[Tagging] Different approaches by other projects

[Minh Nguyen]

Vào lúc 20:20 2021-11-17, Mateusz Konieczny via Tagging đã viết:
> Nov 17, 2021, 22:08 by tomasstraupis at gmail.com:
> 
>     And at the same time we have:
>     * governments opening their data at a very fast pace, and their data
>     is homogenous - stable and very usable (curated by professionals with
>     a clear knowledge on how and where the data is used)
> 
> 
> Maybe it is true for some specific datasets or specific countries.
> 
> If you work with government data and ended with this impression
> you either are lucky or have unusually competent government
> in this area.

Tomas's enthusiasm for open government datasets is admirable, but 
comparing OSM to them is comparing apples and oranges. OSM has a 
different, complementary role to play.

Those of us based in the U.S. are quite familiar with public domain 
government datasets. They have their pluses and minuses. Homogeneity and 
comprehensiveness is a huge plus, but the data comes with the caveat of 
being fragmented into a hundred thousand gratuitously incompatible 
datasets. On the other hand, the national datasets that aggregate local 
datasets are notorious for their unevenness: names like TIGER, GNIS, and 
NHD will elicit groans from many on this list. Nonetheless they remain 
an important resource for us.

The public sector maintains robust data on certain topics but nowhere 
near the full breadth of topics that OSM ideally accommodates. In fact, 
the U.S. Geological Survey recently gutted the federal government's 
canonical database of POIs, removing every airport, church, hospital, 
park, etc. -- everything but the names of populated places. The removed 
POI types hadn't been actively maintained since 2014 due to a lack of 
funding. [1] I think we can legitimately claim to be the ones who've 
been maintaining this GNIS data instead. [2]

Vào lúc 13:08 2021-11-17, Tomas Straupis đã viết:
>    So make no mistake. Community is important, but if that means
> allowing anybody to destroy the quality of the data with no control -
> it will not last long. There are other emerging means of collecting at
> least general/georeferencing/base data (road/water network, landcover,
> buildings, places etc.) which are way more consistent, verifiable,
> stable and homogenic - therefore much more usable - than OSM.

By all accounts, OSM tagging is chaotic, but a lot of that comes from 
being open-minded about what's worth mapping. We don't have the luxury 
of getting all the answers fed to us by an authority. To the extent that 
the alternative methods you mention find success, they could become 
tantalizing opportunities for imports into OSM.

Sure, some of the chaos is our doing, in having multiple toothless 
quasi-authorities between the wiki, wiki proposals, editor presets, peer 
pressure in chat, and sheer elbow grease in the database. But at least 
we can align some of these power centers through good-faith discussions, 
bold wiki cleanups, and well-planned retagging efforts. By contrast, a 
focus on control against impending destruction is just so distant from 
my own experience as a mapper, documentarian, and data consumer that I 
don't see how it would result in any meaningful improvements.

[1] 
https://www.usgs.gov/core-science-systems/ngp/board-on-geographic-names/domestic-names
[2] The USGS also has a well-designed, crowdsourced National Map Corps 
program that's intended to fill in gaps in GNIS, but the 14 feature 
types it collects pale in comparison to OSM's scope and ambition: 
https://navigator.er.usgs.gov/help/WebHelp/structure_def_table.pdf

-- 
minh at nguyen.cincinnati.oh.us