher speeds, although large-scale networks need to figure out how to exploit this to their customer's advantage.

The final part of that margin is that Wi-Fi can be offered to cities for a fraction of the cost of cell data, like EVDO and HSDPA. Even with large-scale subscriptions, cities still pay close to $60 per month for cell data subscriptions per card or user, plus the initial cost of the adapter (from $0 to $100, typically). Wi-Fi-equipped laptops in a metro-scale network can hop on an office Wi-Fi network when a city worker is in the office, and then roam onto a secure city-wide network as needed. Because Wi-Fi works across all operating systems, there's no driver limitation as there is with specific support for 3G adapters.

If Clearwire and Sprint Nextel are smart--and Clearwire has already been bidding on city-wide networks--they'll be able to put together a package for municipalities that guts one of the primary financial legs for citywide Wi-Fi. In Seattle, with Clearwire's current proprietary, pre-WiMax gear, you can spend about $35 per month for 1.5 Mbps down and 256 Kbps. They provide a nomadic adapter; one of my colleagues carries the small AC-powered device with him so he can work from home or at his girlfriend's apartments.

With the right offer for municipalities, a city or town could avoid the whole issue about providing access, dealing with utility poles, coping with community groups and their privacy concerns, and waiting for the buildout. Instead, Sprint and Clearwire will just keep lighting up cities and knocking on their doors. (Oh, and Clearwire and Sprint already have voice integrated into their mobile WiMax plans, just by the way.)

EarthLink's reaction here will be critical. The new CEO said a few weeks ago that he'd have a decision in 30 to 60 days about the company's future direction. Some of us who follow the firm expect that the municipal network division could go into maintenance mode, continuing on active networks, and pulling out of undone proposals. This Clearwire/Sprint roaming decision could affect EarthLink's plans, by removing one of the potential underpinnings of their financial model.