Muni-Fi's core failure is total citywide coverage: The requirement for winning bidders in municipal Wi-Fi projects to build out 90 to 97 percent of a city stemmed from a desire to combat previous institutionalized redlining and ghettoization and to overcome the so-called digital divide. City governments were often complicit in or proponents of restricting minority and poor residents to certain parts of town. Redlining was specifically tailored to prevent minorities from living in certain districts. Subsequently, any effort to restrict services has been dubbed redlining; Kozmo.com was the target of criticism in Washington, D.C., for creating service areas that handled only affluent residents, although the logic there is pretty clear.
When muni-Fi was born in Philadelphia, the notion was that you couldn't create a network designed for the poorest residents and allow a service provider to cherry pick areas to serve. Nor could you require that the poorest and least-served areas of town receive service first, because they would undermine the potential for higher revenue. Thus the nearly-full-coverage requirement became de rigeur.
But it's not financially sensible to sign a multi-year contract, sometimes 10 to 15 years long, with a service provider that locks them and the city into such a deal. Even contracts with milestones to delivery, or test areas that have to work first may be problematic.
Does this mean that economic redlining or underserved area redlining is the only way muni-Fi can work? I think it's unclear. It's more likely that the richest residential areas of a city are the least likely to need Wi-Fi as an alternative. Maybe a proposal that charts out the worst 20-percent coverage area, the downtown district, and a handful of popular venues, and includes some indoor planning--indoor nodes in libraries or tied into libraries, recreational centers, and other places--would be a smarter first move. In that scenario, the poorest and most segregated residents would likely receive the most service first, which would better tie into grants and city subsidies to bring service in.
Remember Willie Sutton's apocryphal quote about robbing banks: That's where all the money is. In this case, the underserved areas are where all the money isn't.
It is difficult to achieve an acceptable ROI when ISPs and/or municipalities deploy wide area Wi-Fi networks throughout suburbia as a DSL replacement targeting the fixed residential market (as happens often in the USA muni Wi-Fi market).
Some of the most successful networks from an ROI and sustainability perspective (see metromesh in Perth, Kordia Metro WiFi in NZ as examples) target areas of high population density (CBDs, metro areas etc, hotel/cafe districts) where the networks are used frequently and by a larger variety of nomadic (and increasingly more mobile) end users such as business travelers, tourists, and mobile locals. This clearly leads to a faster ROI. The simple fact is that economic subsidies are required for ISPs if they are forced by cities to deploy in areas that do not meet their minimum ROI criteria. Cities that try to force coverage maps onto ISPs (without an economic incentive package) will end up with unsustainable networks.
Wi-Fi is just one of the available wireless access media. On most networks, Wi-Fi is being used in conjunction with the other access technologies that are available to the end users (2.5G/3G etc).
Usage is still on the up and the ability to seamlessly hand-over between different types of wireless networks (transparent to the end user) will further accelerate usage, as will the increase in penetration of multi-mode devices such as the iPhone.