Tagging counties and planning regions in Connecticut

Should the admin_level=6 administrative boundaries in Connecticut represent traditional counties, or should they represent council of government planning regions?

Decades ago, Connecticut abolished county government without replacement, but everyone continued to divide the state into the counties for general geographic purposes. Lately, however, the state has been formalizing its COGs into something seemingly more powerful than a typical COG in another state, but less powerful than a typical county in another state. COGs in other states function more like consortia or NGOs, whereas Connecticut has assigned fixed territories to each COG – the planning regions.

Upon Connecticut’s recommendation, the federal government recently redesignated the planning regions as the state’s county-equivalent political subdivision, replacing counties, at least for demographic and funding purposes. The state is now calling counties “historic” but has continued to use counties for some purposes, including public health reports.

Figure 1. Relationship between Counties and Planning Regions. County boundaries are shown as thick lines; names are labeled in standard font in all caps. The planning regions are shaded; names are in italics.

Over the past few years, there have been several involved discussions about whether to retag all of Connecticut’s counties as non-administrative boundaries and map the COGs or their planning regions as administrative boundaries instead. In the database, there has also been some back and forth about how to tag both sets of boundaries, involving multiple mappers.

Currently, counties such as Litchfield County are tagged as:

name=Litchfield County

whereas planning regions such as the Northwest Hills Planning Region are tagged as:

name=Northwest Hills Planning Region
official_name=Northwest Hills Council of Governments

Despite the federal government’s newfound appreciation for planning regions, I really wonder whether Connecticuters tend to use them day-to-day as an intermediate-level geography when describing a place. “Lower Connecticut River Valley Planning Region” is quite a mouthful compared to “Middlesex County”, and I have an even harder time believing anyone would shorten it to “RiverCOG” as a place name.

As a starting point, I would suggest the following hopefully uncontroversial changes:

  • If the counties are no longer administrative but remain in use informally or marginally, then they should be boundary=historic. boundary=political is reserved for electoral boundaries, but these boundaries ceased to have any electoral function a long time ago.
  • The names of the COG organizations should go in operator=*, not official_name=* and short_name=*. I’ve also created a parallel set of Wikidata items such as Northwest Hills Council of Governments (Q124655976) that can go in operator:wikidata=*.
  • The name=* should omit “Planning Region”; the fully qualified name can go in official_name=*. This would be consistent with how agencies such as the Census Bureau are presenting the name.
  • Continuing the separation between these organizations and their territories, border_type=* should be planning_region, not COG.
1 Like

Good summary of the administrative quirks of Connecticut here, Minh, thank you. I met the geography (TIGER) team from the Census Bureau at the Esri Federal GIS conference a few weeks ago and we spoke about this very issue.

The counties are not fully without meaning, despite their COG counterpart’s federal recognition. The team mentioned some court districts and sherrif’s offices are still oriented towards the counties.

I too would be curious about the on-the-ground impact of this change and how people interpret or use those in their daily life. I have some relatives there that I’ll ask and report back.

Agree about dropping “Planning Region” from the name. New York City is officially, The City of New York, yet we name it as simply New York on the map.

1 Like

As a selfish data consumer, I would want the next level up administrative boundary to have a name that makes sense when I’m showing it to a user in hierarchical fashion, for example:


In this case, I think the “Planning Region” is actually a more logical set of breadcrumbs to the user than if we had:

Pomfret, Northeastern Connecticut, Connecticut, United States

We don’t drop the word “County” from counties, why would we drop “Planning Region” from planning regions if that’s how they’re called?

As an actual data consumer, I completely ignore what’s in OSM and walk up the hierarchy using wikidata solely, because I can rely on data property linkages rather than geospatial queries. So, I don’t care too much what gets set in the name tag on these things other than it would be nice to have a consensus and avoid another “Californians grapple with New England quirks” go-around :wink:

In the interest of giving an example of “how data consumers actually use this type of data”, I offer this screen shot from my own application:

The US is complicated and there’s a ton of weird exceptions and corner cases, but there’s a strong need for a “primary” enclosing administrative entity with a “primary” name. Without the “Planning Region”, the names of these things seem quite ambiguous. Do you mean that specific planning region or is this just some general concept of “Northeastern Connecticut”? The “Planning Region” suffix makes it clear what we’re talking about.

Now, my users would probably much rather see “Windham County” rather than this planning region nonsense as it’s more customarily familiar, but there isn’t really a way to do that in a generic fashion without special-casing the quirks of every place on earth…

Regarding the names, here’s table 1 of the official document

1 Like

You could just as similarly dig up a table that has “County” in the header row and “New London”, “Tolland”, “Windham”, etc, in each row and it wouldn’t make us any less likely to name them “New London County” etc.

For a bit of background, Historic counties also exist in England: Historic counties of England - Wikipedia although they seem to be a bit of a different beast from those in Connecticut.

Currently they are mapped as boundary=traditional + designation=historic_county, for example: Relation: ‪Yorkshire‬ (‪8484939‬) | OpenStreetMap. (Not that I’m advocating for tagging these historical boundaries as such, just putting it out there)

I was approaching this from the perspective of whether OSM is going out on a limb, sort of geeking out about a detail that most laypeople would not recognize. Nothing against geeking out, but there’s no reason an interesting detail needs to use the usual tags. Unlike “County”, the “Planning Region” makes it sound a lot like a special-purpose district, so I’m not as confident as you that it avoids surprising the user, but if Connecticut residents have taken to this unlikely terminology, then let’s find out rather than speculating.

In 2020, a Connecticut resident opined that residents still generally use counties as the intermediate geographic subdivisions in that state. Of course, this is just a sample size of one, and it could just be inertia. Apparently it takes a lot for New Englanders to part with colonial-era boundary lines. If we look at the contexts in which someone in another state would locate something by county, some of these use cases are handled by towns in Connecticut, while some others seem to be handled by either county or planning region. Most static maps published by the state only show town boundaries, while equivalent interactive maps show counties from an ArcGIS base layer. Some maps published since the establishment of planning regions purposely show towns by county.

The actual federal regulatory action that recognized the planning regions, and the accompanying state FAQ, make clear that the change only pertains to demographic statistics. This is not nothing – statistics come with funding – but counties are also being used for limited purposes. The state government itself continued to report COVID case counts by county throughout the pandemic. If the state is subdivided into neither counties nor planning regions in general, but it was historically subdivided into counties like the rest of New England in general, then I wonder (but cannot say for sure) if the balance tips in favor of counties.

I agree that there are multiple forms of any place name, depending on context. For this reason, I suggested keeping the “Planning Region” form in official_name=*. But if people are likely to use the full form by default, similar to counties and townships, then the current name=* tags would be fine. I just assumed it was less likely because the names are so verbose. Maybe it turns out that people mostly use neither name because they don’t use the planning regions at all.

For its part, the Census Bureau always spells out “Planning Region”, but then again, they always qualify every place name with its type, pro forma. For example: “Tolland town, Capitol Planning Region, Connecticut”. “Tolland town” does have a cheesy ring to it, but so far it looks like that isn’t what people call the place in general.

I think that would be more accurate than the current boundary=political (which is for electoral districts), but boundary=historic seems to be better established globally.

I will argue the “historic” counties remain more relevant for mapping purposes than the COGs in spite of what the census bureau is doing, because while the census bureau may normally be considered an authority on things like this the reality is that among any official source it is the exception, not the rule - they switched to using COGs at the state’s request because this is a more useful statistical aggregation for funding allotment, but that’s really the only reason why and there is no broader effort to deprecate the counties in favor of COGs.

Look at all sorts of legal documents that call for a county name - e.g., property deeds - and you will see the names of the 8 counties used, not the 9 COGs, and there is no intention of changing this.

The National Weather Service is also still using county names, not COG names, in alerts. For example: NWS Alerts

1 Like

NWS maps also depict county boundaries. But their geocoder splits the difference, returning both “Hartford, CT, USA (Capitol)” and “Hartford, CT, USA (Hartford County)”. Weather maps by commercial media outlets almost invariably show counties, Louisiana parishes, and Alaska boroughs, and sometimes census areas in Alaska’s Unorganized Borough, but apparently never Connecticut planning regions.

As far as I can tell, every broadcast TV station in Connecticut depicts county boundaries on its radar, satellite, and traffic maps, both in broadcasts (WFSB, WTNH, WVIT, WTIC) and in OSM-based interactive maps (WFSB, WTNH, WVIT, WTIC). Similarly, The Weather Channel provides forecasts by county rather than planning region, and their interactive maps only depict county boundaries.

(All these interactive maps use the OSM-based Mapbox Streets vector tiles in conjunction with the proprietary Mapbox Boundaries dataset.)

Granted, one can probably distinguish between weather reporting and administration, but together with public health, the legal system, property records, and more, it does seem like demographics is the odd one out here. The one other map I’ve found that prioritizes planning regions is The National Map, which incorporates the National Boundary Dataset. (Indeed, it spells out “Planning Region” in labels.)

Does anyone have a strong opinion that Connecticut should be primarily divided into planning regions as admin_level=6 administrative boundaries? I didn’t see any discussion at the time the change to favor planning regions was made. Other discussions I’ve seen have always ended in either a stalemate or a consensus that counties should remain the primary boundaries.

Strictly speaking admin_level is about area administration, which makes the planning regions more suitable, but I am not against using the old counties instead if that is what people prefer.

I also agree with all points of the proposed changes except that now I am also leaning more towards keeping ‘planning region’ as part of the name tag.

It’s almost like we need a tag besides boundary=administrative for the primary subdivisions of a region that happen to be non-administrative but still equivalent for mapmaking purposes. :persevere:

1 Like

I’ve implemented the following changes to the nine planning region boundary relations:

  • border_type=COG planning_region
  • official_name operator=*
  • short_name operator:short=*
  • operator:wikidata=*

and the following change to the eight county boundary relations:

  • boundary=political historic

Otherwise, for now, I’ve left the name=* and boundary=administrative tags alone on the planning region boundary relations.