Wikipedia talk:WikiProject Geographical coordinates

RFC: importing coordinates from other language editions
There are over 20,000 articles missing coordinates that have coordinates on Wikidata.

While I don't intend to import coordinates from Wikidata, as I don't believe it is well-curated at the moment, this is suggestive that a substantial number of articles will have coordinates on non-English Wikipedias, many of which have a sufficiently large user-base that they can be considered to be effectively curated by human beings.

I propose that we start importing coordinates from the following Wikipedias, all of which have more than 1 million articles, and have not had articles mass-created by LSJbot:


 * German 2,828,000+ articles
 * French 2,546,000+ articles
 * Russian 1,933,000+ articles
 * Spanish 1,886,000+ articles
 * Italian 1,823,000+ articles
 * Japanese: 1,384,000+ articles
 * Chinese 1,373,000+ articles
 * Portuguese 1,107,000+ articles

As part of this exercise. I plan first to generate lists of candidate coordinates from these that they can be assessed for quality before going any further.

Yes, I know that in theory, Wikidata is the way to do this kind of cross-wiki syncing. But in practice, there's as yet no real cross-wiki integration to enable human curation, and thus Wikidata is not ready to be useful as a way to do this yet. And of course, if I do the interwiki copying, I intend to export any coordinates I find to Wikidata as part of the process, if they're not already present, so that as and when Wikidata is eventually usable for this purpose, we can easily migrate to using Wikidata as the master data source.

I'd welcome any comments other editors have on this. In particular, please let me know now if you have any objections to this plan; I don't want to waste my time implementing this only to for the addition of these coordinates to be rejected by the community. &mdash; The Anome (talk) 17:25, 1 September 2023 (UTC)


 * I'd be interested to hear your comments on this. &mdash; The Anome (talk) 09:35, 2 September 2023 (UTC)
 * Though I think your concerns about Wikidata are unfounded, I support this import. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 18:12, 2 September 2023 (UTC)


 * I've been thinking about this since I saw that you had posted it. Unlike Pigsonthewing, I share your doubt of Wikidata, and I've certainly run across incorrect coordinates in other-language WPs as well. I imagine that a great number of geocoded articles in other languages are already geocoded here, so I'm not sure how many sets of new coordinates you're going to come up with; but I'm willing to hold my judgment. (When I'm looking for coordinates to add to one of our articles, one of the things I usually do is check to see if they're present in a corresponding article in the relevant-language WP, and I can't say that I often find any that way.) Your plan "to generate lists of candidate coordinates from these [so] that they can be assessed for quality" seems a good first step, and I'd like to see how that turns out. Will you post the results somewhere so that others can inspect them? Deor (talk) 18:34, 2 September 2023 (UTC)


 * I fairly often correct bad coords in WD; some of our large imports are indeed dirty. Probably coords that are mass-imported from any Wikipedia will be no dirtier than the average import; might even be cleaner. I'm comfortable with the notion that a load of coords with median errors of 10 or 50 meters are better than nothing. As it happens, a few minutes ago I corrected an error of 50 meters in my own coords for an urban photo, so no use my complaining about other moderately dirty suppliers. You might want to get into touch with whoever thinks they handle coords in WikiData; they might have something important to say, like, nevermind importing WPDE or other WP language coords into WPEN; just put it in WD and use a template to show it in English. Jim.henderson (talk) 01:14, 5 September 2023 (UTC)

Kartographer, infoboxes and Wikidata
I've noticed that some templates that use the Kartographer extension are pulling in coordinates from Wikidata by default. This is bad news, as it represents a slide toward treating Wikidata as the master source of geographic data without there being a robust system of editorial oversight; the coordinates are nowhere to be found in the body of the article, and there is no clear way to detect in edit logs that Wikidata coordinates have changed, or simple user interface for correcting them if they're wrong. This seems to me to break the entire wiki principle that drives Wikipedia.

What to do about this? I'd be happy with using Wikidata coordinates in these if these problems could be fixed, but given the glacial rate of change in the Wikimedia software, I wouldn't hold out much hope of seeing this change. At least with mass-copying Wikidata to here, there's some chance of editorial oversight; this is worse that a mass import, because it's invisible. &mdash; The Anome (talk) 15:47, 5 November 2023 (UTC)
 * Could you give one or two examples of "templates that use the Kartographer extension"? (With my limited technical skills, I'm not sure what this means.) I've noticed that a few infoboxes, such as Infobox power station are set up to pull coordinates from Wikidata by default; when I run across such, I've been leaving them alone if the coords look good but adding local coordinates if they need to be emended or otherwise adjusted (overprecise, a bit mislocated, etc.). They can be identified whenever an article is displaying coords but doesn't have a template in it. As you know by now, I don't think that indiscriminately pulling coordinates from Wikidata is a good idea, but I often feel like Dame Partington trying to push back the Atlantic with a mop. Deor (talk) 16:47, 5 November 2023 (UTC)
 * I've mentioned your concerns in WD chat. I think the people there are happy with their quality control in this area. https://www.wikidata.org/wiki/Wikidata:Project_chat#Pushback_at_using_WD_coordinates_in_enwiki Vicarage (talk) 16:49, 5 November 2023 (UTC)
 * Information imported into infoboxes from Wikidata must have a reliable source in Wikidata, per the 2018 Wikidata infobox RFC. If there are infoboxes that do not comply with that RFC, their code should be adjusted. – Jonesey95 (talk) 16:03, 6 November 2023 (UTC)
 * Thanks for pointing out the RfC, which I was completely unaware of. If there's a requirement for a reliable source citation for the Wikidata coordinates, then I'm fine with it, and this is really good news. Do the templates check for the presence of a reference for the wikidata value? Is there some way in which this check can be done automatically and transparently as part of the value retrieval process? &mdash; The Anome (talk) 08:24, 11 December 2023 (UTC)
 * Infobox bridge provides an example of how to retrieve only sourced values and allow local overrides in articles. See official_name in that template, for example. – Jonesey95 (talk) 13:55, 11 December 2023 (UTC)
 * Fantastic. The syntax is a bit eye-watering, but I can see what's going on. This makes me feel a lot better. &mdash; The Anome (talk) 14:26, 11 December 2023 (UTC)
 * I'm pretty happy with the theory of Wikidata as the master source of geographic data. As with figures for land area, population, and other geographical numbers, this method has the advantage that the data can be entered once for all languages. The same WD coords go into various maps including WikiShootMe and Wikmedia Commons Android App, both of which I use to find photographic targets, including those that are mentioned in articles but have no article of their own. This activity finds many erroneous coords that were imported on a mass basis from other databases. I make a rough WD coord adjustment in the field and finer upon getting to the big computer. No worry about how many languages are using that particular WD item.
 * One problem is, many editors don't know about Wikidata, and the newbies whom I teach find the very idea difficult. I mean, even the infobox itself is difficult to learn, since the Visual Editor is not very helpful. I hope this can be fixed by designing a more direct and obvious way to edit WD from the infobox, and the infobox by VE. Alas, this is likely to take years but when it happens, we can expect faster cleanup of those dirty mass imports. Jim.henderson (talk) 01:25, 8 February 2024 (UTC)

Feedback requested for innovation made
Hello. There is a discussion that might be of your interest and it would be great if you can provide your feedback at Wikipedia talk:WikiProject Maps. Sincerely, Thinker78  (talk) 01:14, 16 March 2024 (UTC)

Improving geodata accuracy on OSM and Wikidata
I have captured my experience about improving geodata accuracy on OSM and Wikidata and shared it through a short OSM user diary and also a long form article on OSM wiki. Let me know your feedback or your questions. Arjunaraoc (talk) 11:04, 10 April 2024 (UTC)

Global Entity Reference System
I see that the Overture Maps Foundation (https://overturemaps.org/) have created a new identifier, the Global Entity Reference System. GERS identifiers are opaque 128-bit handles intended to identify unique entities at any level within the hierarchy of a geographic data system, but don't seem to be UUIDs. (See https://docs.overturemaps.org/guides/gers/). There seems to be some structure within them to be used to ease allocation (and perhaps as a sanity check), but that structure is not supposed to be used as data by end users. (See https://github.com/OvertureMaps/schema/discussions/27)

They also define more structured metadata types, such as places: https://docs.overturemaps.org/schema/reference/places/place/

We should probably follow this iniatiative and work out whether we should link our data with theirs. &mdash; The Anome (talk) 10:17, 17 April 2024 (UTC)

"coord restricted"?
I suggest a new coord family tag, coord restricted. This would indicate that the article describes somewhere with known coordinates, but that those coordinates are restricted from public view with good reasons, and Wikipedians should not attempt to add coordinates to those articles. The primary use would be for archeological sites, whose locations are restricted to thwart vandals and grave robbers. &mdash; The Anome (talk) 09:46, 4 May 2024 (UTC)
 * Is that consistent with the purpose of Wikipedia?. Wikipedia is not censored: we are here to supply information. One productive way (that came from the discussion at Talk:Hyperion (tree)) is to supply approximate coordinates, and highlight the fact that the coordinates are approximate. Barring coordinates entirely from an article seems excessive. — hike395 (talk) 18:56, 4 May 2024 (UTC)
 * An approximation (e.g., d° m') followed by a hidden comment would suffice. If an editor disregards the hidden comment, they would likely disregard this proposed tag. &#8213; Mandruss  &#9742;  19:06, 4 May 2024 (UTC)


 * Re U.S. locations: I've seen cases in which an NRHP-listed house's location was restricted at the owner's request but our article gave the house's street address, and I've also seen ones in which a historic district's location was restricted but the NHRP nomination form gave the district's exact boundaries. Figuring that the cat was out of the bag, so to speak, in those cases, I went ahead and added coordinates. I've also seen cases in which the NHRP restricts an archaeological site's location and I declined to add coordinates, even though I knew where the place is. Re archaeological sites elsewhere: If the site is labeled on Google Maps or OpenStreetMap, I think it would be overscrupulous to avoid giving coordinates in our article. I think your suggested tag might be a good idea, especially to replace tags in articles about places whose "coordinates are restricted from public view with good reasons", but I fear that it might be applied indiscriminately. I also think that approximate coordinates could be useful in some cases, as hike395 and Mandruss have suggested. Deor (talk) 22:08, 4 May 2024 (UTC)


 * I agree with that Wikipedia is WP:NOTCENSORED, but we do not, for example, routinely publish people's home addresses, because there are sensible reasons not to do so. Approximate locations may be the right way to go in some circumstances; I did this recently for an archaeological site, where the exact location is restricted, but it is publicly described as being near a particular locality, by geolocating the article with that of the locality. For other cases, there might not be any locality to use in this way, and perhaps the use of very approximate coordinates might be a good idea for these cases. In any case, if we do have that tag, it should not be applied too widely, and just kept for edge cases like sensitive archological sites whose location is not already well-known. &mdash; The Anome (talk) 11:08, 5 May 2024 (UTC)