For tagging the six Alaska “City and Borough of X” entities according to the “San Francisco” model, I show @stevea and @Minh_Nguyen with a thumbs up and no objection from @willkmis or @ZeLonewolf
I think consensus has been reached on these.
For Honolulu County / Honolulu, I should not have grouped this CCC in with the Alaska entities because Honolulu County / Honolulu are not coextensive, and so there are two entities in OSM and the question is really about how to tag those.
@stevea suggests admin_level=8 for Honolulu as a documented exception, but @ZeLonewolf has an emphatic “NO!”.
The current tagging is:
Honolulu County
– boundary=administrative, admin_level=6, border_type=county
Honolulu
– boundary=place, border_type=city.
So I’m assuming it’s safe to move forward with
the Alaska entities as per the “San Francisco” model
Honolulu County / Honolulu as per @ZeLonewolf (which is to keep the current tagging unaltered)
I’m moving forward with that sometime today or this weekend unless stopped
border_type=borough or borough;city seem uncontroversial for the Alaska borough cases. The Unorganized Borough is a weird case. I’m still not sure whether that’s a borough equivalent, or the name of an area where there’s no boroughs.
At one point, someone was creating boundaries for unincorporated parts of counties, which is a practice we should definitely discourage. So it’s unclear to me whether we should treat the UB as a place (the named area where there’s no boroughs but ones might be carved out in the future) or treat it like a borough and make them all space filling. So my only concern is making sure we’re not setting a precedent for people to map all sorts of other “in the absence of” boundaries, especially since this has already happened.
In short, I tentatively support @Minh_Nguyen’s suggested tagging for the UB since it has an explicit name, as long as we all agree that this isn’t some kind of carte blanche to tag unincorporated parts of places as a place.
Should the lowercase city’s boundary not have place=city to iteratively refine boundary=place? If so, border_type=* might be redundant because it conflates official status with a fuzzier notion of place. Along these lines, if we ever have a need to map the capital City of Washington as an administrative boundary covering part of the District of Columbia, it would probably also be a boundary=administrativeborder_type=city nestled inside the boundary=placeplace=city representing the popular notion of the city of Washington in lowercase.
Anyways, whether to tag the City and County of Honolulu as a city limits depends more on current state law than the origins under territorial law. It’s probably less important than in other states because of, you know, water. Lots of maps intentionally suppress maritime boundaries regardless of the type or level, except to clarify the possession of some minor outlying islands.
Agreed. The justification for the Unorganized Borough as an administrative boundary would be that the state actively uses it to organize the services it provides directly (instead of through a more local government). Such boundaries will always be more marginal and less surprising if omitted than if one with an active government and local identity were to be omitted.
While we have Brian’s wonderful synopsis for Honolulu provided to us by a one-time local that makes current-tagging a slam-dunk, I am almost tempted to make a phone call to the office of Alaska’s office of Secretary of State (if that’s the right place) about what might be correct to do with the six (eight with Haines and Peterborough, nine with UB?) divisions / boroughs in Alaska. I realize that it’s coming up on the weekend, and the answers might be researchable without that phone call (being pretty Western, Alaska IS “still open” this Friday afternoon as I type this), but being authoritative in these matters is what we’re after here. And, we’re doing a great job of this, in my opinion.
As for UB, I’ve read Minh’s justifications for keeping it as is, though it’s so odd (and huge, bigger than any other state!) that I’m inclined to get out of the way and let chips fall where they may. (I’ve been doing a lot of stepping aside — maybe my strong suits were a decade or so ago when I was simply trying to get my arms around the shape and scope of this as I exhorted others to pay attention and offer their more-local perspectives). I also agree that mapping what might roughly be called “unincorporated and/or unorganized” areas (such as UB or the unincorporated areas of counties) is a can of worms we don’t want to open. I’ve seen vast, huge beasts of complexity in Canada along these lines (there’s a link in our admin_level wiki mentioning one of these), and such beasts are indeed something to be discouraged.
So, as I’m not an expert by any means, I’m (happily?!) fading into the background regarding the Alaska six, eight or nine.
I once again offer my sincere, deep thanks to the dozens, perhaps hundreds of people over the years who have championed these causes and taggings so that good…better…best versions of them emerge. “We’re doing fine.”
Probably, yes. Unlike a boundary=administrative which can stand alone, a boundary=place is generally a supplementary object to a matching place=* node. Many data consumers only consume place=* on nodes since the area of a place can often be fuzzy and isn’t really a clear boundary.
border_type=* is only really defined in combination with boundary=administrative and maritime. Some other kinds of boundaries might also benefit from a secondary key that indicates the local classification more specifically, such as boundary=statistical or historic.
For example, when @ZeLonewolf protests that boundary=census should be reserved for census-designated places as opposed to census areas or census county divisions, I get to wondering whether border_type=* could be a useful tool for distinguishing the three, given that only admin_level=* distinguishes them today.
The ahupuaʻa of Hawaiʻi would probably deserve a unique tag of some sort. However, it’s unclear to me whether this information should go in the confusingly named border_type=* versus some other key, or if maybe they should be a distinct boundary=* because they don’t neatly fit into any other box.
Well, Discourse chat and a wiki table entry mentioning the Honolulu ahupuaʻa are at least something! True, these (specifically-named) data are not in our map with that name (though they are as a polygon), but ahupuaʻaare in our project.
I was not aware that anyone was using admin_level with boundary=census. My validator flags that as an error, and I’ve been removing it when I see it.
If we want to settle on a scheme where border_type disambiguates categories of census bureau divisions, I don’t have a particular objection to that, as long as we agree on the values. Ideally we’d make a big pass to add that tag to the CDPs we’ve mapped so far. I’ve been applying border_type=CDP to CDPs recently in anticipation that this might come up. I don’t validate for it because I don’t think we have consensus on it. There is also boundary=statistical, which has been tossed around as an option.
Originally, many CDPs wound up with admin_level=8 simply because all we did was to retag them from boundary=administrative to boundary=census to urgently get them off the rendered maps. I can see the logic in distinguishing the different census geographies by numeric levels equivalent to admin_level=* values. It would enable a geocoder to geocode a location to “Hyder, Prince of Wales–Hyder, Alaska”, as opposed to “Prince of Wales–Hyder, Hyder, Alaska”. The Census Bureau does this with their summary levels too.
That said, I don’t think this information necessarily needs to go in admin_level=*. That key is only documented for boundary=administrative and boundary=religious_administration, which are both “admin” boundary types.
If you mean that boundary=census is for CDPs while boundary=statistical is for all the other census geographies, I can see that approach as a concession that it’s too late to retag CDPs as boundary=statistical too. However, it wouldn’t absolve us from distinguishing the various geography types more explicitly.
If it were the case that boundary=census applies to U.S. Census Bureau geographies and nothing else, then admin_level=* might be sufficient. However, the same tag is also widely used in some other countries that conduct censuses, especially Ireland, where the admin_level=* values probably don’t map to the same terminology.
boundary=statistical is similarly being used for a variety of local and regional statistical areas that don’t have neat equivalents to U.S. Census Bureau geographies. If our TIGER cleanup efforts had used boundary=statistical from the get-go, there would’ve been even more need for a clarifying key such as border_type=*. Otherwise, mappers would’ve seen fit to import things like statistical neighborhood approximations from the local health or planning department, without any way to distinguish them from census geographies.
Other boroughs of Alaska – while I was editing Alaska I added border_type=borough to the 13 other boroughs, so now all 19 Alaksa boroughs have either border_type=borough or border_type=borough;city).
I did not add a border_type to the Unorganized Borough, pending the broader discussion of how to tag the UB.
Wikipedia talks as though there’s some distinction between the borough and the city (the “urban” part):
The central (urban) part of Wrangell is located at 56°28′15″N 132°22′36″W, in the northwest corner of Wrangell Island. The borough also encompasses the entire eastern half of the former Wrangell-Petersburg Census Area, in addition to the area around Meyers Chuck . . .
Offhand this looks similar to Honolulu, where there isn’t quite an administrative boundary but there’s an understanding of where someone is referring to if they say “I’m going to Honolulu” (or “Wrangell”).
Based off that I updated the tagging to be the same as Honolulu:
I added the original boundary to Wrangell (according to the history), but that edit was simply promoting a closed way tagged as a boundary to a relation. The true original edit adding that boundary was this one, and it came from the (in)famous 2008 TIGER import.
If you have a newer boundary geometry for Wrangell, my gut feeling is the other one should probably be deleted. But, I don’t know enough about Alaska to say.
TIGER was right at the time, but only barely: the City of Wrangell transformed into the larger unified borough on June 1, 2008, whereas we imported TIGER/Line data that was current as of January 1, 2008.
That aside, we do have a knack for catching TIGER on a bad hair day:
[I]n the 2008 TIGER/Line Shapefiles release we have identified 1,159 counties that contain new features that are distorted. These features appear to be pulled away from their position to a point and returned to their correct location forming a distorted edge. The problems are a result of the efforts to automatically integrate new data with existing positionally accurate features. We have not identified wide-spread issues in other counties but it is possible that there are additional counties with some distorted edges.
