[Talk-us] TIGER 2022 PLACE dataset

[Zeke Farwell]

On Thu, Jan 19, 2023 at 11:06 PM stevea <steveaOSM at softworkers.com> wrote:

> I will also say that it sounds like you are "right on track" with your
> small-t town as a place node AND a boundary=census.
>

To be clear, I am not advocating for this.  I am in fact questioning the
benefit of boundary=census as a thing that exists in OSM.  I'm not aware of
a data consumer that uses this tag.  If anyone is, please let us know.  The
boundary=census objects I've come across have all been dual tagged as
place=town|village|hamlet and there has also been a node with all the same
tags.  I can see the benefit of mapping a place as an area and a node, but
there needs to be a clear story for how data consumers are expected to
deduplicate.  If I were to map a town, village, or hamlet as an area, I
would probably look at the latest CDP polygon from the census.  If it
looked good I might leave it alone, if it didn't actually seem like a good
representation of the built up (urbanized) area, I would modify it.  I've
seen some CDPs that seem to very closely match the built up area, and
others that seem far too small or far too large.

In summary: Defining the area of a populated place in OSM may provide
value, but the relationship between nodes and areas representing the same
thing needs to be well established and supported by data consumers.  CDPs
can provide a good starting point for defining the area of a populated
place, but I don't see the value  in strictly adhering to census defined
areas or tagging them as such in OSM*.

Zeke

* At least so far anyway!  I'm happy to be convinced otherwise.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20230120/b95f10f8/attachment.htm>