Proposal to demote Indiana municipalities to admin_level=9

Communities United States
,
1

I’m considering retagging every town and city boundary in Indiana, replacing admin_level=8 with admin_level=9. The boundaries already have border_type=town or city to establish a soft correspondence with municipalities in other states. Indianapolis would be the only boundary left at level 8:

  1. Indiana (same as other states)
  2. —
  3. Counties (same as counties in other states)
  4. Townships (same as townships in other states)
  5. Indianapolis
  6. Other cities and towns

This change is necessary because of the complex mess that is Unigov: Indianapolis is in nine townships (level 7), and in turn 10½ towns (level 8) are in Indianapolis (the “included towns”). Under Indiana law, towns and cities are peers regardless of county, even in Marion County. With this change, we’ll finally be able to draw the correct Indianapolis city limits. Currently, we depict the included towns as enclaves of Indianapolis, as if they’re all excluded towns.

The downside of this change is that some border towns will compare unintuitively. For example, the city of Union City, Indiana, is much larger than the adjacent village of Union City, Ohio, but the Indiana city would end up with a numerically higher admin_level=* tag than the Ohio village. Arguably both boundary relations should share the same place point, following the Texarkana model, but this approach won’t work for some affected border towns that don’t share a name, such as West College Corner, Indiana, and College Corner, Ohio.

admin_level=9 enjoys somewhat less software support than 8, but major global renderers do support it. Like some less mature styles, OSM Americana only goes up to level 8, so it would need a tweak to add level 9. This would bring in neighborhood boundaries in many cities, so it might be necessary to distinguish Indiana’s level 9s from other level 9s on the server side in OpenMapTiles.

2 Likes
Incorrect Indianapolis city limits
2

I’m confused. When Indianpolis is made up of townships, then it clearly should be above level 7. Shouldn’t it?

3

Technically, in Indiana and several other states, there are two parallel administrative hierarchies overlaid on each other:

  • State
    • County
      • Township
    • Municipality (in Indiana: city or town)

Nothing stops a municipality from overlapping multiple townships or even multiple counties. Across the country, hundreds of municipalities cross county lines, but this doesn’t mean they sit above counties in any hierarchy. Unfortunately, this notion of parallel administration is incompatible with OSM’s administrative boundary tagging scheme, so we’ve flattened it into a single hierarchy per state for convenience:

  • State
    • County
      • Township
        • Municipality

This is plausible to the public because most municipalities are smaller in area than townships. Indeed, it’s pretty common to describe a town like West Harrison as being “in Harrison Township”. Indianapolis is one of several large cities in Indiana that span many townships. In neighboring Ohio, a large city would typically withdraw from the townships it overlaps, punching a hole in them or abolishing them completely, but Indiana has no such provision. Regardless of size, every town or city is a peer of every other town or city in the state – except for the uniquely consolidated city of Indianapolis.

1 Like
All the Ohio townships
4

Interesting. I didn’t realize the there were two separate but overlapping administrative boundary systems in some states. I wonder if municipalities could have varying admin levels depending on size. If townships were consistently level 8 (instead of the current 7), then large municipalities spanning multiple townships could be level 7 while smaller municipalities spatially within a single township could be level 9. Just like the current flattened hierarchy it wouldn’t really be a correct model of reality, but it might be a bit more flexible. There is already some precedent for large municipalities having a lower admin level than smaller ones with consolidated city-counties being level 6 and the multi-county New York City being level 5.

New York City and its multiple counties seems somewhat similar to Indianapolis and it’s multiple townships (albeit at a different level). The NYC counties retain a level of independence and serve different administrative purpose from the City of New York. It sounds like the the townships of Indianapolis are similarly not subordinate to Indianapolis despite being spatially within it.

5

In Indiana, there’s nothing special about a municipality that spans multiple townships or counties. It’s still a municipality like any other. Indianapolis almost completely overlaps Center Township, but it would be incorrect to say that Center Township is “in” Indianapolis, other than spatially.

Besides, the number of overlapping townships has more to do with the town’s location than its size. Lots of small municipalities also span multiple townships or counties just by happenstance. Aurora is in three townships despite a population so small that we should probably classify its place point as place=village. Conversely, Carmel has 29 times the population but lies in only one township.

Every consolidation is a unique situation. New York City followed a different model, retaining a borough within and coextensive with each of the five counties.

2 Likes
6

“Within spatially” is what the word “in” means to me in this context, so that statement seems correct to me[1]. But if there is some other connotation that it raises in your mind I’m sure I will understand how it seems incorrect to you.

I should have been more specific. By “municipalities spanning multiple townships” I really meant municipalities fully containing multiple townships. This would not include small municipalities on the edges of multiple townships and overlapping a portion of each. Maybe only Indianapolis would qualify.

Anyway I don’t have strong feelings on the matter. It would just seem a little odd to have most of a state’s sub-county admin boundaries in admin level 7 and 9 while just one lives in 8. In most states level 8 is widespread. Given the parallel administration schemes it doesn’t seem unreasonable for either townships or municipalities to be admin level 8 as they are both typically the next most local level after county. But if those more local than me insist this doesn’t work, I’m not going to argue.

  1. Despite enclaves making it a bit less than 100% technically correct in a very specific sense. ↩︎

7

The administrative hierarchy isn’t purely spatial. There are reporting structures too. We try not to get in the business of mapping government org charts, but it matters that the township government answers only to the county and that the municipal government also answers to the county in some matters. Otherwise, why would we bother tagging admin_level=* at all if the hierarchy can be computed by ST_Contains?

South Bend covers almost all of Portage Township, but this doesn’t make the city’s relationship with Portage Township any different than its relationship with Penn Township, which it overlaps far less. It’s essentially a dotted-line relationship, both types of jurisdictions responsible for different areas of responsibility. Even so, a layperson would expect to see a location description like “Washington High School, South Bend, Portage Township, St. Joseph County, Indiana”, not “Washington High School, Portage Township, South Bend, St. Joseph County, Indiana”. Likewise, “Marion County, Indianapolis” would be very strange.

I don’t think we can ignore the fact that level 7 has been used successfully for the vast majority of minor county divisions in the country for decades, despite the prevalence of municipalities that cross jurisdictional lines. This is just one of those things a mapper from overseas occasionally comes by to sound the alarm about, only to be told that it’s perfectly normal and nothing to get alarmed about.

In a different universe, the counties and townships would be a different kind of boundary=* than the municipalities, with independent admin_level=* scales, as we’re doing with reservations and their subdivisions in other parts of the country. After all, in some parts of the country, people primarily use school districts as geographic references within a county. But who can say that counties are more essential to political geography than municipalities or vice versa?

Indianapolis fully overlaps Washington Township if we count the included towns as part of Indianapolis (contrary to how they’re currently mapped). Indianapolis doesn’t fully overlap any other township. It comes close with Center Township, but part of Beech Grove, an excluded city, also overlaps that township.

The city of Anderson does fully overlap Anderson Township (while extending into other townships).

8

A great many OSM contributors as well as my experience talking to people in the real world about their opinions on the structure of “counties” (in all 50 states, including Louisiana’s parishes and Alaska’s divisions) indicates there is a STRONG identification with the belief that “counties are the de facto and de jure subdivisions of states.”

This remains true even in cases / places where it isn’t strictly true, such as the subtle distinctions about counties “versus” RCOGs in Connecticut, perhaps a handful of counties in Massachusetts and maybe some oddities about counties in Rhode Island. OSM seems to have “settled” upon some consensus tagging and wiki-documentation about these edges, and peace now reigns over the land (map). Minh has been instrumental, even crucial, in being peacemaker in many circumstances here; many thanks!

Somehow, the identification of “counties” (in all 50 states) as state political subdivisions is so embedded into the psyche of “US political geography” that even where these are called something else or don’t exist in the “traditional sense” (they might in some limited sense, like “for judicial boundaries only”), many (most?) of us — including the US Census Bureau, which has a tendency to make EVERYTHING a county, whether it is or isn’t one — believe all 50 states should “have” counties (or county equivalents). And they should be — or SEEM LIKE they should be — 100% “space filling” (green shaded) even when they are not (as in Alaska, purple shaded).

It IS complicated out there. Documenting complications goes a long way (or even the whole distance) to solving “it’s complicated.” Oh, and let’s additionally “tag well” to the best of our ability: BOTH are important for us to complete.

9

Or someone like me from the northeastern United States :grinning_face_with_smiling_eyes:. In my area of the country, level 8 is the norm for county subdivisions. Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut, New Jersey, and Pennsylvania all use admin level 8. New York is the odd one out subdividing its counties with level 7 instead. If level 7 represents the vast majority you’ll have to forgive me for getting a different idea based on my surroundings. Or maybe that is a bit of an overstatement.

Looking west I see that Ohio, Iowa, North Dakota, and South Dakota all use level 7 county subdivisions heavily while Michigan, Wisconsin, and Minnesota all use level 8. There is some use of level 7 for county subdivisions in a number of other states but coverage is quite spotty. Indiana appears to be one of these with subdivisions only mapped in a few counties.

I expect a layperson would find both of these location descriptions rather odd for listing four administrative entities when normally just two are expected (city, state). Wouldn’t “Washington High School, South Bend, Indiana” be what most people would expect?

10

I feel like everyone on this thread should watch my talk from SotM US in Boston last year where I talked about exactly this problem and how the admin_level structure works across the US:

Of course, if we change Indiana, my talk will be out of date :grimacing:

4 Likes
11

Ohio’s mappers just finished mapping all the state’s townships a couple days ago. It was quite a feat! Over the years, there were multiple proposals for automated imports of external datasets, but we could never trust those datasets because of the Byzantine system of paper townships. We had to do it the hard way, by hand, consulting a variety of sources.

Fortunately, things are easier in Indiana, but just. There are no paper townships, no withdrawals. Township lines rarely change. We can map townships using nothing but the USGS topographic map layer and TIGER as a finding aid. However, it’s still more complicated than in Michigan and Wisconsin, where “strong” townships are equivalent to and mutually exclusive of municipalities. New York is actually a closer analogue. At least there a municipality can be nestled inside a minor county division (town).

The TIGER boundary import omitted minor county divisions by mistake, or because the authors didn’t know about them or what to make of them. Otherwise, Indiana would be completely covered in admin_level=7 like its neighbor to the east. As it stands, coverage of manually mapped townships is merely a reflection of the local community’s size rather than any argument for or against level 7. Help wanted!

I intentionally called it a “location description” – that is, the kind of thing that Nominatim outputs, as opposed to an address. If the POI is qualified by just city and state, the user would reasonably expect it to be an address, so the city would actually be a postal city. But if they see something more longwinded, they’d expect it to reflect the political geography more precisely: no postal city nonsense, and an acknowledgement of counties if not also townships. After all, nothing prevents two communities from sharing the same name. Indiana has more cases of that than probably any other state.

The townships are well-known as a local unit of government, whether in an rural or urban area. @ZeLonewolf’s StreetFerret is emblematic of the kind of use case that needs this political geography as opposed to addresses. No one with the misfortune of living outside any city or town would want to have to run all the streets in unincorporated Indiana. However, in an incorporated area, the municipality is the more relevant local geography, even if it covers a bigger area than a township.

The combination of the municipality and township can help to pinpoint a location, but it is unusual to combine them in the same breath. That’s what we have to work with in the single-threaded model of boundary=administrative admin_level=*. It’s just a fact of life that the municipality comes before the township and county in any such description.

12

I’ll take your word for it, but you have described them as parallel admin schemes. If that truly is the case it would suggest that the order doesn’t actually matter between municipality and township. Anyway I don’t have local knowledge of Indiana so I’ll leave it at that!

13

First I’m hearing of “parallel admin schemes” though that wouldn’t surprise me.

Minh, I’m not sure what you mean by “the municipality comes before the township and county in any such description.” Or, maybe I’m misunderstanding: you say “in an incorporated area” just previous. (Meaning your perspective starts at 8…but I’m now starting to parse “parallel admin schemes”).

There is a set, it includes these three: the municipality, the township, the county. We say 8, 7, 6 (usually, about those, with those values). The ordering is just that. “One’s perspective” looks at 8, 7, 6, another perfectly valid perspective looks at 6, 7, 8. That’s about the size of it: a set ordered like that, you could start at one end and go up, you could start at the other and go down. Sometimes, it makes sense to be looking at a smaller set of only two of these, like 8, 7 or 7, 6, but the same things about ordering them is true.

Which “such description?” admin_level is single-threaded without trying to be anything besides what it is.

As we work with admin_level, let us not forget that “everything is its own case.” It can be hard to learn. At admin_level=4, for example, we have our own set of fifty nesting dolls and several “other things” like Territories and a District (wholly different than those fifty). We do well both discussing and documenting, like we do here and in our wiki (plural). I’m guessing that Brian’s identified “life is more difficult here” states can still use some mapper love, that’s OK and fertile ground for good growth ahead. Continuing to tease these apart as we have experienced the growth (in understanding and documenting our admin_level structure in many, even most states) is rather tasty in the strength of our data, if you ask me. Green lights ahead!

I think we’re doing well so far (especially as I wiki-link and re-watch Brian’s patient presentation). Though Indiana (with Unigov, et al) is a chewy one, 'tis true. Discuss, document, ta da!

(As the eggs crack open, we cook in a bit of butter and olive oil and they become delicious…this is more or less what we’ve been doing in the realm of US_admin_level for well over a decade). Finite lists and even complexities grow, yet can be followed: that’s what matters. As it continues to grow, it continues to work (as it is understood to do so, because we generally wiki-document it). Onward to solving Indiana! (As its own thing, because it is and then we document it so all of us know). Agreeing that a “full blanketing of the state of Indiana by admin_level=7 polygons” is good growth. As they are a thing that we map in our world, let’s map them in our map.

14

The operative word is “technically”. Technically there are parallel hierarchies. Municipalities are established in IC Title 36, Articles 4 and 5, municipalities in Article 6. The Census Bureau takes pains to keep these branches separate. But in practice, if someone must describe a spot using a single phrase that includes both the municipality and the township, the municipality comes first. Call it a convention, I guess, but it has a pretty obvious translation to our boundary tagging scheme: a higher number in admin_level=* than for townships.

15

So I think the scheme of Indy at 8 and the subordinate towns within Indy at 9 is probably something we can all nod our heads yes at. The real question for me, is whether we can boil this down to which of the two are true:

  1. The towns within Indy are just like all the towns that surround Indy, and Indy is a special case administrative creation.
  2. Indy is just like the towns that surround Indy, and the subordinate towns are like districts or boroughs of a city.

From @Minh_Nguyen’s descriptions, I’m hearing that #1 is the correct mental model and therefore the proposed demotion of the whole municipality layer down to 9 makes sense to me.

The discussion about setting the townships at level 7 seems like a red herring – they’re mapped at level 7 everywhere in the US where townships are a thing in the map. That doesn’t seem controversial.

1 Like
16

It was at least a tangent. Sorry for being difficult :sweat_smile:. It wasn’t an argument against moving municipalities to level 9 though. Moving townships from level 7 to 8 could only happen if municipalities also move from 8 to 9.

In Vermont we have Villages within Towns which are essentially mini municipalities within other municipalities. Residents of a Village deal with two overlapping local governments. These are mapped as admin level 9 within level 8. So the proposal to have level 9 towns within a level 8 Indianapolis makes sense to me. The rest of Indiana having level 9 municipalities and nothing at level 8 would be different from every other state that has some kind of municipality at level 8. But it would be good thing to have more level 9 out there to force better support from data consumers.

Anyway, with Indiana needing three levels below county, municipalities at level 9 seems like the only viable option.

17

Yes, (1) is correct. To be clear, I would greatly prefer to limit the fallout to just Marion County if we can. We could alternatively introduce a level 7.5 just for Indianapolis. Indy is an oddball already, why not say so. But I just don’t have enough confidence that the software ecosystem is ready for such an innovation.

(2) would become feasible in January once Cumberland becomes an excluded city. Unfortunately I still don’t think it would be an accurate reflection of reality. The remaining included towns are puny and some of them contract with Indianapolis for things like fire protection, but they are still towns in their own right. They aren’t comparable to the neighborhood council areas that we map as administrative boundaries in many cities.

On that note, some mappers started adding level 9 and 10 boundary ways in parts of Indianapolis to represent some neighborhoods. They aren’t administrative either, since the Minigov plan never got off the ground. I’m thinking of converting the Near Eastside area to a boundary=place place=quarter relation. Meanwhile, smaller areas like Emerson Heights are already redundant to named landuse=residential areas, so I’d probably just delete them.

In that case, I’m surprised Vermont didn’t follow New York’s example, promoting the space-filling layer of towns and cities to level 7 so that the sparse layer of villages could be level 8. Indiana doesn’t have that option because Indianapolis coexists with the townships.

I did wonder if the municipalities should be level 9 or 10, since the even numbers are intended for the “major” administrative area types. But if Vermont is already using 9 for municipalities of a sort, then Indiana might as well too.

18

That’s my concern also. My software would be able to handle a 7.5 just fine, but at present there are no uses of admin_level in taginfo that have non-integer values. If there is any data consumer out there treating this value as an integer, it could break on handling the boundary of a major American city.

Puerto Rico already has their municipios somewhat arbitrarily at admin_level=9 and that seems to work OK.

19

And I’m surprised New York didn’t follow Vermont’s example so the space-filling layer of towns and cities could be level 8 like every other state and province in the region. It all just depends what we’re used to I guess. Anyway, enough about Vermont and New York in a topic about Indiana.

20

I think the New York choice was made more or less arbitrarily by @ke9tv when he mapped those boundaries in the first place. We could certainly reconsider it.