Right; the premise of this conversation is the edge cases where this isn’t obvious. For the vast, vast majority of places this entire conversation is moot because the addr:city such as we know it in OSM isn’t in question, and can just be easily gained through a spatial query.
This isn’t a tangent, I think you’re driving at the very heart of the matter, and this is a great example of where you really do need to distinguish with that addr:city tag. Given there is a distinct numbered street grid in San Pedro, even though it doesn’t have its own municipal government, I would presume anybody in the LA area would distinguish San Pedro from the rest of LA because otherwise you wouldn’t necessarily know if you meant an address downtown instead.
All a “9-1-1 address” is is the address you would give a 9-1-1 operator in an emergency. I would presume if you called 9-1-1 for an ambulance and they asked where you were, if you’re in San Pedro you’re very pointedly going to say so, and they’re going to know precisely what you mean.
Certainly, if it were true that the city name used in an address was always the same as the containing municipality name (setting aside those in unincorporated areas) then an addr:city tag containing that name would be redundant. However, as this discussion shows, this is not the case. Populating addr:city in the many cases where the address city name differs from the municipality name is clearly beneficial, but even when they do happen to match I’d argue there is still a benefit.
The municipality of Fayston, VT is covered by two different zip codes - 05673, and 05660. For each of these zip codes, the USPS calls out Fayston and North Fayston as city names to avoid. For the 05673 area of Fayston, the city name Waitsfield is used instead and for the 05660 area, Moretown is used. A data consumer doing a spatial query for the containing municipality name here would end up with the non-preferred city name of Fayston. The 05673 zip code also fully covers the neighboring municipality of Waitsfield. There a spatial query would give the preferred address city name since it happens to match the municipality name. Querying for the containing municipality name is a good fallback when addr:city is not tagged, but in this case there is no way for a data consumer to know whether it is the preferred city name for addressing or not. Despite being redundant, if addr:city=Waitsfield is explicitly tagged within the municipality of Waitsfield this indicates that indeed the preferred city name for addressing matches the municipality name, unlike neighboring Fayston where it does not.
The reason I started this thread is that (in my state at least) the state E911 database and US National Address Database (NAD) do not contain the postal city name, only the containing municipality name. Addresses from these data sets are being added via MapWithAI/Rapid and are populating non-preferred city names in these edge case rural areas. While these city names clearly aren’t exactly “wrong”, they will likely not be what locals expect or search for when attempting to find an address.
Sometimes those postal cities are actually neighborhood names. That is because some residents will only use their neighborhood as their address because they don’t realize or want to admit where they actually live.
It’s probably impossible to answer to this question definitively without asking everyone who lives or works in the area what their home or work address is. However, it is pretty clear that there are places in the US where the city name commonly used in addresses is different than the containing municipality name. Using the latter in an address would be understood, but it would seem weird to people familiar with the quirks of their local addressing system.
From context of this thread, I think we are dealing with the USPS preferred name for the ZIP code. You can look it up for any given ZIP code at ZIP Code™ Lookup | USPS
And this is precisely my point, my friend: we’re talking about the preferences of the United States Postal Service.
My overarching opinion is that the ZIP code database is but one source among (potentially) many. At the end of the day the ZIP code database is not telling you where something is: it’s only telling you how to get a piece of mail delivered.
Absolutely it’s not a simple straightforward thing, and that’s why you started this topic and solicited the input of the greater community, right?
What I’m really trying to say, at the end of the day, is that addr:cityshould be “the most commonly-used, to people familiar with the weird quirks of their local addressing system”. I’m also trying to say “don’t presume that the USPS knows better”.
To be clear: The USPS considers their 5 Digit ZIP and City State databases to be proprietary and protected by copyright. While the actual copyrightability of this data in the US is…unlikely, at best, they still can’t be used in OSM.
Welcome to my world (rural north Florida, specifically Gilchrist county)
Our county has two (maybe three) incorporated communities. One of them spans across the county line into the county south of us (Levy county). Each county is responsible for the 9-1-1 PSAP, and to accept 9-1-1 calls from within that county. That can get messy, when the calls originate from cell towers. So anyone calling 9-1-1 has to (when they get the wrong PSAP) ask to be transferred to the correct PSAP.
USPS has two post offices in this county, each with one 5-digit zip code. Different physical locations are contained within three different zip code, plus one of those two post offices extends south into an adjacent county.
Because the street addresses are laid out in a county oriented center-point scheme, that means that the same zip code can have addresses just north of the county line tagged as SE or SW, while on the other side they are all NW (with street addresses based on the respective county).
Lastly, 9-1-1 addresses are only given out when a building permit is issued. That means that large agricultural holdings have no real 9-1-1 address.
