I'm not the first to state this, but the chorus is growing that municipal Wi-Fi will rise or fall by EarthLink's success in Philadelphia: Phila. is likely to be the first major city in the U.S. to have nearly full Wi-Fi coverage in a model that has many different purposes and different revenue streams. (Taipei is the first major city worldwide with Wi-Fi throughout, but it was much less of a partnership with the city, and they haven't done the full rollout of voice services that the firm building the network planned to drive network usage.)
Tropos picked up on my words about Phila. in their regular newsletter, which I found fascinating--not at my words being quoted, but rather that the metro-scale equipment vendor is also pinning its future on whether the Philadelphia experiment really works.
The Philadelphia Inquirer's Miriam Hill, who has written extensively about the network as it's been built, quotes a few reliable sources on Wi-Fi who have concerns about how a Wi-Fi network this size will perform given a lack of knowledge. Philadelphia once looked like it would be the first big city network in the U.S., and teach other cities how to (or if to) build such networks. Now, with networks being built at the same time, it's likely lessons will be learned simultaneously. This could lead to some networks being abandoned mid-stream or scaled down tremendously if cost structures turn out to be different than anticipated.
Tropos once said 20 to 25 nodes per square mile would work in a city; they now put that number at 30 or further north. EarthLink told me last year that 35 nodes was their benchmark; Novarum, the muni-scale testing service quoted in this article, noted in their public research and in an interview that the best networks they found had a node density even higher.
Now in this Inquirer article, EarthLink admits to an average of 42 nodes per square mile. Novarum in its independent testing--hired neither by EarthLink nor Philadelphia--says that node density increased between two separate tests they conducted in that city.
Moving from 20-25 to 42 nodes per square mile means double the expense fairly linearly. And as the Inquirer found out in testing in this separate article, even 42 nodes doesn't guarantee reliable (anecdotal) outdoor coverage.