Talk:United States admin level

From OpenStreetMap Wiki
Jump to navigation Jump to search

Township boundaries

At least in Ohio, townships and cities/villages don't overlap (except in the rare situations where they do). So I think we should draw the township boundaries so that they stop at the city limits, where the welcome signs would be. I've seen maps that indicate the townships' original (square-like) boundaries, but those are historical boundaries. They don't correspond to any current administrative jurisdiction. – Minh Nguyễn (talk, contribs) 10:18, 1 December 2010 (UTC)

So do they or don't they overlap? If they do in real life, we should map them that way. --NE2 12:34, 1 December 2010 (UTC)
There are only a handful of places in Ohio where townships actually overlap with cities or villages, usually only by a few parcels. They're the exception rather than the rule: the municipality wanted to annex the property for the additional taxpayers, say, but didn't want to run afoul of the township by taking away theirs. (In the overlapping areas, property owners would pay taxes to both the township and municipality, as they belong to both.) I just wanted to point out that the overlap was possible, but it's separate from the fact that cities and villages occupy former township land. – Minh Nguyễn (talk, contribs) 09:07, 2 December 2010 (UTC)
After some additional research, it looks like townships should be admin_level=7 and should exclude cities that have withdrawn using paper townships. Census maps like this one of Hamilton Twp., Warren Co. can be pretty useful for figuring out which cities have withdrawn. The scenario I mentioned above would be implemented by including the parcels in the city but excluding them from the city's paper township. But I'm not sure whether cities with paper townships should be admin_level=7 or 8. – Minh Nguyễn (talk, contribs) 09:08, 5 December 2010 (UTC)

Planning areas

The master planned city of Irvine, CA, uses "planning areas" (also referred to poetically as "villages" by the real estate marketeers) to denote large parts of the city, each with own infrastructure and a bit of character. There are about 30 in all, this number may slowly grow. I am planning on mapping these out as boundary=administrative + admin_level=9 unless someone stops me. —Preceding unsigned comment added by Ponzu (talkcontribs)

I don't see these as quite the same as administrative boundaries, more like school districts or water management districts. Perhaps boundary=planning_area would be more suitable? Predominantly residential areas can of course be tagged landuse=residential as well. --NE2 06:01, 6 April 2011 (BST)
If I map them as boundary=planning_area, this would be the first example of such usage. And given the number of master planned communities around the world, it may not be of use to anyone else. I think my city is pretty unique for California, can't speak for the world. Check out this document (incl. the map): Doesn't it look like administrative division? And yes, I am planning on tagging the neighborhoods inside the planning areas landuse=residential--Ponzu 00:36, 7 April 2011 (BST)
This usage reminds me of the well-defined "neighborhoods" that many large cities are divided into. They're usually associated with neighborhood councils and don't necessarily coincide with wards or voting precints. Due to that analogy, I find your proposed tagging scheme reasonable. Irvine covers enough area that the additional boundaries wouldn't really clutter up the map that much. – Minh Nguyễn (talk, contribs) 08:30, 10 October 2011 (BST)

Townships, Cities and Villages in Michigan

Under Michigan law, townships and cities are at the same level under the county level. That is to say, in Michigan one is either in a township or a city. There are no areas that aren’t in either a township or a city and there are no areas in which one could be in both. When a city is incorporated or annexes territory from a township, that land is no longer in any way connected with the township, not even in the sense of “paper townships” like in Ohio. If the entire township is incorporated or annexed to a city, that township ceases to exist in every sense.

Villages are incorporated municipalities subordinate to townships. That is to say, when one is in a village he or she is also in a township. No part of a village shares territory with a city, however. As stated previously, incorporation as a city or annexing territory to a city removes if from the township and thus, if a village becomes a city it no longer is part of the township—in effect, it would be “elevated” from a level below the township to an equal level with the township.

Here are some complications: Cities in Michigan can be part of more than one county. Thus, although a city is clearly a subordinate administrative subdivision to the county on an equal level with the townships, when a city is part of more than one county it makes it more complicated to designate this. Creating two adjoining areas and giving them the same tags to indicate that they were the same city would still make it appear that they were two separate entities and not one—just with the same name. Villages in Michigan can lie in more than own township, and also in more than one county. There are many villages in Michigan that lie in more than one township within the same county and a few that lie in more than one township in two counties—yet, it’s the same village municipality. The same complications regarding cities would apply here.

Note: Michigan has two designations for townships: general law townships (usually simply called “townships”) and charter townships. There is no practical distinction between the two—particularly regarding mapping in Open Street Map. Charter Townships are not an incorporated municipality in the sense that a city is. Rather, they are townships that have exercised some local government options provided by state law and have some protection against being annexed by a city.

My approach: For my mapping, I have given cities and townships the same level: 7. I have given villages level 8. When a city spans more than one county or a village spans more than one township, I will split the area of the city or village into separate areas for each higher-level administrative unit in which it lies, but will apply the same tags to both (except for the "Is_In" designations).

Is there a better way to do this? --Vox Sapiens 01:47, 30 May 2012 (BST)

"Cities in Michigan can be part of more than one county." I think this is true in most states. Look at New York City - it comprises all the land of five counties. --NE2 10:16, 30 May 2012 (BST)
I think cities should be tagged level 8 as suggested by Admin level page so as to keep Michigan consistent with other states. Killian441 00:18, 30 November 2012 (UTC)

Updated table

I found it hard to read the table as it was so I updated it to its current form. I also alphabetized it. A couple of issues I was grappling with: New York City boroughs. According to the Admin level page, NYC itself is level 5 and its boroughs are level 6 as they are most similar in function to counties. I updated this table to reflect that. If anyone has objections let me know.

Counties in CT - Counties were abolished in 1960 but still used for courts, except in some places. I left a note there but not sure if counties should be included or not. Same with the regions that came after them which have only a limited authority.

Boroughs in VA - Boroughs in VA seem to be the same level as neighborhoods. The are officially recognized but don't seem to have any real authority. If this is the case and we want to map them then I don't seem how this is any different than neighborhoods, wards, voting districts, or census designated places. Perhaps we want to include these as well.

Independent cities - All cities in VA are independent and therefore level 6. But what about wikipedia:Consolidated city–countys? Independent cities are not considered part of any county and therefore it makes sense to have them level 6, but Consolidated city–counties are in fact counties so should they be level 6 as well?

Cities in MI - I tried to keep the discussion on this page in mind as I updated but I think the one conflict is cities in Michigan. I don't see a reason to leave them at level 7 and moved them to level 8. Again from the Admin level page municipal cities are level 8. My understanding is the levels determine how to render an area. Its seems logical that even though cities in MI have similar responsibilities as townships, they should be rendered the same as cities in other states, therefore level 8.

Killian441 00:14, 30 November 2012 (UTC)

admin_level=5 proposal

We may need to extend how New York City — a conglomeration of Level 6 entities (boroughs) — defines a new entity (Level 5 "city-glom"). This entity is still "below" the next admin_level up, Level 4 (state), hence the choice of NYC being admin_level=5.

Similar entities (which seem to fit best at admin_level=5) are created by both state and federal legislatures to serve a variety of purposes and include:

  • Probably state-designated Councils Of Governments
    Councils Of Governments are regional bodies that exist throughout the United States. These typically serve an area of several counties, addressing issues such as regional and municipal planning, economic and community development, cartography and GIS, hazard mitigation and emergency planning, aging services, water use, pollution control, transit administration, and transportation planning.
  • Likely MPOs, federally mandated and federally funded transportation policy-making organizations in the United States that are made up of representatives from local government and governmental transportation authorities
    (Local authorities + US Department of Transportation)
  • Possibly PSAs, CSAs, MSAs, µSAs, federally-designated
    (US Executive Office of the President, Office of Management and Budget)

—Preceding unsigned comment added by Stevea (talkcontribs) 02:20, 31 December 2012 (UTC)

See discussion on talk-us.
But don't all of these match existing city or county boundaries? --NE2 15:56, 4 January 2013 (UTC)
No, NE2. A COG, for example, might be a logical union of existing cities: made up of cities, but a new unit at a different admin_level which includes two or more cities or counties. The federally-designated PSAs, CSAs, MSAs and µSAs very often encompass areas quite different than city or county boundaries. They are calculated not by municipal or state legal limits, but rather a statistical edge, such as population density around a centroid. --Stevea 11:18, 19 February 2013 (UTC)
I'm using MSAs to replace the previously mapped 'statistical_area's imported in 2010 by NE2. -Valerietheblonde (talk) 19:41, 14 July 2016 (UTC)

The bottom line: it has emerged as consensus that COGs, MPOs, similar "special purpose districts" and statistical areas defined by the US Census Bureau are tagged with neither boundary=administrative nor admin_level=* of any value. Stevea (talk) 20:18, 20 July 2017 (UTC)

School District admin level

I was thinking of adding the boundaries of a few school districts in my local area, but found there doesn't seem to be documentation on the wiki on the proper designation for them. It looks like they would fit here as an administrative boundary under level 9 or 10. Does it make sense to fit a School District under admin_level 9? Or is there another designation that should be used for this sort of boundary? --Midnightlightning (talk) 17:31, 17 February 2015 (UTC)

boundary=administrative is used for corporate boundaries, the kind of boundaries that have welcome signs. Overlay districts don't fit neatly into this admin_level hierarchy. Some school districts cover only part of a city while others cover entire counties or span county lines. In many states, they have independent boards that answer to the state education department rather than a city or county government. The same is often true of fire, water, and sanitation districts. You could use another tag like boundary=school, but consider that school districts can be more fluid and less on-the-ground verifiable than other kinds of districts, which partly explains why so few have been mapped. – Minh Nguyễn (talk, contribs) 10:33, 18 February 2015 (UTC)
Well, I spotted my first school district welcome sign. [1] – Minh Nguyễn 💬 17:19, 7 May 2019 (UTC)

Consolidated city-counties beyond a single county: admin_level=5?

Consolidated city-counties (CCCs) in the US are complex, with many subtle distinctions about how they are organized, such as "consolidated since their creation," "merged," and "merged with some independent municipalities." (CCCs are distinct from independent cities). New York City (NYC), presently OSM's only US example of admin_level=5, seems unique in the sense that each of five boroughs (admin_level=6) is coterminous and co-extensive with one and exactly one county. However, county-level NYC government is non-existent as all executive and legislative power is exercised by the (parent) city government throughout the five boroughs, conveying that admin_level=5 is correct, even as admin_level=6 is correct for (administratively meaningless?) borough boundaries.

Borough boundaries are not entirely meaningless administratively. Boroughs of New York city still retain some independence. The Borough Boards have (small) discretionary spending budgets. Each borough has its own county court, and retains its own district attorney to prosecute crimes. Brooklyn and Queens have their own public library systems independent of the New York Public Library. Other government functions have been consolidated and redistributed several times since the Great Consolidation of 1898, and could hypothetically be again. (For each borough to have one vote on the Board of Estimate was found to be unconstitutional in 1989, and that house of the city's legislature was abolished, with most of its responsibilities devolving upon the City Council.) Kevin Kenny (talk) 22:45, 16 March 2017 (UTC)

Other CCCs exist in the US, although most seem to stay within one county. This discussion asks and intends to address: Are there CCCs in the US which extend across more than one county, thus deserving an admin_level=5 tag? An initial possibility might be Unigov as greater Indianapolis, since it includes Marion County and portions of Hancock County. Or is Unigov more of a CCC best tagged with admin_level=6? If the answer is "No, there aren't any US multiple-county CCCs besides NYC" then this discussion is moot. Stevea (talk) 21:55, 16 March 2017 (UTC)

Great Question. NYC is the exception to the rule with CCCs. NYC consists of 5 county-boroughs. There are 8 CCCs within the US. Their boundaries could be downloaded here Consolidated Cities. There are many cities that are in many different counties but they remain separate governments. For more on Consolidated City-County please read Wikipedia or Census --Jonwit (talk) 14:17, 17 March 2017 (UTC)

Thanks, Jonwit. There are links to the Wikipedia pages you mention embedded in our wiki already, they were actually helpful to launch my understanding of what I already do know about these, although the page seems to indicate there may be more than 8. Again, "many subtle distinctions" seems to be the key phrase here. I also received email from somebody in the Indianapolis City GIS office who says that Unigov is a single-county (Marion) CCC, even though Cumberland is in Hancock County. Stevea (talk) 17:15, 17 March 2017 (UTC)

Although not a CCC, in Oregon we have Metro, which is a regional government around Portland that includes areas in Multnomah, Clackamas, and Washington Counties. According to them, Metro is the nation's first directly elected regional government. I don't think this boundary is in OSM yet, but I did talk with some locals a while ago and the general consensus was to make a boundary relation for Metro with admin_level=5. Also interesting in Oregon is the fact that cities often span more than one county. Cleaning up admin boundaries has been on my TODO list for a while. --Dobratzp (talk) 17:35, 17 March 2017 (UTC)

Hi Peter. Oregon's Metro is identified in its charter as a "metropolitan planning organization" rather than a "full" government. If such a boundary were entered into OSM, I would say (and have, please see the corresponding Page to this Discussion where this is documented) that instead of Metro being tagged boundary=administrative + admin_level=5 that it be tagged boundary=MPO. If Greater Portland's Metro were actually a "city-glom" (such as NYC made up of multiple boroughs/counties), or an actual CCC, or even an "independent city," then boundary=administrative would be appropriate. But as a planning agency, mmm, no. My opinion, but bolstered by an earlier proposal I made to promote such entities (California has "LAFCOs" which are similar) to admin_level=5 and it pretty much went nowhere, dying from an utter lack of consensus. And correctly so, I now agree. Stevea (talk) 17:56, 17 March 2017 (UTC)
Although Metro serves as the MPO for the Portland region, they are truly a form of regional government that does much more than manage a small portion of federal transportation money. Also notable is that the president and six counselors of Metro are elected by the 1.5 million people that live in their administrative area. Metro collects taxes and has a budget of $21 million with a staff of over 800 people. I don't think California's LAFCO's are that similar to Metro.--Dobratzp (talk) 20:13, 17 March 2017 (UTC)
I certainly grant that Metro's responsibilities are large compared to other special-purpose districts. (Although $21 million is not huge, especially with 800 people; if 100% of Metro's budget pays for salaries, and it can't, each of the 800 earns much less than a living wage). However, it is not a single exclusive government providing all services and offering complete representation to the People who elect its "Agency Executives." It is an agency. Its predecessors were a "Metropolitan Service District" and an "Association of Governments" (the Page to this Discussion already has identified a suggestion for tagging "Councils of Government"). Neither of these, nor Metro in its present form, is truly a government, what admin_level=* describes. My water agency has a budget (MUCH smaller than Metro!) and provides multiple services (sewage, water, garbage) and yet neither is it a government, it is a "special (purpose or purposes) district" for which admin_level=* is not the right tag. I do maintain (but do not insist, I am very much in listening mode here) that boundary=MPO is an appropriate tag for Metro, although if you wanted to use boundary=COG or coin boundary=SPD for "Special Purpose District" (as the Wikipedia page you point to defines Metro as "Regional Special-purpose district and Metropolitan planning organization") I certainly do not have a problem with that. Truly, rather than digging my heels in, I listen here, but I do ask for substantially more evidence that Metro is a government rather than the direct confirmation that it is as I originally described it should be tagged, as an MPO or something besides boundary=administrative. To your points, the Oregon Blue Book does call Metro "a regional government responsible for managing issues that cross city and county lines" but this seems to straddle both of our perspectives: a weak version of "government, but with limited responsibilities" as well as a weak "non-government, but vast." Also, voters gave it "home rule" which I am finding elusive to define in this specific case. (Oregon is a Home Rule state, but Dillon's rule does not apply. So what does "Home Rule" mean in Oregon? Such questions are not my strong suit.) A wider discussion seems prudent. Stevea (talk) 21:12, 17 March 2017 (UTC)

These are the 8 identified by the 2016 U.S. Census Consolidated Cities

  1. Nashville-Davidson metropolitan government
  2. Butte-Silver Bow
  3. Louisville/Jefferson County metro government
  4. Greeley County unified government
  5. Indianapolis city
  6. Athens-Clarke County unified government
  7. Augusta-Richmond County consolidated government
  8. Milford city

Any others that are identified (i.e. Philadelphia, NYC, San Fransisco) are not official County-City Governments but for the purposes of OSM should get admin_level=5. I guess the larger question is if a city spans multiple counties (many do) Do they get upgraded to admin_level=5? --Jonwit (talk) 18:48, 17 March 2017 (UTC)

The perspective of the Department of Commerce is appreciated, Jonwit! Sometimes this does not exactly align with OSM consensus, as the 2012-13 talk-us dialogs mentioned in the Page of this Discussion indicate: OSM now tags with boundary=census for Bureau of Census boundaries, rather than boundary=administrative. Some interesting things about your most recent paragraph above. Your list above says Greeley County and Milford city are both CCCs, while Wikipedia does not. Wikipedia says that Jacksonville-Duval County Florida and Macon-Bibb County, Georgia are both CCCs, while your list does not. Also, Wikipedia lists DOZENS MORE (well, let's say "over thirty more") than your list of only 8. Please don't get me wrong: I'm not saying that the Census Bureau is "all wrong" and Wikipedia is "all correct." The truth lies somewhere in good discussion like we are having here. It does seem to me (and not only me, but wider consensus) that these (and independent cities) are correctly admin_level=6 UNLESS they span more than one county, like NYC, which seems to be the only CCC which does. If, as you say "many do" I ask you to kindly enumerate these. Stevea (talk) 19:39, 17 March 2017 (UTC)

There are 35 Cities\Towns just in the state of Tennessee that span multiple counties. It is silly to name them all... here are two Johnson City, Spring Hill. Both much smaller than larger cities in the state. Counties are mostly an independent place boundary from cities. As for the differences it lies in the definition of what a CCC is. I would stick with population or area when distinguishing between admin levels.--Jonwit (talk) 19:59, 17 March 2017 (UTC)

Ah, Jon, I think we are having a minor misunderstanding; perhaps I haven't worded my original question well enough. The examples you offer in Tennessee are single cities which cross boundaries of ("span") multiple counties. As these are a single entity at the jurisdictional "level" of a city, I would tag (and I believe other informed OSM volunteers would tag) these with admin_level=8 precisely because these ARE cities. And their parent counties should be concomitantly tagged admin_level=6 precisely because they are counties. But those examples are not exactly what I am asking for. What I am asking for are CCCs which are made up of multiple counties (or county-equivalents), like NYC. Earlier (yesterday, as I started this discussion) I thought Unigov was such a thing because I misunderstood that it was made up of at least a part of a second county, but I have since been corrected by somebody quite knowledgable. So, I continue to ask: are there CCCs in the USA made up of MORE THAN ONE county? What admin_level=* is all about is "does this entity in the hierarchy fall above or below another one in the hierarchy, and if so, what is its level?" Finally, while the Census Bureau (and even some states as they define "First Class Cities" and "Second Class Cities," etc.) might find population or area convenient methods to statistically align/compute differing levels of political administration, neither population nor area are part of the definition of admin_level=* in OSM. Thank you for the continuing excellent discussion. Stevea (talk) 20:19, 17 March 2017 (UTC)

INCOG around metro Tulsa, Oklahoma may count. Paul Johnson (talk) 11:42, 22 March 2017 (UTC)
Paul, one might easily think so, but it is a COG (boundary=COG) as described in the Page: its website's front page says "INCOG is one of 11 Councils of Governments in the State of Oklahoma, and one of several hundred regional planning organizations across the country. INCOG provides planning and coordination services to assist in creating solutions to local and regional challenges in such areas as land use, transportation, community and economic development, environmental quality, public safety, and services for older adults." I can see new three-letter acronyms being coined for these things: SPD as described above. However, as I don't fully understand the subtle distinctions between these, we might begin by sticking with the established ones. (And in the case of INCOG, it does self-identify as COG, though it also calls itself an MPO, if that is indeed something distinct). Finally, this section intends to discuss what are perhaps more-clearly stated to be "consolidated city-counties made up of multiple entire counties or county equivalents" (as NYC is). INCOG is not that sort of entity, though I certainly concede that such distinctions are not always easy to see. We did discuss (two sections above in this Discussion) whether COG-ish things are admin_level=5 four+ years ago, consensus emerged they are not. Similarly, NYC emerged as "yes, it is." This discussion seems to be heading distinctly to the conclusion that NYC is the only admin_level=5 in the USA. Of course, additional discussion enlightened by true insight and real research is welcome. Stevea (talk) 18:34, 23 March 2017 (UTC)

Article needs to be streamlined

Over the past nine months, this article has steadily grown in size and detail to the point where it more effectively educates readers on the arcana of governmental organization than on the task of editing OpenStreetMap. The pervasive use of jargon and acronyms serves to alienate inexperienced mappers who are often pointed to this article as a primer on boundary tagging. (In fact, I've taken to linking to this revision fully a thousand edits ago for usability's sake.) This wiki should defer to Wikipedia for complete definitions of each of these governmental or statistical structures and focus on how to translate bona fide administrative boundaries into boundary=* and admin_level=* tags. Concepts ineligible for inclusion in OSM should get nothing more than a sentence saying so. Most likely, any references to specific laws and dates can be removed without impacting the quality of the map. – Minh Nguyễn (talk, contribs) 23:33, 2 July 2017 (UTC)

Fair criticism, Minh, though I am disheartened that your choice of "fully a thousand edits ago" could just as well have been stated "from several months ago." (The difficulty of editing wikis, especially with complex tables, simply requires trial and error editing, increasing the number of edits to "very large," I agree).
Point by point: the arcana of governmental organization in the US is complex, the reasoning is described in a few lines of introductory text which are not excessive in my opinion: I find this text to be precise (if not exactly terse) to abstract the upcoming complexities. Would you agree or not? I ask that we begin your streamlining proposal by you suggesting what might be trimmed from that intro text as excessive.
Jargon should be specifically identified as such (by you, here, please) and in my defense I say that acronyms are properly introduced and used instead of lengthy jargon. Just because a concept is needed, and needing to be described, does not make it jargon. And an acronym for that jargon when it is repeated is perfectly acceptable language, especially when introduced properly. There are several such concepts in this wiki which are distinctly new for many OSM readers, that is exactly why they are introduced and abbreviated as they are. I welcome improvements, though I question deletions, unless they truly clarify.
The article has numerous references (and essentially, deferences) to Wikipedia, where appropriate. The syntax for inline linking is used correctly, I believe, though I welcome improvement. If you believe these excessive, please identify them and why.
Please identify what you mean by "mixed messaging" regarding non-administrative boundaries (in your Summary). If I am not mistaken (and I might be), the article identifies distinctions between these, and asks users to tag boundary=administrative where it is identified (in the text or table), and to tag with another boundary=* value when this is not the case. I believe this is both explicit and somewhat succinct, but again, I welcome a tightening up of the text, though not necessarily the concepts nor their wholesale deletion, as you did with MSAs and similar entities (μSAs, CSAs, PSAs, CBSAs). OSM has not achieved a true consensus about what to do with these, and it is precisely because of this that I welcome their entry into the map database by way of explanation here while simultaneously welcoming further discussion as to whether they are "proper" (having achieved wide consensus, really) entities for entry.
You say "since [statistical] boundaries always correspond to the boundaries of counties and county equivalents, they can be viewed as mere collections of administrative areas rather than boundaries in their own right." This is incorrect: statistical boundaries nearly never correspond to the boundaries of counties and county equivalents. While you are correct regarding the subsequent link to "Relations are not Categories," I am not sure I agree with your deletion of mine that suggests we no longer enter statistical boundaries with specific tags that identify them as such. These are pervasive in the world of mapping (especially with our US Census Bureau) and while we might reach a consensus that these absolutely do not belong in OSM, I don't believe we are there yet. TagInfo does not show occurrences of these, but various wikis and discussions I have had demonstrate that some want to enter them. Let's decide (here and now? once and for all?) whether these belong in OSM.
Please identify cases of "specific laws and dates" which can be removed without impacting the quality of the map. I welcome deletion of these as they are identified.
This wiki was never intended to be a "primer" on boundary tagging. As its title states, it is a comprehensive wiki on admin_level=* tagging, especially selecting the proper value for a given jurisdiction, in the United States. There are thousands and thousands of these, it is complicated, pains and steps have been taken to keep the table uncluttered given the challenges. In the spirit of your wish to clean up this article (and perhaps make a rudimentary "primer" out of a complex topic), I have stated an explicit instruction to users who wish to "skip to the chase." Right after that intro text, a one line instruction is offered to wiki readers, (paraphrasing myself): "Tag an administrative boundary (like this), guided by this table and wiki. (And then the table, first)." As any given boundary is going to be a single cell in a single row, both readily determinable, that does seem to be a streamlined focus on how to "translate bona fide administrative boundaries into boundary=* and admin_level=* tags" (as you suggest), doesn't it?
Thank you for your proposal to streamline. I trust it will yield fruitful results. Stevea (talk) 00:20, 3 July 2017 (UTC)
First of all, I apologize for my rant above. That was unnecessary, especially the cheap shot about the number of edits you had made. In lieu of a detailed critique of the article, I've prepared a separate article, WikiProject United States/Boundaries, which is closer to what I'd be comfortable showing to less experienced mappers. I don't have a strong opinion on whether it should complement or replace this article, but I would like for there to be an article similar to the one I've written. Even with the new instructions, the existing article would surely overwhelm someone mapping their first or second boundary.
As you can see, the new article doesn't necessarily contain less prose, but it does focus more heavily on the practical aspects of editing OpenStreetMap. The idea is that a mapper already knows the legal status of a given place or can look it up somewhere else (on Wikipedia or an official site). The new article simply translates that knowledge into tagging guidance. It also gives a small amount of historical context around the state of boundary tagging in OSM, but it steers clear of the distinctions between various administrative structures, since Wikipedia is better positioned to provide that context.
Editorially speaking, I deemphasized or omitted any tagging that has yet to take hold, even if it would make plenty of sense, like townships in Minnesota. Separate proposal pages, talk pages, and mailing list threads are more appropriate venues to discuss and document proposed tagging schemes. On the other hand, it isn't a static document: if someone comes along and maps some townships in Minnesota, we should absolutely update the article accordingly. I also omitted probable tagging errors and moved exceptions to the bottom, since they've already been mapped to completion.
Specifically regarding statistical areas: every mailing list thread over the past several years regarding CDPs (which are linked from this article) has ended with the unchallenged assertion that they aren't administrative boundaries and probably aren't verifiable enough to be in OSM. Moreover, the community has been actively removing or retagging CDPs since at least 2009. So I find it unlikely that there would be much movement in the other direction in the foreseeable future. On the other hand, it wouldn't bother me much if I started seeing a novel tagging scheme for MSAs that consists of a non-boundary relation with county relations as members. But let it happen on OSM first before we start documenting it. An article would muddle its own message by saying, "Don't do this, but if you were to do this, here's how."
In short, I think the new article trades precision for efficiency. It's descriptive rather than prescriptive of OSM. It isn't perfect, and your feedback is welcome as always.
 – Minh Nguyễn (talk, contribs) 10:51, 3 July 2017 (UTC)

I begin by replying with my thanks for your reply and congratulations an excellent start to the Boundaries wiki. I like it and it can and should certainly coexist with this wiki, at least for now: both are living documents. OSM has room for many voices speaking together about a large topic and a large number of topics and we often have specific audiences in mind as we do so as wiki authors. "There are many books in the library" seems a correct place for us to continue. We often and usually seem to complement each other in our efforts in this project, I see that continuing here and now and in the future. Stevea (talk) 19:41, 3 July 2017 (UTC)
That said, "descriptive vs. prescriptive" is a sharp dichotomy which is not necessarily sharply true here (either way) but more like "a deliberate and even delicate balance of both in each and every case." (At least with our good, better, best efforts). Wiki authoring is both to offer guidance and to be comprehensive. As we are prescriptive: "do it like this," we also wish to be (must be) descriptive: "here is what the community does, is doing, is talking about doing and might do in the future given the wide spectrum of conversations on this and related topics." We do our best to "get out of the way" as we strive to set a good example of "do it like this." This is a medium- to longer-term conversation. I am in no hurry to "solve" it, rather I wish to see it continue to evolve as it is evolving. And it evolves. I will have more feedback as the future unfolds, as will you and others. Looking forward to it. Stevea (talk) 21:13, 3 July 2017 (UTC)

Nine state improvement

I am posting in talk-us a request to improve the nine states' admin_level tags identified here:

Here are what exist in state constitutions/statutes/the real world, map well onto OSM's admin_level scheme, yet do not exist in OSM's data:

Rhode Island 7/Town, 9/Village: all are marked as 8/City when perhaps some are 7/Town or 9/Village
Massachusetts 7/Town: all are marked as 8/City when perhaps some are 7/Town
Maine 6/Unorganized territory and 6/(unincorporated) Plantation
Vermont 8/Village: all are marked as 8/City when perhaps there are 9/Villages in some 8/Cities
Pennsylvania 7/Township, 7/Borough are missing throughout, 8/Town subordinates to Borough, 8/Village and 8/Hamlet both subordinate to 7/Township
Connecticut 6/Region (not County), or both? Harmonize these
Minnesota 7/Township, 7/Town (it appears simply that none have been entered)
Illinois 7/Township, 7/Precinct?

New Hampshire, 8/Town: shouldn't these be 7/Town (as inTownship)? Are there 7/Organized Locations?

To read this, then perhaps participate in first discussion, then possibly "solve" these issues, take the second line of Massachusetts as an example. Massachusetts has done the MassGIS import, which included "City" boundaries and set their admin_level values to 8. However, I assert (politely) that Massachusetts also has "Town" boundaries (sometimes called "Township" and by consensus yielding an admin_level value of 7) which either are or aren't in OSM (I can't tell) and which should have their admin_level set to 7. But they do not. Again, OSM seems to need to identify "which, whether and how" we do this, on a state-by-state basis, in identifiably (only) nine US states.

Please discuss here. Thank you. Stevea (talk) 19:22, 8 July 2017 (UTC)

I've lived in Rhode Island and New Hampshire for a while. I don't think the notion of a cities being subordinate to towns as in Connecticut applies to the rest of New England. There's really no level of government between the county and municipal level.
I think we want to retain county boundaries even if many functions of the county government have been moved up to the state level or down to the municipal level.
Here's my thoughts about New England summarized in a table:
State: 4 5 6 7 8 9 10
Connecticut N/A County
Town City District N/A
Massachusetts N/A County N/A City Ward Precinct
Town Precinct N/A
Maine N/A County N/A City District Precinct
New Hampshire N/A County N/A City Ward N/A
Town District N/A
Rhode Island N/A County N/A City
Town Village N/A
Vermont N/A County N/A City
District Ward
Town Village N/A
--Dobratzp (talk) 02:31, 9 July 2017 (UTC)

Thank you, Peter: that's a very nice table. Just to put a bow and ribbon on it, you publish that table to mean "how you believe admin_level=* values OUGHT to be in OSM," right? Stevea (talk) 05:33, 9 July 2017 (UTC)
Yes, these are my tagging suggestions. I think all of the admin_level=6 and admin_level=8 in Massachusetts and New Hampshire agree with this table. The admin_level=9 and admin_level=10 vary a lot by municipality (and in many cases do not apply).--Dobratzp (talk) 05:44, 9 July 2017 (UTC)
Administrative subdivisions in New England are historically complex, the linked Wikipedia article makes that abundantly clear. I had done some research, mostly Wikipedia, distinctly avoiding how the Census Bureau categorizes, and I believe OSM should value the understandings and communications of its very own contributors, as in Peter's table above. However, this seems to require ongoing dialog as to what is actually authoritative and how that "properly should" map onto OSM's admin_level=* scheme. My research continues, more contributions and suggestions are welcome here. Stevea (talk) 18:12, 9 July 2017 (UTC)
Limiting the discussion for the moment to New England and using the Dobratzp table above as a template plus adding additional research (especially the previous link), I have discovered some discrepancies with his table:
In Connecticut, I can't be sure that there is District at admin_level=9, so I have placed a ? by it. However, I am reasonably sure that in the whole of New England there are no administrative subdivisions at admin_level=7 (sometimes called Townships, especially in the Midwest), this is consistent with the rest of his table. Connecticut is also one of the two states in New England that can be said to have an incorporated subdivision of a Town (which I am comfortable tagging admin_level=9), called a Borough. Vermont also has this, called (incorporated) Village. Also in Vermont, I could find neither evidence of an admin_level=9 subordinate to a City (as he says) called a District, nor an admin_level=10 subordinate to a District (as he says) called a Ward.
In Massachusetts, I (temporarily) concur with his table, although I could find neither evidence of an admin_level=9 subordinate to a City (as he says) called a Ward, nor an admin_level=10 subordinate to a Ward (as he says) called a Precinct, nor an admin_level=9 subordinate to a Town (as he says) called a Precinct.
In Maine, I (temporarily) concur with admin_level=5, 6, 7, 8 in his table, although I could find neither evidence of an admin_level=9 subordinate to a City (as he says) called a District, nor an admin_level=10 subordinate to a District called a Ward. I also (unusually) explicitly include (unincorporated) Township, as these "leftover" areas may (or may not) be characterized as admin_level=8, I'm not sure.
In New Hampshire (and Vermont and Maine), the items of Purchase, Grant and Location (and Gore, Surplus, Strip) are "leftover" areas, not included in any town. I believe they may be characterized as unincorporated state territory and so are not included for any value of boundary=administrative. This leaves "Organized Location" and "Unorganized Township" as remaining values which deserve admin_level=8, although I am certainly open to discussion on that point. I could find no evidence of an admin_level=9 subordinate to a Town (as he says) called a District, although I am open to discussion on that point.
In Rhode Island, I fully concur with that row in his table, with the exception that there really do not appear to be counties in that state.
As a lifelong Rhode Islander, I object to the exclusion of Rhode Island counties from admin_level=6. They are truly first-class subdivisions that divide the state and group cities and towns together. If we allow government-less CT and MA counties to exist with admin_level=6, RI should be tagged the same way. In addition, we allow, for example, NH towns with zero populations to have admin_level=8 even though there clearly is no government in place. In the US, the common understanding of admin_level=6 means "counties and their equivalents". OSM should not be adjudicating whether or not a county has a sufficient level of government function to "qualify" as a county. If it's commonly understood to be a county, it should be tagged as a county. Let's use common sense here and not take an interpretation that's so pedantic that the average citizen would understand it to be wrong. ZeLonewolf (talk) 02:40, 4 September 2020 (UTC)
This is a delicate subject and I (and others) have struggled with it several times already (in Connecticut and Massachusetts, now Rhode Island). However, I assure you, "struggle" is NOT what I want on this topic. And Rhode Island seems the most clear-cut of these three: quoting from Wikipedia (the original source is a dead link): "Although Rhode Island is divided into counties, it does not have any local government at the county level. Instead, local governance is provided by the eight cities and thirty-one towns. Counties in Rhode Island have had no governmental functions since 1846 other than as court administrative and sheriff corrections boundaries which are part of state government." Accordingly, it would appear that admin_level=6 is truly inappropirate in the state, "common understandings" notwithstanding. (And, alternative tagging can and did express this accurately, which I believe it should, following good and already-existing OSM conventions). I believe OSM DID accurately tag Rhode Island — as it excluded admin_level=6 — without being pedantic at all. An average citizen can read Wikipedia, for example, and discover that for almost 175 years, the state has had "no governmental functions" (what admin_level=* means) at the county level. I can understand that, I believe you can, I believe average citizens can, too.
Additionally, the state's counties do exist in OSM, they are tagged similarly (last I checked) to how the Connecticut and Massachusetts counties which also have "no governmental functions" were tagged: with boundary=region. However, it appears (at least Kent, I have not examined the other four) that you (ZeLonewolf) have recently (in the last week) changed them to boundary=administrative + admin_level=6 as appears to be your desire. I will ask you to please note once again that it is the Rhode Island Secretary of State (via website not readily available, again, so denoted on Wikipedia's article "List of counties in Rhode Island" who says: "Although Rhode Island is divided into counties, it does not have any local government at the county level." With great respect to you, this makes an assertion of boundary=administrative distinctly incorrect. I ask that you redact your changes to these counties, changing to boundary=region + border_type=county (and delete admin_level=6) as is (or was) done in Massachusetts and Connecticut. Thank you. Stevea (talk) 03:26, 4 September 2020 (UTC)
I made the change to all 5 counties almost three weeks ago in a single changeset, upon reading the discussion in the talk-us archives from June, which I reference in the changeset. What I understood from that discussion was that the change was made unilaterally, and by the end of the thread there was at best no clear consensus and thus I changed the map to the status quo. I am well aware of your position, which has been stated repeatedly, nevermind the quoted "other than as court administrative and sheriff corrections boundaries" which are governmental functions and no different from adjacent Tolland County, CT. Many reasonable arguments were made against removing boundary=administrative + admin_level=6 but those voices were steamrolled over by a small number of users that desire to enforce this -- in my opinion -- unreasonable interpretation of boundary=administrative. I believe this issue is significant enough that it ought to be raised to a higher level working group as the consensus process is at best producing an inconsistent interpretation between essentially similar jurisdictions, and at worst imposing a minority interpretation by outsiders. If you insist on forcing this change in RI unilaterally, I will not interfere and instead log my objection as a changeset comment -- I have no interest in starting an edit war. As a side note, I would expect that future RI mappers (unaware of these discussions) will almost certainly "discover" the lack of boundary=administrative + admin_level=6 on our counties and "fix the problem" which will re-open this sorry debate again and again. Respectfully, ZeLonewolf (talk) 05:14, 4 September 2020 (UTC)
We simply discuss here, there has been no "insistence" (on my part) of "forcing this change in RI unilaterally." (And, see below). I don't wish to start an edit war, either, I much prefer discussion (here is good), or, as you suggest, that this topic be "raised to a higher level working group." (Though I don't believe OSM is structured to have "higher levels" unless we include the DWG or LWG and they get involved; they are not now involved). You can see in this very wiki page how I have striven for many years, with much sweat and toil, with much patient consensus-building to effect "more correct tagging" in this realm (Nine state improvement, where admin_level values are distinctly more difficult compared to 41 other states). On this subject, it feels diplomatic for me to largely "bow out" now: firstly, because of the topic's obvious contentiousness and secondly, because I live on the West Coast it appears I am seen as an "outsider" to having much say about New England political jurisdictions. (Though, like most anybody and as I allude to above, I can read statements which seriously bolster my position made by the Secretary of State of Rhode Island in a Wikipedia article, if not the state website by the Secretary him- or herself — and for over 170 years, too). Please note that contrary to "unilaterally" or "edit war"(ring), it was not me who last edited these county tags, but you, ZeLonewolf. I defer to this most recent tagging of yours, even as I politely and heartily disagree with you. I am curious to see if / how you "elevate" this to a working group, as we might all learn something from that endeavor, whether in the realm of political jurisdictions, OSM consensus-building machinations, both, or something else entirely. Stevea (talk) 05:37, 4 September 2020 (UTC)
In Vermont, I fully concur with there being a Village at admin_level=9 subordinate to Town, though I do clarify by explicitly preceding it with "Incorporated." I could find neither evidence of an admin_level=9 subordinate to a City (as he says) called a District, nor an admin_level=10 subordinate to a District called a Ward.
Peter, here is a table which captures these understandings so you/we may make a direct comparison. Note that these changes are already in the "big table" in the Page for each state noted here; the table is simply my proposed "states structures should look like this," having started from your table. I ask you for your comments here and now. To everybody else reading this, I ask for constructive criticism on how admin_level=* tags SHOULD be, given the actual law and political reality in those states (starting with New England). Apologies for the length and tedium, this is a fair bit of sweat and work. Stevea (talk) 01:21, 10 July 2017 (UTC)
State: 4 5 6 7 8 9 10
Connecticut N/A County
N/A City District? N/A
N/A Town Incorporated Borough N/A
Massachusetts N/A County N/A City Ward Precinct
Town Precinct N/A
Maine N/A County N/A Town
New Hampshire N/A County N/A City Ward N/A
Organized Location
Unorganized Township
Rhode Island N/A N/A N/A City Ward
Town Village N/A
Vermont N/A County N/A City N/A N/A
Town Village N/A

Postscript: Peter, I agree with you that in Rhode Island (and Connecticut) we wish to retain the county boundaries. Especially in Rhode Island, I believe the simple solution is to not tag them with boundary=administrative nor admin_level=6. Please see here for data using tag border_type=county, which seems to have emerged as consensus on what to do for this specific case.

Here are some references of sub admin_level=8 division possibilities in New England. These tend to vary a lot by municipality. In some cases, there are districts or wards which directly correspond to elected members of local government and in some cases it's not clear what the divisions are supposed to indicate, apart from a voting location.
--Dobratzp (talk) 02:15, 10 July 2017 (UTC)

Awesome! I'm now "sold" thanks to your examples. I wanted to say that before I retire for the evening; it's late here. More tomorrow and during the week, I'll have more to say later. Stevea (talk) 07:27, 10 July 2017 (UTC)

In "the big table" (in the page, not directly above) I have added District and Ward to 9 and 10 for Vermont's 8/City. That's the only change "back" to Peter's assertions that his excellent evidence has boosted into the "should be like this" we are entering in this wiki, so that is now completed. G'night. Stevea (talk) 08:02, 10 July 2017 (UTC)

Re: Illinois- I'm familiar with rural Illinois, where townships are a well defined governmental entity with distinct boundaries. I'm not well versed in internal government workings, but never recall precincts except in a voting context. Cook County / the city of Chicago may be special. MikeN (talk) 21:03, 10 July 2017 (UTC)

Thanks, Mike. I had thought so, too (from Chicago-based TV cop shows?!), then I saw this. It says that of the 102 counties in Illinois, 17 (west central, southern) of them are divided into 261 precincts and the remaining 85 counties are divided into 1433 townships. I don't know exactly where this leaves us, except it appears that the current entry in the "big table" is accurate. At least, it is harmonious with that Wikipedia article and "the rest" of what we know. As for what is actually entered into the map, I don't believe we are there (or even close) yet. Stevea (talk) 21:22, 10 July 2017 (UTC)

Re Boston in the Nine States

Massachusetts looks correct in the small table to my eye in terms of legal entities, Wards and Precincts are primary fine-scale legal entities. I know my Ward and Precint numbers in Boston, so can confirm they exist. OTOH there are no "signs on the ground" for Wards or Precincts.

As noted above, there are also Council Districts in Boston, but their mapping onto Wards/Precincts will *change* for re-gerrymandering after each census (which in Boston is an on-going process, we don't wait for Federal census to count noses!) and could be easily abolished if we opted for all city-wide seats again. Wards and Precinct boundaries are less flexible; deeds reference Wards, absolutely necessary since street names are reused within the city (due to absorbing adjacent towns); Precincts are the fundamental unit that City Council, State House, State Senate, State Governor's Council, and US House district gerrymanders are built from; but still even Ward&Precinct boundaries are adjusted periodically if a precinct suddenly is built up or industrialized.

Neighborhoods are also formally defined by city planning dept in Boston, but of which are hard to find good maps. There are a couple city employees assigned per neighborhood satellite city-hall, and aside from Election volunteers and party organizers, there are no Ward officers, so an argument can be made that in Boston Neighborhoods are more real at level 9 than Ward. They would be interesting to map as something, if not as admin_level. There are also interesting historical boundaries where former towns have become neighborhoods when amalgamated into the adjacent city (and disincorporated).


comments on Township and Range, etc are to my knowledge historically correct.

--i had a .sig when sigs were cool (talk) 18:17, 27 July 2017 (UTC)

Minor Civil Divisions, distinguished by inclusiveness

There is an important distinction that exists among the 17 states that do have Minor Civil Divisions (MCDs), whether towns or townships, and I am wondering whether this should be reflected in some way in a state's admin_level=*structure.

  • In a state like Indiana, a township is the first division underneath a county. This means that all individuals in an Indiana county live in a county and township. They may also live in an incorporated city or village. (See this example. In this scenario, especially where reverse geocoding is concerned to determine the location of a coordinate pair, it makes sense to me that we use the system we have today, which is:
    • County: admin_level=6
    • Township: admin_level=7
    • Cities, etc.: admin_level=8
  • In a state like Wisconsin, towns are a division underneath a county, but they are mutually exclusive with incorporated villages and cities. A person lives in either a town or a city/village, but never both. (See this example. What we're doing in Wisconsin so far is doing exactly as above (as would be done for Indiana), with very careful use of inner and outer relation memberships, to ensure that reverse geocoding does the Right Thing at any given coordinate set, returning one OR the other. That said, I have been wondering of late whether it is appropriate for all towns, cities, and villages in Wisconsin to exist at the same admin_level (probably 8), because their areas are mutually exclusive, and because towns do not exist "above" cities and villages in Wisconsin. They are at best on equal footing (see mutually exclusive). Is it better where towns and townships do not exist above incorporated entities, for all of them to have the same admin_level?

tl;dr: should whether towns supercede – or are mutually exclusive of – incorporated settlements, change what admin_level they should be tagged with? Skybunny (talk) 00:33, 9 June 2018 (UTC)

I'm no political scientist, but I have studied U.S. history as well as done much research on how to earnestly "blend" OSM's admin_level=* tag with the relatively complex political hierarchies we have in the USA. (Look at that: in the same sentence I used two different abbreviations for the same country!) So, Skybunny, my off-the-cuff (but reasoned) reply begins with suggesting that OSM entries in any given state (or territory) stay consistent within that state (or territory). Seems obvious, but I think important to point this out, given your question. Next, cities/villages and how they do or don't subordinate to townships is a subtle and complex topic: these really must be considered on a state-by-state basis (not case-by-case, as I believe we can stay consistent within a state). Clearly, a "true township" will always be admin_level=7, subordinate to a "true county" (or "county equivalent") at admin_level=6. What is much less clear is whether a city/village entity at admin_level=8 is or is not "mutually exclusive" with its geographically-enclosing township entity at admin_level=7, if there is one: maybe it is politically subordinate to it, maybe not. I believe an important concept here is "incorporation," that is whether the city/village at admin_level=8 is a "body corporate" and hence independent (to a certain degree) of its "parent" political entities (township, county, state and country). Yes and no, for certain and myriad reasons, and I believe all can agree to the complexities of what some of those are, and why (the 10th amendment to the US Constitution, "Home Rule" and what it means in any given state, how a "body corporate" as a municipality may be "chartered" in state law...).
It seems a large reason you are asking is to assure "correct" geocoding. While a noble goal, this seems it could be fraught with the same perils for which OSM has the tenet "don't code for the renderer." Why? Because we want the data to be correct, even if the (a, any...) renderer doesn't necessarily render them "as one might expect." Here, we want to enter data into OSM regarding admin_level=* as accurately as possible, and if geocoder results are wrong (or seem so), it is likely it is the geocoder itself which needs improvement to accommodate the complexities of USA admin_level=*. I don't know if this helps or throws a monkey wrench in the works. I welcome wider discussion. Stevea (talk) 18:13, 3 August 2018 (UTC)

If Wisconsin’s towns and cities are mutually exclusive and on equal footing in terms of scope, then it does seem appropriate to give them both the same admin_level, even if that means towns in Wisconsin will have a different admin_level than towns in another state.

Wisconsin and Indiana seem like pretty clear-cut situations compared to some other states. In Ohio, we have a situation where cities and villages can withdraw from townships, so that they are exclusive of any surrounding township. Even though withdrawn cities are in equal footing with townships in some sense, so far we’ve opted to keep them at the same admin_level as other cities because the distinction between withdrawn and subordinate cities is fairly obscure. On the other hand, in Virginia, where cities are independent of counties, cities have different admin_levels than in other states, because independent cities are well-known to be independent.

 – Minh Nguyễn (talk, contribs) 19:29, 5 August 2018 (UTC)

Any attempt to impose a strict hierarchy on my home state of New York is likely doomed to failure. It's messy. The entire state is divided into Counties - every resident lives in exactly one county. Nevertheless, there is the special case of New York City, which has annexed the whole of five counties. These Boroughs have ceded all executive and legislative jurisdiction to the city, but retain independent judiciary. (Confusingly, their names as Boroughs are different from their names as Counties: New York County=Borough of Manhattan; Kings County=Borough of Brooklyn; Richmond County=Borough of Staten Island; Bronx County=Borough of The Bronx. Only Queens County=Borough of Queens has an identical name.) For this reason, New York City gets slotted in as a sui generis entity at admin_level=5, with admin_level=6 being counties.

Counties are further subdivided into Towns (called 'townships' informally, but formally titled Town of Xxx) and chartered Cities, plus some Indian Reservations. No two of these overlap. One City (Geneva) has annexed land across a county line. Aside from that, no City or Town crosses a county line. Indian Reservations are a politically charged case. The Akwesasne Kanien'kehá:ka nation crosses not only county and state lines, but an international border. It's called the St. Regis Mohawk Reservation in the USA, and the Akwesasne Indian Reserve in Ontario and Québec. It is not recognized even as a 'dependent nation' by either Canada or the US, largely because its constitutional form of government is unfamiliar to both nations (and the inhabitants rejected a more conventional written constitution at a general election). It is a member of the Haudenosaunee Confederacy, which claims national status under the Jay Treaty, issues passports, and has from time to time fielded athletic teams in international competition. Cities and Towns are admin_level=7. Native American reservations are, for the most part, unmapped owing to the political controversies. Akwesasne, at least, arguably belongs at admin_level=3!

There are also a large number of Villages, which are mutually exclusive (but most of New York is not in any Village). A Village has only limited governmental independence of the Towns, but can enact local legislation on certain matters, elect an executive branch to administer those matters, and consitute a village court. Every inhabitant of a Village owes taxes not only to the Village but also to whatever Town the resident inhabits. (Usually, the Town collects taxes for both entities and disburses the receipts to the Village.) Most villages are in only one Town, but many cross Town lines. They slot in at admin_level=8, as do the wards, districts, precincts, etc. of Cities that have them. (Most Cities elect all their officials 'at large'.)

Some large Towns also designate Hamlets - these are not governmentally independent, but have fixed and signed borders, and the Town often enacts legislation that pertains to them specifically. While not self-governing, the fact that they are signed, and that legislation so often refers to them, means that they still ought to be mapped. They never overlap with Villages, and so can share admin_level=8 with them. The most populous Hamlets are, in fact, cities that never troubled to draft a charter and incorporate. The largest are Brentwood (population over 60,000) and Levittown (population nearly 52,000).

There are a great many varieties of special-purpose entities: school districts, library districts, police precincts, water, sewer and sanitary districts, and so on. All of these cut willy-nilly across other administrative borders. They are not hierarchic at all, neither do any of them cover the entire state. For the most part, their field observability is questionable, so they remain unmapped.

Summary: New York doesn't lend itself to division on a strict hierarchy, and will not fit any data model that demands one.

4 5 6 7 8
New York State New York City Borough (none) Community Board
(none) County Town Village*
City** Ward, district, precinct

*The City of Geneva crosses a County line; New York City subsumes five Counties.

**Many Villages cross Town lines, but Village inhabitants are also all governed by some Town. To the best of my knowledge every Village is in one and only one County.

 – Kevin Kenny (talk) 2018-08-08 01:34Z

I have incorporated Kevin's observations in his "small table" into the "big table" on the Page, as well as the appropriate footnote. (Although, I'm not sure if the current listing of "Hamlet, City of Sherrill" in the same cell is exactly correct. If somebody wants to improve this, have at it). Despite Kevin's no doubt accurate observations in these regards, in keeping with OSM's "no consensus has emerged" tone regarding Indian Reservations in the USA (and Canada), no changes were made to the "big table" or OSM data. (Thanks, Kevin — interesting and informative, all around!). Stevea (talk) 17:14, 20 August 2018 (UTC)

Recently added Connecticut COG (Regions) as 5 and CDP as 10 should be deleted

The "Not all boundaries are administrative" section describes how many years ago COGs (and their ilk) were not value=5 government entities (NYC is a unique example of an admin_level=5 entity in the USA, by long-ago and fairly wide consensus) and so I would continue to agree that the Regions in Connecticut note footnote is confusing. I don't believe Region should be in the table at value=5 because it implies that admin_level=5 should be applied to these, along with its "bound" tag boundary=administrative. We state that latter tag isn't (quite) appropriate on these (they have "limited power" and "are not a government," having "most government state (4) and local (8, 9) levels." NYC is a 5, whereas these are "not quite" (so, not). Because of this, I believe "Region (COGs)" should be removed from this intersection (Connecticut, 5) of the table. Let's clarify the footnote text, too (if we can). I am certainly open to "yes, NYC is a 5, so are these, that's just the way it is in Connecticut" because these are governments. Limited governments, yes, but governments nonetheless. However, that argument must be quite persuasive to keep Region in the table. Discuss!

The same deletion is asserted for CDP as listed in Connecticut's row for value=10: I'm OK with Neighborhood being here (to the extent a larger Connecticut city has neighborhoods, and they are truly "local government administration at a small scale"). I'm not OK with this intersection of Connecticut and 10 including CDP. CDP is a federal Department of Commerce Census Bureau designation and is statistically defined, not defined by a polygon as OSM must define such an area. Therefore, I'd like to see CDP be deleted from this entry in the table (as has been said many times and agreed to widely, "CDPs are statistical, not administrative boundaries").

I welcome further discussion about this, especially by someone from Connecticut who knows such regional and local politics / government / administration with direct personal experience and can articulate how those realities logically map to the values (and intent) of admin_level=* as we describe it in USA's political context and OSM's tagging context. Really, I'm more in listening mode than being prescriptive about "what I am certain is true" (I welcome new knowledge and fresh perspective, especially when local). However, we have had these fairly-exactly-the-same-as-before discussions (about 5 and 10, though it was boundary=census and admin_level=8, not 10 — which together don't make sense) and we seem to have explained them. So, I'd like to see these entries in the table be deleted, unless they have more firm ground to stand on. Stevea (talk) 02:03, 3 May 2020 (UTC)

As someone with local knowledge of CT all I can say is that COGs are administrative regions that are publicly funded and have their own form of government. Therefore they they absolutely belong into administrative regions. All those numbers 1..10 are just arbitrary and therefore it is unclear why NYC (and only NYC) has to occupy level 5. But as a compromise COGs could be level 5.5 and counties 6 to preserve some levels of consistency. When it comes to CDPs this becomes less of a clear cut. Many CDPs in CT were originally villages or boroughs (some maybe cities) that were consolidated into towns and their area is clearly defined. I don't think I strongly care about CDPs. Although, I still feel there is no harm to have them included under level 10 or 11. --Mashin (talk) 03:20, 4 May 2020 (UTC)
Thank you for your reply, I find it prompt and polite. I'm delighted you have local CT knowledge! "Their own form of government" is like saying "a unique form of government," yes? I think that might be a temporary meeting place, yet will eventually fall away for the same reasons as described before. (Limited government, not "all or most of the powers granted, as in Home Rule." (See Dillon's Rule knowing that this sort of political science can be complex). NYC is unique (sometimes styled sui generis in this wiki, in this context), too. We assign 5 to NYC but only did so after a lot of listening, a lot of "it doesn't fit anywhere else, and really IS a (full, not limited) government between 4 (state of New York) and 6" (the more local boroughs, or county equivalents). So NYC got 5. Townships got 7, again, correctly and by wide consensus. For a while, a lot of imported census CDP boundaries were additionally tagged with admin_level=8, but again, with wide consensus this was determined to be incorrect (in the context of the semantic described by admin_level=*) and these tags were deleted. This was all hashed out before, the footnotes and links to discussions on talk-us pages (years ago) are all in this/these wiki. Briefly to reiterate, you might assign boundary=COG (or value SPD for Special Purpose District or other three-letter acronyms as they are variously known around these united states). However, due to their limited government, COGs are not nor do they have a boundary=administrative semantic as OSM understands it. We have discussed it and have reached consensus / conclusions regarding such entities. CDPs are not "less clear cut" (maybe they are in your mind), they are even more clear-cut as "not these," (administrative boundaries) which is why many are now tagged boundary=census. I disagree with "many CDPs in CT were originally..." as CDPs are defined exclusively, by definition, by the US Department of Commerce's Census Bureau. Please do not conflate CDP with what a village might geographically define, these are two different things. There are no decimal (as in 5.5, or "let's split the difference") values for admin_level=*, only 1 (never used) through 11. Our wiki explain this, our talk-us pages (linked here) explain the rest. Please read them and present a strong(er) argument that you believe CT COGs are administrative governments. I'll immediately agree with you that they are governments of a strictly limited nature, which automatically makes then "not worthy of admin_level" tagging. OSM also (very rarely) has boundary=school because a school district boundary is a boundary, but it is not an administrative boundary which is the key-value pair (of boundary=administrative) that admin_level=* associates with. There is "harm" at putting in CDPs in at any admin_level, as it is actually incorrect to do so (and discussed many times as exactly that). Same for COGs (at 5 or any other level), though I am still listening for cogent, persuasive arguments for that. Nothing stops you from entering (what I and others consider correct) a polygon tagged boundary=COG for these: that is what they are. The numbers 1..10 are not arbitrary, their values are intended to describe a hierarchy, although it has been acknowledged many times that often this is insufficient to capture reality. (For example, indigineous people's lands both do apply and don't fit into admin_level, in a sort of "parallel sovereignty.") I said it is complicated, it is. However, I still have not heard persuasion. Being unique and a government (as is NYC) yields a value of 5. Being unique and not really a government (not as admin_level means it), no, we don't give it an admin_level=* value, any value at all. Stevea (talk) 03:47, 4 May 2020 (UTC)
The problem here is that there are only 8 counties (admin_level=6 though in practice they have few administrative functions in Connecticut now) while there were originally 15 COGs, though it's now down to 9, so each one was a little smaller than a county. It would make more sense to put them at admin_level=7, since they are smaller than the local admin_level=6 counties --Jeisenbe (talk) 21:38, 4 May 2020 (UTC)
I also appreciate your polite response (as I learnt this is not default here). You guys really like to write a lot. Though in the end this is not necessary, I somehow feel I need to reply to most of it point-by-point.
  • You tend to interpret boundary=administrative in a narrow sense of areas that need some form of government. This requirement is nowhere stated in the definition, nor was this tagging definition voted for and accepted by OSM community (same for US specific implementation). In fact other countries use it more loosely and even for statistical areas (see e.g. Slovenia). In fact every level of administration has limited government (maybe except monarchies and dictatorships). By this definition even federal government does not have absolute control so should we remove USA from admin_level? Where is the cut off and who decides?
  • admin_levels are arbitrary and set by community. No one tells us we can't use more than 1 to 11. In fact even intermediate levels where introduced if circumstances required it (e.g. this).
  • I don't know what "defined, by exclusive definition" mean, but in CT several former cities and villages were disincorporated and their boundaries turned into CDP (see former cities).
  • COGs were created after counties were cancelled and everybody realized some level of higher organization is needed (see here ). COGs are essentially reincarnation of counties, but with only limited power. Their full name is Regional Councils of Government. Here is again more info.
  • As a general reflection over this, maybe we should think about what do want to achieve with admin_levels tagging. What do we provide to the user of OSM database. Information about government structure of USA or an easy mean how to subset data based on common regions with level from big to small. I personally think that that the former can be found on wikipedia and we should provide the latter. If it was up to me, I would include as many different regions as possible that would then allow easy map rendering or GIS analysis.
  • As I said I don't really care about CDPs, but in my opinion by following the strict interpretation of admin_level tagging, you are creating more mess and less intuitive data structure. (I also don't care about NYC tagging either).
I was also considering admin_level=7, but opted for 5 since I thought it would create less mess (silly me right?). I am fine with level 7 (or any level though higher than 8 would make more sense) if some agreement can be reached here. --Mashin (talk) 02:40, 5 May 2020 (UTC)
I'll address User:Jeisenbe first, as he uses size ("since they are smaller") as a criterion for what value to assign to admin_level=*. Nowhere does OSM have anything to say anything about geographic size determining this value. I appreciate that "low numbers tend to be larger (geographically) and high numbers tend to be smaller (geographically)." But size isn't what this key denotes. What this key denotes, as is stated in the initial sentence of its wiki definition is this: "The admin_level key describes the administrative level of an object within a government hierarchy." And, "a lower level means higher in the hierarchy." As conversely, higher level means lower in the hierarchy, rather arbitrarily choosing 7 for something (COG) which was originally (and here, and years ago, by myself personally) thought to be 5 seems almost random, not really based on reason. (CDPs were once so-arbitrarily thrown into the 7 "bucket," but OSM determined this was incorrect). I think it is safe to say that not everything that appears to be a "government-like thing" deserves an admin_level=* tag. The disagreement stems from whether an object IS (or is NOT) "within a government heirarchy." COGs (and CDPs) are not. Stevea (talk) 19:00, 5 May 2020 (UTC)
Now to address Mashin's points. What is meant by CDPs being exclusively defined by the Census Bureau is exactly that. Their Wikipedia entry states (again, as its initial sentence): "A census-designated place (CDP) is a concentration of population defined by the United States Census Bureau for statistical purposes only." They are not governments. They do not fit into a government hierarchy. They MIGHT (as in the case of Connecticut, I don't really know but I'd be delighted to be enlightened) happen to overlap with what Connecticut [villages, boroughs, "some maybe cities"] "did" at one point in time, but if true, OSM prefers to have the [village's, borough's, city's] boundaries, NOT a CDP's (which are statistical only), unless added as a boundary=census.
admin_level values are anything but "arbitrary." (Being so arbitrary, almost random, is what I say User:Jeisenbe attempts by suggesting that COGs become 7). No, these values are carefully-crafted by the consensus of a country (at the state-level, too, in my experience) and its community of OSM users as to how values 2 through 11 actually map logically to real-world government entities which are in a political hierarchy. Obviously, these differ for each country. At the state level, we have an interesting case with Connecticut and its history. Neighboring Rhode Island offers "the ultimate:" no counties whatsoever for a solid counterpoint. 46 states have counties, the dominant level-6 entity in this context. Two states (Louisiana and Alaska) have "parishes" and "divisions" which both OSM and the Census Bureau are happy to call "county equivalents." Which leaves Connecticut apparently unique in all the 50 states. But this is illusory: Connecticut (like Rhode Island) does NOT have counties, it has no counties, plus it has what many other states have: COGs. And OSM (in the USA) decided years ago that COGs are not administrative. (What OSM means by things enclosed by boundary=administrative polygons is that they are "within a government hierarchy").
Mashin says I "tend to interpret boundary=administrative in a narrow sense of areas that need some form of government." Again, the initial sentence in our very definition of admin_level is "The admin_level key describes the administrative level of an object within a government hierarchy." So, my interpretation (as "needing" some form of government) is rather "ARE a form of government." So, yes it isn't merely an interpretation, it (being a form of government) is actually stated in the definition. More importantly (in my opinion of applying admin_level=*) is being true to "fits within a government hierarchy." I don't know where the "cut off" is, but I think you'd agree that a school district isn't "a government" which is why OSM (rarely) maps school districts with boundary=school. For COGs (and SPDs, MPOs, etc.), tag them boundary=COG similarly, I don't have a problem with that. I do have a problem with them being tagged boundary=administrative, largely because in 2017 we decided we don't do this. As I mentioned earlier on this page: my city water district (limited government powers over water distribution, garbage collection, sewage treatment...) isn't a government, though it does have those limited powers. So assigning its "range" as a polygon tagged boundary=administrative and admin_level=* is incorrect. Same for COGs: they aren't really governments, they are Councils Of (smaller) Governments, their powers are strictly limited. So, if you must, and you have an accurate (multi)polygon boundary, tag them accordingly (with boundary=COG). OSM strives to tag accurately what is in the real world. We have "plastic" tagging, let's use it here (we do in many other places).
A single example of a historic=yes-tagged 9.5 (Rača) is odd indeed. However, we do not speak here of historic entities, we speak of modern-day COGs, found in many states.
The word "Government" appearing in "COG" refers to the local governments in them as members, not the whole. The whole is the Council, which if it is a government, has very strictly limited powers and quite arguably does not fit into a government hierarchy (the criterion for using admin_level=*).
The examples of South Norwalk, Rockville, Willmantic and Putnam as "Former cities in Connecticut" and their present disposition of "now a CDP in the town of..." are not mutually exclusive. In other words, it is consistent (in the real world, in the Census bureau, in OSM) for all of this to be true and accurate, as long as the CDPs are not entered into OSM as boundary=administrative, but rather boundary=census. (Though if they were ONLY "former" cities, why would we do that?) One reason these might enter OSM (as a boundary=census) would be to help geocoders. But do you see the trend here? What is entered into OSM are tags of boundary=school, boundary=census, boundary=COG (where true), NOT boundary=administrative. Except boundary=administrative DOES get applied to real, full governments (not limited to landuse planning, sewerage or schools). Please read the main Page's last paragraph in the "Not all boundaries are administrative" section: it states that the Census Bureau offers a helpful-to-OSM recognition of five local government types in the US. "Three are general purpose governments: counties, townships and cities, [then there are] 'special-purpose' governments like 'special districts' (COGs, SPDs, MPOs...) and 'school districts.' The first three are tagged with admin_level=*, the latter two are (not)."
I agree with you that after Connecticut's counties were dissolved, a vacuum of power existed at this regional level. But what replaced counties (COGs) were not counties nor county-equivalents. They were and are limited-power councils of government, these limitations were deliberate. OSM must recognize differences between these (counties and COGs). By assigning admin_level=* to these COGs, we truly lose this distinction, the main reason why we shouldn't do this. COGs are not "reincarnations of counties," they are quite different from them. For example, as counties are divisions of states, it can be said they derive their power from the state, "top down" (despite USA being a "by the People" body politic). By contrast, COGs (by their very name) are councils of governments which derive their power from the "bottom up," as local governments act together as a council. So, COGs are not "county-like," not by a long distance. True, some of their activities might overlap with what counties do (landuse planning...), but as they do not have the full powers of a county, but rather very strictly enumerated powers, they are not a county. The USA has since its inception made a distinction between the federal and state governments and it largely rests with "strictly enumerated powers" (and the brief, succinct US Constitution's Tenth Amendment). COGs in Connecticut are arguably "not counties" in precisely the same ways that the federal government is different from state governments: strictly enumerated powers (for COGs) vs. broad, inclusive, not-enumerated but granted (by the Tenth Amendment) powers (for states, and, by extension, counties, depending on a state's definition of Home Rule and application of Dillon's Rule). The bottom line is that COGs don't fit into a political hierarchy (are they subordinate to 4 as 5s? no; are they the same as counties, as 6? no; are they subordinate to counties, which don't exist? no, can't be true; are they superior to 8 as 7, well, maybe, as they are a collection of 8s acting like a 7, sort of, but if COGs were somehow a 7, such a 7 would not subordinate to a 6 or 5, which don't exist, and does not even subordinate to 4, does it?, so, no, not really). Hence, admin_level=* doesn't apply to COGs, they really do not fit into a political hierarchy, a requirement for the key. If someone can persuasively suggest that COGs are both collections of 8s acting like 7 (maybe, then we're half-way) AND COGs as 7 subordinate directly to 4 (the state of Connecticut), well, I'm listening. But nobody has made that argument. It seems possible, but a longshot or a new way of thinking about or describing COGs I've never heard anybody utter. Their strictly limited powers seem to override this and the consideration I extend here. Though, I could be mistaken or not thinking about this in a certain way others might be able to better explain. Interesting I "meet back" at a possibility of 7 (both considered by the other two participants in this thread), but I reasoned out how I get there as a possibility, something I saw neither of the other participants engage in. We need others with deep knowledge of American political science to participate in this Discussion.
A serious problem with calling COGs admin_level=7 (we shouldn't) happens in states with both Townships (also 7) AND COGs. I haven't enumerated those, but there must be some. Like I said, "arguments must be persuasive." I think it fails at the start, but I listen.
I'm all for "what does OSM want to acheive with admin_level=* tagging?" being further discussed. I have been a passionate contributor to sorting this out for years and I remain in listening mode. I am relatively well-educated about local, state and federal government structure in the USA as well as OSM and its boundary=* and admin_level=* tagging. While I'm not a political scientist, I suggest we actually need one here! Stevea (talk) 20:05, 5 May 2020 (UTC)
My early brush with this was in 2012 when I thought California LAFCOs might be admin_level=5 although today I see how 7 might've been a better value (though 7 is township). Then, there are "joint powers boards" for rail transportation here in California (Peninsula Caltrain, LOSSAN, Metrolink, SANBAG, SDTI/SDMTS, the one for ACE in Stockton, I'm jumping around...) and all sorts of these for watersheds, mosquito abatement and garbage collection districts. Their complexity in the many states are myriad. I don't think OSM wants to call these governments (though they are "small," limited ones, sort of). Let's tag them boundary=* (to the extent we know this) and boundary=COG or MPO, SPD, JPB, MAD... (if we really must map mosquito abatement districts, I mean they are a real thing), with no admin_level value at all. This has worked for years. Stevea (talk) 02:05, 7 May 2020 (UTC)

"...the member municipalities of each planning region are authorized under separate state statutes to establish a formal governance structure known as a regional council of governments (RCOG)." [1]
  • Are COGs administrative units within government structure? : yes
  • Are COGs superordinate to towns? : yes
  • Are COGs subordinate to state? : yes
=> COGs belong to admin_level. Case closed.
--Mashin (talk) 04:38, 7 May 2020 (UTC)
If you assert that what I wrote was long, I agree. If you assert that what I wrote was TOO long for you to read, then I don't have time to discuss this with you, either. Especially as you say "case closed" to your own opinion after it has been the way that it is for years. We need more opinion on this rather than one person (me) who has painstakingly listened and documented over years plus another person who comes along and says "nope, that's all wet, we're done." There MIGHT be more to go here, COGs MIGHT eventually be assigned some admin_level value (is it 5? 6? 7?), but you asserting this and saying "case closed" is not consensus as OSM builds it. I remain in listening mode. Your argument is succinct, yet you do not address "limited powers." If I follow your logic blindly, now I can enter my garbage collection district (oh, it also does sewer and water) as a government. I don't do that because "I don't think so." Same with many others who have had the opportunity to add these, but thought better of doing so, so they didn't, either. Jeisenbe was here for a bit, anybody else? Because if it is simply Mashin and User:Stevea we disagree. I remain polite, if loquacious about it. Simply "case closed" -ing this seems less than polite, especially as we already agreed we don't do this years ago. Stevea (talk) 05:02, 7 May 2020 (UTC)
Please allow me to introduce a simple concept which I hope helps us describe the dimension of "limited scope government" as not really being boundary=administrative-worthy (and hence admin_level=*-worthy): "full spectrum" (as in a color spectrum) government is welcome as boundary=administrative + admin_level=*. However, "just water/sewer and garbage" is like a thin sliver of blue and a touch of indigo. One or two thin colors does not a spectrum make. State governments and federal governments (that offer soup-to-nuts governance, "all the colors in the rainbow," or almost all), yes. Those get admin_level=*. I will say, COGs and MPOs do many things (some overlap with what counties do, I suppose what some townships do or might do, I'm not sure), but my understanding is that landuse planning, and certain other limited powers are what these limited sort-of-governments do. They might be an impressive selection of orange, some blue and a bit of green, but they are not full-spectrum governments, they are limited to quite-specific tasks. Let's not start endless discussions whether this transportation-corridor agency that crosses a bunch of jurisdictions is or is not a government, because we can agree (we HAVE agreed) that BECAUSE it is ONLY limited to transportation (maybe rail, for example, or bus-now, rail-studies-for-the-future...), it ISN'T what we call a government (in the USA). We might call it a "special purpose district" (and indeed, the Census Bureau does, this is helpful to OSM, as mentioned). But an SPD (MPO, COG...even at the "medium hefty" level of 5 or so, and I'm not saying these are or aren't that) is a limited, special purpose thing (with a name, codified in statute as a thing distinct). It ACTS LIKE a government for CERTAIN and LIMITED purposes. I realize that sounds hair-splitting, but I haven't seen this exact topic addressed and it needs addressing (here, now, soon). I believe this has to do with the state's definition of [W] Home_rule_in_the_United_States and the state's application of [W] John_Forrest_Dillon#Dillon's_Rule, but that's where I run out of intellectual gas on the topic and say we need political scientists who understand what we and the greater world (of OSM) attempt to do with admin_level=* who can chime in with sharp focus and help here. Connecticut is both a Home Rule state, AND Dillon's Rule applies. Now out of gas. We need help here. Stevea (talk) 05:22, 7 May 2020 (UTC)
I have requested help at talk-us. Stevea (talk) 06:59, 7 May 2020 (UTC)
It would be easier to participate in this discussion if the posts were more concise and focused. --Jeisenbe (talk) 16:22, 7 May 2020 (UTC)
Sure, though these topics aren't exactly tweet-length. Thank you for pointing out tht concision can be beautiful. Stevea (talk) 19:29, 7 May 2020 (UTC)
I have left intact the data in OSM for Massachusetts, though I call attention in (2020-Q2's) Footnote 18 to "those eight counties" vacillations at having been not tagged with admin_level=6 in the recent past (circa 2019), but are now, deliberately by local mappers (and a link to Greg's post on talk-us). In Rhode Island and Connecticut, I reduced a lot of (admin_level=6 ways tagging as redundant in their parent (containing) relations (ways don't need to be tagged x when they as members are in a relation tagged x). These relations also deleted admin_level=6 and kept intact border_type=county (they are, indeed, borders of counties in relations of type boundary, though geographic regions named "counties," not administrative boundaries). So, they are also no longer tagged boundary=administrative, rather they are tagged boundary=region. Additionally (nice work, user:Mashin and others!) boundary=COG was set on (R)COG boundaries. There may be strays, I think I got this mostly right, though perhaps some "eraser crumbs to blow away" (minor touch ups still to go I simply didn't see). There were some maritime boundary Validator complaints I couldn't find or Resolve, I didn't mean to break maritime but maybe that needs touching up, too. Stevea (talk) 02:59, 9 May 2020 (UTC)
Look, I realize that you have spent some time working on this. Not so many people have the motivation to keep OSM data organized and it is very needed. If it wasn't you, there might have been no one to keep this in place. So I understand you feelings when someone comes along and presents their ideas.
This conversation was for me frankly quite frustrating, not only by how much unorganized text there has been produced (it seems I was not the only one who noticed), but because you choose to ignore most of the arguments given to you. You create your own arbitrary rules, that you contradict right after. And then keep asking for more "persuasive arguments".
From what I could distill from the text, the sole problem lies in some concept of "full spectrum" government. As I tried to argue, almost any level or form of government is not complete and has some limitations. That being true even for US or CT state government. There is no defined cut off or definition for this concept. Nor it is defined, agreed or been voted for in the admin_level tag description.
When it comes to by you mentioned areas, my opinion is that if they have a defined boundary then absolutely they should be part of OSM as that might be useful information for OSM data users. I don't know anything what about what they do, but from what it seems to me Connecticut water authority is just a state owned company that provides services. You pay, they give water.

What I find absolutely unacceptable are your OSM data edits that were undiscussed and not approved! This is exactly the opposite of how OSM community should work and they will have to be reverted! --Mashin (talk) 19:49, 9 May 2020 (UTC)
One at a time, in order: "presents their ideas." What Mashin (why is that red-linked?!) didn't do was "present his ideas." What he did do was change the wiki to his opinion (wholly undiscussed by him with anybody, and contradicting what was already established consensus, right here), while editing the map (with decent data, though much extraneous tagging, imho) to coincide with that (dismissed as incorrect) opinion. Rather than immediately reverting or editing them into correction as I did yesterday, first, I engaged in conversation here, to encourage the author of the incorrect wiki changes (which again, directly contradict consensus) and the incorrect editing to read what the community has discussed, decided and documented we do (for COGs). Years ago. I said that to re-open the issue, persuasive arguments must be made, that remains true. Nobody in OSM should have a problem with this so far. I do not ignore the arguments made, I respond(ed) to them. I also call for more: more persuasive, more scholarly as virtually everybody (I've spoken with: in OSM missives, private emails, the talk-us pages...) agrees with that. Nobody in OSM should have a problem with that. I do not create arbitrary rules, I resonate with established consensus, or facilitate achieving it if there remains disagreement or misunderstanding. For good or bad, admin_levels of 2, 4, 6 and 8 (roughly "full-spectrum" government, which is why there was no question whatsoever to assign these to federal, state, county and city/town) became well-established. 5 as NYC took some discussion. 7 as township took some discussion. Some states' 8, 9, 10 (especially in New England) took quite some discussion. COGs/SPDs/LAFCOs/Dog Catcher, Garbage Collection Districts... took some discussion and were decided to not be admin_level (by the community, not me, in fact, I even thought a LAFCO was admin_level=5 back in 2012, it isn't, so I can and do accept that I get these things wrong, can be corrected AND that the issues can be complex). There IS a defined cut-off for this: 2, 4, 6 and 8 are clear admin_levels. CDPs are not. Neighborhoods at 10, maybe, leaning towards yes, especially if there is a "neighborhood council." Wards, Precincts at 9 or so, I'd say "not," though if truly "voting only" I believe those get boundary=political instead of boundary=administrative, we haven't fully discussed it and much remains undone. COG-like things (at 5? 7? you haven't really said, though you did tag them in the map as 5 without one word of discussion), no. They are not admin_level value candidates, they are not admin_level-deserving (multi)polygons. This has been already decided and to reverse that decision we'll need much more than what Mashin and I have discussed here so far. Are you familiar with Home Rule? Have you said a single word about Dillon's Rule? (No, you haven't). Those really are relevant here. Addressing "if they have a defined boundary then absolutely they should be part of OSM," well, yes, I agree. And so they do, now, with boundary=COG as we discussed, decided and documented almost three summers ago. Really, this is a win for OSM: you took the time to find, upload the COG boundary data (that's what they are), you simply mis-tagged them (they are not administrative by consensus) and so I tagged them as we have agreed to. That's all. And that's GOOD. Nothing has slammed the door closed on further discussion as to whether COGs are administrative (or how much, or "how many colors of the rainbow" they might be to make a "full enough" spectrum to be called a "government that we call administrative.") In fact, I continue to welcome such discussion. But with more participants than you and me. Nobody in OSM should have a problem with that. So in the meantime, we revert to the status quo, and discuss (maybe, maybe not; patience is very thin). Nobody in OSM should have a problem with that. Reverting yesterday's edits would be very hostile to what is correct process in OSM (I have been patient and discussed this AT LENGTH!) Not to mention, there were many errors of redundancy on the tags of the objects (members of a multipolygon don't need the same tags of a polygon on the member ways, though multipolygons tagged type=boundary might be exceptional here to avoid double-, triple- and quadruple- rendering of admin_level dashed/dotted lines, where applicable). The (multi)polygons and ways might deserve some fine-tuning / improvement, though I assert they are largely correct. Addressing "OSM data edits that were undiscussed and not approved:" that isn't what happened. They were discussed and "approved" (by consensus) years ago (starting EIGHT years ago) and documented here. Apparently, Mashin didn't read this wiki (though he edited it into what he believed was correctness according to the tagging he did, neither the tagging nor the wiki entry were correct). He simply went ahead and both entered the incorrect data and changed the entry in Connecticut/5 to contradict established consensus. That isn't how OSM works. The good news is that we've had a "decent save" and the data Mashin entered are now correctly tagged. Reverting them would change these data into a "you've been told not to do this" edit, and they would be reverted to their present state. I remain open to further discussion, we need it here. Stevea (talk) 20:30, 9 May 2020 (UTC)
As an example both of how to promulgate an argument like this in better fashion (rather than "I'll revert you for removing my presently-incorrect opinions!") as well as to show I am open-minded about the topic and possibility to change, how about the following? BECAUSE Connecticut doesn't have admin_level=6, this CONTRIBUTES TOWARDS (but by no means is the sole reason) the decades-later-replacement-for-them RCOGs (and decades after that reduction in number of them) is being considered to be tagged admin_level=5 (though, maybe 6 is better, where "Regional COG" is "like" a county, but isn't one). I mean, there are no other states without counties that also have these "flavors of COGs" (with no county structure in the hosting state) and that uniqueness lends credibility to the argument that "these fill the vacuum as administrative." That's a good start, and might get people to listen/read, perhaps agree and even participate. Bolster with understandings (and application, or not) of Dillon's Rule (this is nuanced and complex, I agree, and a bit out of my league, probably is for many unless you are a legal scholar or political scientist) and other sensible points why these really are ADMINISTRATIVE rather than simply a geographical boundary of a quasi-governmental organization of smaller political units, which as a whole has statutorily STRICTLY LIMITED scope. There: now we have what we need (a good start, anyway) to continue here, the question is, will we? Stevea (talk) 22:14, 9 May 2020 (UTC)
I added boundaries for CT to their true real state into OSM and then went wiki to reflect that change. Counties and regions were in the wiki table already and I simply moved regions up admin_level not to be mixed with counties, all in good faith. Then you start dispute it (which is fine that is what we do) and you right away went on and changed someone's else data in OSM and wiki changes while they were being actively discussed. That is what most of people would call hostile behavior. So please, you are no moral position to lecture anyone not alone to threaten by forbidding them anything.
In fact CT counties existed in OSM data as admin_level=6 for 10 year (and they were along with regions in the wiki table) without giving anyone trouble that they needed to change them, until you came along. So what you did by doing such undiscussed changes to CT, RI without discussing with local communities is an act of vandalism. (And it is not only me who call you out on that.)
Persuasive arguments were made. Just read above. But you go on and on creating your own requirements (like the need to be subordinate to state) and when provided. You choose to ignore them and create another requirements (like some illusory concept of "full spectrum" government). And when those are challenged you come with another. Until where will this go? Now you came up with some laws. I don't think anyone wants to turn OSM community discussions into Supreme court hearings. Unless there is a law that forbids us to assign COGs/counties a admin_level they make little relevance to how we as OSM community interpret our tags. And also please stop using garbage collection districts, .etc as your straw man. Discussion is about CT COGs and counties.
Also can you please provide a link where a community vote as been made on CT COGs and counties before so I can look why others were so against it. It is really hard to look in those archives. On other hand I found at least 3 separate instances where other have proposed to include COGs, but you single-sidedly dismissed their requests or arguments. For now it looks more like you are trying to gate-keep other OSM users, but maybe I am wrong.
But I appreciate your start of conversation about admin_level value for CT COGs. As I said I am open to any of the options of levels 5-7, even half numbers if that is the only solution. --Mashin (talk) 15:25, 12 May 2020 (UTC)
What we have now (and as honestly as I have been able to channel long-standing consensus as I understand it) are counties in both Connecticut and Rhode Island tagged as boundary=region (and border_type=county). Especially in Rhode Island, I believe that is correct, along with a dearth of any boundary=administrative (multi)polygons which might also be tagged admin_level=6. In Connecticut, this is less clear to me, though it seems clear to Mashin (and others, he says, though we don't hear from them or on talk-us, except for User:ke9tv), as Connecticut has BOTH "regional counties" (counties which are disputed to be administrative and so instead are tagged boundary=region) AND we have RCOGs, which might be called "a limited, weak government at approximately or accurately admin_level=6." So the question is, is the suggestion that RCOGs remain tagged boundary=COG and the counties revert from boundary=region to boundary=administrative and admin_level=6? That makes a certain amount of sense to me (though I'd like others to confirm here that it does to them, too). But that's not what Mashin did (he tagged the COGs admin_level=5), it seems more like what he is proposing in light of the controversy and events that have occurred. I appreciate that he did so, "moved up COGs not to be mixed with counties," but this was done with no discussion. Can we agree to tag counties boundary=administrative and admin_level=6 along with COGs being tagged boundary=COG (and no admin_level=* tag)? Or is the insistence that both counties be tagged boundary=administrative and admin_level=6 AND that RCOGs be tagged boundary=administrative and admin_level=5? The latter really does fly in the face of what we agreed to three summers ago (COGs aren't administrative), so I'd still like to see persuasive arguments for that (the "case closed" statements above were a seed, not the whole argument) if we were to agree to start tagging COGs as administrative. One reason is that as Paul Johnson states, INCOG in Oklahoma is "toothless" and doesn't deserve administrative. Will we leave the dozens of COGs around the US up to local determination to say whether or not they deserve admin_level=5? (Which collides with NYC as "an agglomerate of county-equivalents," but I think we can concede that entities in one state at one administrative level are OK to not equate to others similarly). As I have said, I thought LAFCOs were admin_level=5 in 2012, but the community dismissed this. Maybe in Connecticut, both counties are 6 and RCOGs are 5 and I'm clueless, but I don't think so, I have long perspective on this. I DO continue to think that wider opinions (more scholarly, less "dug in") besides a small number of us here would be very helpful in this endeavor. Stevea (talk) 19:05, 12 May 2020 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── I’ve only been able to skim this discussion so far, but I just wanted to point out that OSM doesn't consider boundaries administrative primarily based on home-rule powers, legislative authority, etc. Otherwise, most subnational boundaries in OSM would have to be deleted, especially in less federalist countries, such as France and countries with a French legal heritage. So would townships in most of states that have them.

As a rule of thumb for the U.S., I think it would be useful to distinguish between regular administrative boundaries and special-purpose districts. Examples of the latter include school districts, COGs, and yes California’s LAFCOs. Some school districts and fire protection districts actually have welcome signs at their boundaries, which is a good argument for mapping those boundaries, but that doesn’t mean we’d tag them as boundary=administrative or try to fit them in the admin_level=* hierarchy.

From a data consumer standpoint, it would be inconvenient for there to be a gap in the admin_level=* hierarchy for just one or two states, but that's only problematic inasmuch as it differs from the situation on the ground. Based on my admittedly limited understanding of Connecticut geography and law, it sounds like Connecticut's RCOGs are a subtype of COG that has more authority than, say, metropolitan planning organizations (MPOs). But does that additional authority change how they're used in the real world by users of data consumers of OSM?

  • When users search for MetroCOG, do they expect to get its boundaries or the location of its headquarters as the first result? (It's the latter everywhere else I look.)
  • Do Bridgeport residents say they live "in" MetroCOG or merely that their town is a member of MetroCOG? This is important because relations are not categories.

The end-user considerations are what we should focus on when curating administrative boundaries in OSM. The legal distinctions are far less important to what is ultimately a geography project. Moreover, boundary relations may be useful for subsetting OSM data, but that doesn't mean they should be influenced by extraction needs. After all, Geofabrik divides the U.S. into five extracts along state lines and California into two extracts along county lines without there needing to be an explicit Midwest or Northern California boundary relation. My favorite Overpass queries involve geocoding 17 counties at the same time, no need for a combined statistical area boundary relation.

If we do map RCOGs as administrative boundaries, that definitely doesn't open the door to mapping COGs in other states. Taking my hometown as an arbitrary example, Loveland, Ohio, is a member of two COGs: RCOG manages taxation for over 300 municipalities and dozens of special taxation zones that dot the entire state, and OKI is an MPO whose members are spread across three states. (Two of these states have admin_level=7 townships too.) It would be a royal headache to map these COGs, and for questionable benefit. That raises an interesting question: if the motivation for including Connecticut's RCOGs in the admin_level=* hierarchy is to be consistent with other states, then wouldn't it be just as inconsistent to map only Connecticut's RCOGs as administrative boundaries and not COGs in other states?

 – Minh Nguyễn 💬 16:10, 13 May 2020 (UTC)

(I have no idea how to properly indent in these discussions). Minh has largely got it, IMO. His "it would be useful to distinguish between regular administrative boundaries and special-purpose districts" is spot-on. We can map the boundaries of COGs, school districts, fire protection districts (especially when there is a sign) WITHOUT calling them an administrative boundary (or trying to fit them in the admin_level=* hierarchy). That is a major point of agreement. Where we might disagree because of an apparent misunderstanding Minh has about MPOs and their relation to COGs: MPOs exist to meet a requirement to receive federal transportation dollars. It is my understanding that Connecticut COGs do this (effectively "encompassing the MPO purpose-of-being"), so there is no "subtype" relationship between them. I did note in talk-us (pulled from a wikipedia quote) that some COGs and MPOs in Connecticut actually share staff members.
Whether one says "I live 'in' MetroCOG" or not IS important. I doubt people say this, another reason why I don't believe that a COG is an administrative entity. We all say we live in our [2, 4, 5 in the case of NYC, 6, 7 in the case of township, 8, 10 in the case of a neighborhood council] category. Saying you "live in" a particularly-named [country, state, NYC, county, township, city, neighborhood] is another "evidence point" for "a government, something we identify as belonging to as a citizen." COGs? No. I sincerely doubt people say that.
I also agree with Minh that we might agree to tag COGs in Connecticut with boundary=administrative and admin_level=5, 6, or 7, but we disavow doing so in other states. It may very well be that not having counties (administrative counties) in Connecticut aids in reaching that conclusion. I wouldn't say this, but others might, and doing so might assuage the seething masses. I could say more, I'll stop there. And thank you for your contributions, Minh. Stevea (talk) 18:07, 13 May 2020 (UTC)
To clarify, I'm not necessarily expressing support for tagging Connecticut's RCOGs as administrative boundaries. Rather, I'm asking why we would do so, hypothetically, in the name of consistency when it would create even worse inconsistency. To me, Connecticut's counties are conceptually a better fit for the administrative boundary hierarchy, but I'm on the fence about whether they should actually be tagged that way. – Minh Nguyễn 💬 18:15, 13 May 2020 (UTC)
Right, I wouldn't express that support, either. But I still await a cogent argument for doing so, where I might expect a primary element among that to be "well, Connecticut has very little administrative/political association with its counties, save judiciary districts, which means there is a vacuum at this level of government, so COGs fill the vacuum." That IS an argument, but I believe it has weaknesses that would allow it to be shot down. So, yes, I, too am asking "why would we do so?" Stevea (talk) 18:18, 13 May 2020 (UTC)
I believe some US mappers are taking too strict a view of what can be mapped as a boundary=administrative. I've asked about this on the Tagging list to get more input, but it appears that this tag can be used for any "Subdivisions of areas/territories/jurisdictions recognized by governments for administrative purposes", especially those that fit in a hierarchy. Home rule or independent law-making power does not seem to be a requirement in most countries. So it seems reasonable to map the counties of Connecticut, Massachusetts and Rhode Island as admin_level=6, even though they have lost much of their political power, as long as they are still officially recognized by the State government and used for some administrative purpose. --Jeisenbe (talk) 21:33, 13 May 2020 (UTC)
Thank you for this valuable input. As is Minh, I am "fence sitting" as to whether these specific examples (especially Rhode Island, which really shouldn't be administrative, IMO) should be admin_level=6. However, I appreciate the perspective that "as long as they are...recognized...and used for some administrative purposes" lends legitimacy to their being tagged with admin_level=6. How about this: can we leave Connecticut COGs tagged boundary=COG and seriously consider tagging Connecticut counties from boundary=region to boundary=administrative + admin_level=6? That seems like something many can agree upon (though I'm not thrilled, I see the point: there ARE judicial boundaries set here by the state). Rhode Island, let's leave its counties alone for now, as there is simply no administration at the county level: all five Rhode Island counties are simply geographic, not administrative/political boundaries. There are eight counties in Massachusetts which (in 2019) also had their level-6 tag stripped, but these were "restored" and I am happy to shrug and say "if the locals think so, let it be." Seems like we are moving in a better direction here. (Good for us!) Stevea (talk) 21:46, 13 May 2020 (UTC)
I am glad that finally I can present my ideas on a normal level. As I said before, boundary=administrative is for tagging forms of administration and so CTs COGs definitely fit into that. In fact they were created as a replacement for counties. So both are admin_level=6. However, for the reasons of consistency and their still wide use (CT government still for some reason prefer showing their statistical data based on counties), I would think it is a good idea to keep them as admin_level=6. For COGs, I thought to include them in the less used and less cluttered admin_level=5. But if that creates problem (not sure what is so scared about that number), I would propose admin_level=6.5, which would have advantage:
  1. It would not get rendered, so less people would be irritated.
  2. This would keep counties accessible for data users.
  3. Would indicate that they both are on the same level.
As it comes to arguments how people identify them selves with some name has little relevance with administrative divisions. There are villages in CT where people identify themselves with the town the village is in. And then, are we going to do citizens polls to find what the identify themself to give a boundary some tag?
CT COGs are definitely not geographic regions like Housatonic Valley, Greater New Haven or Greater Bridgport (people actually identify with these names). They are not political boundaries - they are not a voting district. They serve to administer given region and therefore boundary=administrative.

When it comes to other regions like school districts, that is not really part of this question, and should be solved separately based on what those districts actually do. Same for COGs from other states.
btw. I found this old proposal, so maybe for those cases it could serve as inspiration --Mashin (talk) 02:13, 14 May 2020 (UTC)

@Mashin: Choosing an obscure admin_level=* value just because it doesn't render in openstreetmap-carto would be considered tagging for the renderer, which is a nonstarter. I don't think any data consumer would consider admin_level=6.5 and 6 to be at the same level.

I never suggested polling residents about their feelings about a particular RCOG. That was a thought experiment and, in my opinion, a useful rule of thumb. If RCOGs are not geographic territories, that disqualifies them as boundary relations, so I don't think we should even consider whether to tag them as boundary=administrative. Maybe it would be better to tag each MetroCOG town's Wikidata item with a member of (P463) statement set to MetroCOG (Q54954343)?

It's been argued above that data consumers find it problematic that there's a gap at level 6 in Connecticut. I'd contend that it would be even more problematic if COGs are mapped in one state only but excluded everywhere else. Besides, these data consumers would be just as annoyed that some states use level 7 for townships, whereas most states have nothing to put there. I'd hope we wouldn't dig up obscure administrative-territorial concepts for those other states just for consistency.

Please don't dwell too much on the word "administrative". This tag isn't meant to capture all the nuances of government administration, which are far too complex to squeeze into a linear hierarchy. It's probably better described as "the lines or fill colors you see on a typical political map" (as opposed to an electoral map, census map, ecclesiastical map, viticultural map, etc.).

By the way, do you have a particular use case in mind for these RCOG boundaries? I've gotten the impression that you intend to use these boundaries to subdivide OSM data into more manageable chunks. Rest assured that there are solid ways of doing so even if there are no administrative boundaries to group Connecticut towns. But if you feel strongly that Connecticut is divided into groups of towns in practical everyday usage, what do you think of tagging counties as administrative boundaries, leaving data consumers to figure out RCOG boundaries in postprocessing?

– Minh Nguyễn 💬 03:07, 14 May 2020 (UTC)

No didn't mean to say that you want to do polling. Just bring up that is is really hard to sample feelings of population and they shouldn't be deciding for us here. (I bet a lot of Scots don't feel like British, but we don't want to change boundaries there) Hmm, I guess I wanted to say geographic region? Meaning informal areas usually based on geomorphology or closeness to town. (I don't know much about wikidata. I don't know how to do what you suggested)
Not really tagging for renderer, it would just have advantage of creating less bad feelings. I agree that 6 and 6.5 is not optimal. Before, I was thinking having both as 6 and separating them by an additional tag, but that seemed to me even less intuitive. What about 6 and 6.1, where these are not ordered values, but a categories (e.g. 6.1, 6.2 ... 6.11, 6.12)? This would allow room for potentially other similar level areas. If that is properly documented in wiki then that would eliminate confusions. I don't think I dwell on administrative tag, I more tend to interpret it loosely. Some countries include statistical or cadastre regions (I suspect they want to have some form subdivision if the official is absent).
Originally, I was trying to avoid discussing other regions, but I feel your concern what to do with the other states. They could be included if they are really the same as RCOG. I don't know if there are other regions like that in CT alone. I saw MPO which is a bit harder to place to a level.
I am working on building import for WestCOG (though procrastination is strong). Subsetting addresses for merge with buildings and so on. But obviously it is not for me, but others might find for their purpose useful and having those few extra relations is not going to crash OSM database. --Mashin (talk) 06:29, 14 May 2020 (UTC)
Oh, my gosh, please. MPOs are impossible to assign a value for admin_level=* because they are absolutely not boundary=administrative, they are areas which are statistically aggregated to qualify for (or not) federal transportation dollars. And a COG in one state is almost certainly guaranteed to be absolutely different from a COG in another state. (That isn't true of counties, they are "divisions of a state" and that is agreed to by everybody. In some states that division is administrative, in at least two, Rhode Island and Connecticut, it isn't, except that in Connecticut, it IS used by the state administratively / governmentally to delineate judicial districts). And yes, I agree with Minh: it appears that tagging COGs in Connecticut by Mashin IS tagging for the renderer, I see little other reason why he wants to do this. And let's agree that it is impossible to predict what data consumers want, as that does not guide our tagging, nor should it. Stevea (talk) 06:41, 14 May 2020 (UTC)

@Mashin: It seems like you're trying too hard to shoehorn these RCOGs into the existing admin_level=* hierarchy, but administrative boundaries are not the only kind of boundaries. (Technically, the state calls these divisions "planning regions"; each RCOG serves a planning region.) Meanwhile, there is already a better candidate for tagging as administrative boundaries – counties – if only because they were designed to fit in a hierarchy in a manner consistent with other states. Can we just map the county boundaries as administrative boundaries and tag the planning regions as something else? If a data consumer really needs to work with RCOGs as boundaries, they can do that with boundary=planning even more easily than with a novel application of boundary=administrative and admin_level=*. We should take the general needs of data consumers into account, but the bar is higher than "just in case" for the gymnastics that are being proposed above.

As to the COGs in other states, please see my earlier comments on why COGs aren't generally suitable for mapping as boundaries in the first place. I'm heartened to see that we do have coverage of COGs in other states – as offices.

It's worth noting that the inclusion of any administrative boundaries below the state level is not without controversy in OSM, but we've gotten away with this exception to verifiability because of the obvious utility of certain kinds of boundaries in mapmaking for a general audience. I've yet to come across a general-audience political map of the U.S. or even of Connecticut that labels or delineates the planning regions. The only ones I've seen are specifically maps of RCOGs and their planning regions, but you can find similar maps for plenty of other kinds of boundaries that we don't tag as administrative boundaries.

Kudos for working on a building import. In case it helps, the Bridgeport boundary relation is tagged wikidata=Q49174. I've already added a "member of" statement to that Wikidata item; ideally all of Connecticut's towns would get the same treatment. If you're unsure how to contribute to Wikidata or use it in postprocessing, please swing by the #wikimedia channel in OSMUS Slack for help.

 – Minh Nguyễn 💬 08:00, 14 May 2020 (UTC)

I've asked the question already, but I will again (to put this out of its misery): what if we agree to tag counties in Connecticut from presently region=county to boundary=administrative + admin_level=6? (Let's leave alone as-is boundary=COGs, as well as present tagging in Rhode Island). Minh, again, thank you for facilitating. Stevea (talk) 08:09, 14 May 2020 (UTC)

@Stevea: Just when the discussion was so enjoyable :(
@Minh Nguyen: Hmm, maybe you're right. But isn't planning essentially an act of administration? Otherwise what is planning.
To be honest, I don't quite understand the negative sentiment towards COGs. Did something happen in US? To me it seems like we are trying to make a new administration tag just for COGs. I can feel how unsatisfying it would be if admin_level gets cluttered when all levels of administration would be added there. In that case seems to me understandable to have two tagging schemes: one for administrative levels that we like and second for those that give us weird feelings. But are there really that many and is anybody going add them into OSM any time soon?
Thanks for that. I recently started to really appreciate the connection to wikidata source, which would save us from adding lots of tags like population count or elevation. I am definitely going to look into that.
btw: we haven't made it public yet (one persons suggested adding addresses, so working on that), but if you want to have a look and feel free to give critique on tagging or anything else: Western_COG_Building_Import --Mashin (talk) 15:09, 15 May 2020 (UTC)
Martin, it would be helpful if you sign your posts here (end with four ~ characters). There is no @ before my moniker here; the "ping" directive seems unhelpful. Your sentence-fragment "Just when the discussion was so enjoyable" is nonsensical. It seems like an attempt at sarcasm, I can't be sure. Planning is one aspect of what government does. Agencies (like COGs) which LIMIT their activity to only planning or planning plus one or two other activities aren't really a government, they are more like a planning department. There is no "negative" sentiment about COGs in OSM, merely the desire to tag accurately. As COGs (and SPDs, MPOs, LAFCOs...) are not governments, we can (and do) tag them boundary=COG (or boundary=SPD...) instead of boundary=administrative, which is reserved for actual "levels of government which fit into a hierarchy." You, not the OSM community, are the only one with anything to say about "giving us weird feelings." That isn't a guiding principle for how we tag. Nor is whether anybody is going to add them to OSM anytime soon. Stevea (talk) 08:47, 15 May 2020 (UTC)

@Stevea: A little off-topic, but FYI, {{Ping}} is intended to add send the named user a notification about a reply, which comes through Special:Notifications, the inbox icon at the top of every page, or by e-mail, depending on your preferences. Some users rely on such notifications because they don't have the time to curate their watchlists. {{Ping}} can also clarify who a reply is addressed to, in a conversation involving multiple people where the indentation starts to get messy, like this one.

Also, please assume good faith in this conversation. Mashin is trying to be constructive just like you and I are. Mashin was referring to the idea of maintaining two parallel tagging schemes, one for jurisdictions that easily fit into the admin_level=* hierarchy and another for jurisdictions that don't. That distinction sounds pretty much like what you and I have been suggesting to avoid the slippery slope of special districts and planning regions as administrative boundaries.

– Minh Nguyễn 💬 20:02, 15 May 2020 (UTC)

@Mashin: Planning is an aspect of administration, one of many. Tagging planning regions as boundary=COG is already a compromise between the position that they shouldn't be mapped at all and the position that they should be considered administrative boundaries. Some mappers have taken the same compromise with census-designated places because of outcry when CDPs were getting deleted. Planning boundaries probably form a hierarchy of their own; after all, many cities have their own planning areas that may not necessarily line up with where neighborhood boundaries are delineated on the ground. So probably boundary=planning planning=council_of_governments would be more flexible. But at least avoiding the boundary=administrative tag lowers the stakes by avoiding any mistaken treatments by data consumers, so the broader community doesn't need to worry so much about 100% coverage or accuracy. Data consumers would still be free to make use of boundary=planning boundaries, just as they can with CDPs (to the extent that anyone is still maintaining them). – Minh Nguyễn 💬 20:02, 15 May 2020 (UTC)
Hmm, it seems we are in deed spinning in circles. On one side me, Jeisenbe (though I can't say what his preference for COGs are) and large portion of world OSM community tend to interpret the in a loose sense and on other hand you guys (for some reason) prefer some form of stricter categorization. Personally, I don't think that the discussion is about about whether they can be in OSM, as they clearly have known and defined boundaries that are relatively stable and verifiable from the documents. So nothing really prevents us from adding them to the database other than the consideration whether they are useful for the downstream users.
Though I agree that boundary=COG is not systematic if we were to use something it would have to be a bit expanded. When I looked at that old region-tagging proposal, I also thought of something similar to yours. Though I thought that having boundary=administrative, administrative=COG would indicate that this region is part of administrative system, just not the kind that people like to put admin_level=* on. It would also use the existing system rather than creating a new tag that maybe no other country would ever use. --Mashin (talk) 01:00, 20 May 2020 (UTC)
For better or worse, too many prominent data consumers already interpret all boundary=administrative relations as the kind that you put admin_level=* on. It's far too late to redefine that tag away from its de facto usage. – Minh Nguyễn 💬 18:11, 21 May 2020 (UTC)

@Jeisenbe: I would tend to agree with that blanket statement, except that I don't think special-purpose districts should be mapped as administrative boundaries, because they're designed to be overlays over the usual hierarchical administrative units. (In some contexts, they're literally called "overlays".) Also, unlike Mashin, I remain wary of tagging the RCOGs as administrative boundaries, even though they serve an administrative function and can be associated with a geographic area, because they're more strongly associated with their office locations.

I would've been strongly in favor of tagging Connecticut counties as administrative boundaries from the outset, but what gave me pause is that I had only heard of them being used for divisions of individual state agencies rather than anything at a lower level of government, and some state agencies are using RCOG service areas instead. In that sense, the counties are more like the divisions of each state's transportation or health department or state police, which we probably wouldn't map as any kind of boundary. (Why not? Because then San Francisco might have a few dozen extra boundary relations around it, even though such districts would be easier to manage in postprocessing or mashups.)

However, maybe I'm missing something because I don't have local context. If Connecticut counties are still widely used outside a clerical context, and it isn't just a coincidence that the judicial system's divisions correspond to county lines, then the county lines have cartographic value as administrative boundaries. In that case, I would favor tagging them as such; the fact that they double as judicial district boundaries becomes less important in my opinion. It goes back to that question of whether a resident understands that they live "in" a county.

I wonder if, a generation from now, mapmakers would come to a different conclusion about these counties. Virginia's magisterial districts were originally townships until the abolishment of townships in 1874. After some time, it seems, people stopped thinking of the magisterial districts as vestiges of the township system and instead saw them as divisions similar to the state police divisions I mentioned earlier.

 – Minh Nguyễn 💬 03:07, 14 May 2020 (UTC)

After reading the talk-us thread on this topic, I think there's a much better argument for tagging counties as administrative boundaries than tagging RCOG planning regions as administrative boundaries. Apparently counties do still figure in public consciousness as more than a way to divide up a government agency's org chart. [2] Moreover, [3] makes the point that we could look to technical standards for guidance. The relevant technical standard in the U.S. used to be FIPS 6-4, which has been superseded by INCITS 31:2009. I don't have access to the published standard, but the Census Bureau still shows INCITS-based codes for Connecticut's counties. [4] All these observations point in the direction of retagging the Connecticut counties as boundary=administrative admin_level=6. – Minh Nguyễn 💬 20:02, 15 May 2020 (UTC)
I'm down with that. (Meaning "I agree"). If there are no objections, I'll tag Connecticut's counties from boundary=region to boundary=administrative + admin_level=6. Not so much because I think what the Census Bureau has to say about things should (heavily) influence what OSM does (there are agreements, there are slight disagreements, we do a fair-to-good job of noticing these and denoting them), but more because there are "greater than zero" administrative functions at this level: I refer to judiciary boundaries. I think it helpful to indicate in the wiki that this is at least part of the reason that OSM tags like this in Connecticut. As well as that Connecticut follows in New England's propensity of states having "weak county / strong town" forms of government (largely at levels 4 and 8 or 9, as is already mentioned in the wiki). I also think it is important to include in that exact same context that Connecticut has seen fit to evolve over about sixty years a collection of RCOGs (the local moniker) which perform limited planning functions that have been found during those sixty years to be agreeably and efficiently performed by RCOGs (legislation freely describes them as able to continue to evolve). So, we tag these boundary=COG as we might (or do) in other states — though because of statutory creation and definition of scope, a COG in one state is certainly quite different than a COG in another state. This "a COG in one state isn't like a COG in another state" reasoning is one more buttress in the argument that COGs becoming administrative is a messy nightmare for OSM to manage. Let's call a COG a COG (or an SPD an SPD...) and leave it there. This is tagging accurately, it allows both COGs to evolve (state by state) AND it allows OSM to continue discussion about how we might "sharpen up even more" the tagging on these. (Which for now, seems sufficient). Peace, Mashin? Minh, again, thank you very much for facilitating here (and educating me on ping-in-wiki). Stevea (talk) 20:22, 15 May 2020 (UTC)
The change is complete. Stevea (talk) 20:30, 15 May 2020 (UTC)