ISP perspective: There are physical limitations to how many discrete, independent facilities-based last mile carriers can be built in a city to individual residential premises. In many cities this ends up with being the local telco, and the local cable company running two parallel overlapping networks. For instance, in Seattle, Centurylink (former USWest/Qwest) running a GPON FTTH network, and Comcast or Wave running a hybrid fiber-coaxial DOCSIS3/DOCSIS3.1 cable last mile plant.
This ends up with a defacto duopoly for many consumers. If you are in a less optimal situation you might have something like the terrible choice between 10Mbps ADSL2+ over degraded, long loop length copper POTS wiring, or 100-800Mbps service from Comcast, Charter, RCN or similar. In which case it's not really a choice at all, you effectively have a cable operator monopoly.
Much like how it is impractical and impossible to have more than one last-mile electrical grid operator in a city, it's highly impractical, costly and expensive for more than 2 to 3 maximum facilities-based ISPs to run parallel competing overlapping networks serving end user premises.
If a city is going to have a 3rd carrier run by the municipality, it needs to be an open access network that sells layer 2 transport to a myriad of independent ISPs. In this arrangement, the city and municipality builds a FTTH network, but only sells wholesale capacity to ISPs.
The city is responsible for the OSI layer 1 outside plant cable, running the drop cable to each house and the termination box, and a basic GPON CPE. Individual ISPs can sell layer 3 (IP) services on top of that and provide the billing, customer support and value-added features like triple-play TV/phone etc. In this arrangement the city ISP runs what is effectively a layer 2 network only, and hands customer traffic in aggregate to an ISP at an NNI.
I have to admit that I am much more skeptical of the concept of cities suddenly developing the deep, long term knowledge base and experience to run last mile access IP networks at a scale of hundreds of thousands of customers. I just can't see many cities successfully getting their own ASN, setting up infrastructure at major IX points, acquiring significant chunks of ARIN ipv4 and ipv6 space, and so forth. Not unless they really throw a lot of money at poaching network engineering talent away from ISPs. If you take a look at peeringdb and pick a random residential last mile broadband specialist ISP with 50-100Gbps+ of traffic, I've literally never seen in my life a city government with the on-staff skill set to successfully implement something like that.
If you want to see how this has been done successfully on the municipality-as-wholesaler end, take a look at Grant and Douglas counties in WA state, which operate 1Gbps FTTH last mile networks much as I described in the above paragraphs. In those two cases, the local county government is also the electrical grid last mile operator. So they already have on staff crews of bucket trucks, pole placing equipment, all of the lineman-related stuff required. Not to mention the support infrastructure like HR/admin/accounting is shared between the county last mile fiber network, and the last mile electrical grid service. There's economies of scale in doing something like that as a combined effort.