On Thu, Feb 21, 2019 at 3:46 AM Mark Wagner <
[hidden email]> wrote:
> When you did your query for hamlets, I'm afraid you ran headlong into a
> quirk of American political geography. Historically, the postal service
> would only deliver mail to buildings within a certain distance of a
> post office, while people further away would be responsible for
> visiting the post office to pick their mail up. As a result, it was
> quite common for a group of farmers or ranchers to get together and
> have themselves declared a community in order to get a post office.
Pardon my ignorance, do those hamlets typically correspond to OSM's
description (100-200 inhabitants), in contrast with other possible
values (place=locality for no inhabitants, place=isolated_dwelling for
less than 3 households)? I'm seeing from Bing imagery that Osborne
Corner has several households, as does Nille Corner. They are close to
the generic threshold for being considered isolated dwellings, but
still pass. I'm not familiar with the exact details of how place=* is
assigned in the US. In Brazil we still use the "generic" rules for
place=* (even though I've tried pushing the adoption of our national
Geography Institute's criteria).
In any case, I could try to find another, more populated pair of
place=hamlet and the generic idea would still apply more or less.
Going over a large number of cases (I would suggest choosing a
manageable large random sample), some sort of general verifiable rule
would emerge. But more importantly, the general rule is only a
guideline to aid in choosing such a general rule. Perhaps the main
local roads connecting those communities have some sort of property
(physical quality, official classification, maintainer) that would
bring out a system that is closely related to the idea of "connections
between hamlets".
For comparison, in Brazil most hamlets (some of which look a lot like
what you just described) would be connected by unpaved municipal
roads, excluding highways that connect larger settlements. Thus,
unpaved municipal roads are highway=unclassified unless something else
would push a particular road's class up (some large industry, a port,
or the main route to a village/subprefecture, etc.).
> There are thousands of these "paper communities" scattered across the
> country, and they don't exist to any degree beyond the minimum
> necessary to make someone else responsible for delivering the mail.
> Many of them don't even exist to that extent any more, and are merely
> names on the map.
So, are they really hamlets or usually just uninhabited localities?
Nonetheless, place=locality would typically also be connected by
public highway=unclassified routes (but I expect exceptions in
restricted areas, such as in national parks).
> (It's also somewhat misleading to say that Del Rio Road and Y Road
> connect Nilles Corner to Osborne Corner. Rather, they were built
> to connect the farms in the area to the Grand Coulee/Coulee Dam/Electric
> City area, incidentally providing access between Nilles Corner and
> Osborne Corner.)
In most countries I've seen unclassified roads providing access to the
entrances of farm properties.
In previous debates, I've agreed that such places (work places, ports,
airports, etc. in this case a large dam which is a workplace for many)
can indeed work like place=*, that is, they are traffic generators.
But in the case of those roads, they would need something equally
important at the other end to justify raising their class in OSM I
think.
So, using this area as an example, what would be a more sensible
highway classification for you? I don't think it is correct (based on
the original intention) to classify roads that have only a few houses
spread between farms as highway=residential. The wiki says (and I
agree) that residential streets typically have lower speed limits and
sometimes traffic calming devices, designed to ensure the safety of
dwellers. As such, highway=residential typically shows up in more
dense urban areas, even small ones, but not over large expanses of
farms.
Also, to me, the nearby roads like X Road Northeast [1] seem closer in
purpose to highway=track as their main function is to provide
agricultural vehicle access to cropland [2], but perhaps not in terms
of access restrictions. Tracks are usually considered access=no, but
their default access type varies by country (from "yes" in the
Netherlands to "no" in Denmark to "destination" in Germany) and the US
surely could opt for whatever default access value makes most sense
there. [3]
[1]
https://wiki.openstreetmap.org/wiki/OSM_tags_for_routing/Access-Restrictions#United_States_of_America[2]
https://wiki.openstreetmap.org/wiki/Tag:highway%3Dtrack[3]
http://openstreetmap.org/way/5875745--
Fernando Trebien
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging