Lev Gonick of Case Western writes about how city-led large-scale Wi-Fi is the wrong approach: Gonick has been involved in expanding OneCommunity since its launch in 2002, and it now spans "thousands of route miles of community owned fiber" with a strategy based on connecting stakeholders who anchor the network. The assessment, availability, and use of fiber assets is one of the missing pieces, he argues, from the municipal plans that were developed and have failed. (Some muni-Fi includes fiber as one component, but rarely as the key starting point.) With OneCleveland, fiber anchors Wi-Fi which flows from appropriately connected institutions--"libraries, schools, universities, community centers, and health care facilities."
This is partly what happened in San Francisco: fiber appeared as an option (at great expense, mind you) after the parameters around a city-wide Wi-Fi network were fully understood. It's unclear whether there's the will in San Francisco to spend the money, but there's little reason to argue against fiber, as it's the current ultimate technology: You can't beat it, but can you afford it? Companies don't locate in cities where fiber is scarce. Cities can't efficiently run their often far-flung sets of offices today without a fiber infrastructure; they fritter a lot of money away on expensive wired hook-ups. I've heard a lot of talk from municipalities about how putting in fiber--or having a fiber provider move into the city--pays back rather quickly for that reason.