Thanks, Brian – I apologize for the for the disruption on the OpenMapTiles and I (along with many others I’m sure) appreciate you quickly reverting those changes to get that service operating as it should!
I know it’s a separate subject, but it might be of interest to note that I didn’t change Nova Scotia from border_type=state to border_type=province because it already had the value of province: Relation: Nova Scotia (390558) | OpenStreetMap
Thinking that Nova Scotia was correct (with “province”) and the other Canadian provinces were wrong with “state” led me to “correct” the others, causing the OpenMapTiles issues.
I’m curious as to if Nova Scotia does or doesn’t show up on OpenMapTiles.
At any rate, I’m sure it’s a good example of the frustration that occurs on your end when a (well-meaning 
although not necessarily harmless 
…) new member changes something without the awareness of how it impacts anything beyond what they’re working on!
But I’ll keep this OpenMapTiles example in mind going forward and be more mindful that I’m working in a broader ecosystem.
Ok, I reverted most of the changesets I did from the last 24 hours, which is namely “reverting the revert” of the deletion of San Francisco County and Nantucket County from three-ish years ago and reverting my addition of Denver County and Broomfield County, Colorado.
For reference here are the four relations:
San Francisco County: Relation: 396487 | OpenStreetMap
Nantucket County: Relation: 2387088 | OpenStreetMap
Denver County: Relation: 18351788 | OpenStreetMap
Broomfield County: Relation: 18351840 | OpenStreetMap
Not to get too confusing, but of note is that Denver County and Broomfield County had also existed in the database at one point, but were deleted (or rather their county portion (admin_level=6) renamed to the city and the city portion (admin_level=8) deleted) several years ago and the OSM revert tool could not revert those, so I created them.
See version 65 where the admin_level=6 was turned into Denver from Denver County: Relation: Denver (1411339) | OpenStreetMap
Previous Denver entry that was deleted: Relation: 253750 | OpenStreetMap
See version 18 where the admin_level=6 was turned into Broomfield from Broomfield County: Relation: Broomfield (1411321) | OpenStreetMap
Previous Broomfield entry that was deleted: Way: 33153017 | OpenStreetMap
Point being, the county and city portions of those CCCs were on OSM at one point, just the same as the other 27 are currently.
Further, I can’t find any discussion at either community.openstreetmap.org or help.openstreetmap.org about deleting them, so I’m not sure what, if any, community consensus those deletions had.
Concerning the Wikidata solution – I’m aware of and was initially using property P131, “located in the administrative territorial entity”, but found it at times to be almost as inconsistent as well, which led me to the solution I currently have, which is to download the geojson from Nominatim and calculate it myself using the spatial features in MySQL.
This has worked relatively well – as long as I can get the geojson data for those four CCC counties on OSM! 
Ok, just to sum up where I think this discussion stands after a day:
ICs
It seems the only feedback so far is that border_type=city is the proper tag to use over place=city.
CCCs
Aside from the same feedback the border_type=* vs place=*, the main point of contention seems to be whether the county and city should be listed separately.
Aside from Alaska, having them listed separately is the case for 27 of the 31 remaining CCCs. The other 4 have the city only listed at admin_level=6 and border_type=city, the same as an IC.
One question that stands out to me is whether the county and city are coextensive.
Cusseta / Chattahoochee County, Georgia, form a CCC and their borders are not coextensive. Athens / Clarke County, Georgia, form a CCC and Clarke County has other municipalities still within the county.
Just a quick check shows at least Macon / Bibb County, Augusta / Richmond County, New Orleans / Orleans Parish, Philadelphia / Philadelphia County, and all five boroughs of New York City are coextensive with both county and city entries in OSM.
Statenville / Echols County, Anaconda / Deer Lodge County, and Camden / Camden County, all have both city and county in OSM but the county is a relation while the city is a node, as is Honolulu County and Honolulu although i recognize that’s a slightly different situation.
At any rate, I don’t see a difference between the four counties in question and the others that I’ve listed that are coextensive.
There may be options I’m overlooking but I would think we’d choose either to:
- include San Francisco County and the other three counties to match the existing 27, or
- delete the counties of CCCs that are coextensive with the city (such as Bronx County, Philadelphia County, Orleans Parish, etc.) to match how the city of San Francisco is entered, while keeping the ones that are not coextensive such as Athens / Clarke County in order to map those two features separately
Thanks all and looking forward to more discussion!