I just came face to face with Wikipedia when consulting about America in English Wikipedia and Spanish Wikipedia, Questionable differences where the only thing that was made clear is that there are two different points of view derived from different educational systems.
If I ask an English speaker about America, they tell me that they are two continents where Central America is a minor subregion of North America, which is equivalent to saying that Panama is part of North America.
If I ask a Spanish speaker about America, they tell me that it is a single continent subdivided into three regions where Panama is part of Central America.
What is the correct concept of America as a continent?
Geologically, America can be divided into North and South America at the border between Panama and Colombia. The Mexico-USA border is a cultural border: south of it is Latin America. Central America is the Latin American part of North America.
I just came face to face with Wikipedia when consulting about America in English Wikipedia and Spanish Wikipedia, Questionable differences where the only thing that was made clear is that there are two different points of view derived from different educational systems.
@dieterdreist was joking, of course, but thatâs a really neat encapsulation of the problems that we see with wikidata in OSM in lots of other areas, too.
Do you consider that because Wikipedia in Spanish says something different than Wikipedia in English, Wikipedia in Spanish is wrong?
As indicated above, they are different educational systems where children in the USA are taught that America is two continents and Latin American children are taught that America is a single continent with 3 subregions (South, Center/Caribbean and North).
The point I was making was that different languagesâ wikipedia articles for the same physical thing can have different nuance, and that can have implications for OSM.
Another example is Kosovo - the geographical extent of Serbia is very different in the maps on the wikipedia pages in Serbian and Albanian, for obvious reasons. OSM has not just names for countries but actual boundary relations.
To take âNorth Americaâ as an example, the wikidata has various weasel words in various languages https://m.wikidata.org/wiki/Q49 as to whether it is a continent, a subcontinent, or a bit of both.
At a very different geographical size, âvillagesâ and âcivil parishesâ near me have one wikipedia page, and hence one wikidata entry (since here** that was just created from wikipedia). It doesnât map well onto OSM because OSM has them as separate objects (sometimes there are multiple villages per civil parish), which means that wikidata links for OSM objects can be a bit woolly.
We could similarly discuss whether Europe and Asia are actually one single continent. I donât think weâll find objective answers to these questions though.
That is true, but in OSM we should take a neutral position, and Felipe is showing us a clear example of biased data from one perspective.
As part of this discussion, I propose that we follow the same reasoning for the place= tag as for admin_level=1 and prevent having place=continent.
The nodes describing continents have been prone to vandalism in the past: https://www.openstreetmap.org/node/36966063/history
This implies an unnecessary effort for the DWG to keep these nodes clean for inaccurate data.
As part of this discussion, I propose that we follow the same reasoning for the place= tag as for admin_level=1 and prevent having place=continent.
âŠ
If the definition of a continent is unclear, OSM should not reflect a different thing and should be neutral in that.
as weâre on it, with this reasoning we should probably also remove the place=village/town/hamlet objects.
I can see that for a continent, as there are so few of them, leaving them out is the most practical solution (indeed in the first global map I created from OpenStreetMap data I already decided to have the positions for these in a separate, curated file rather than using OpenStreetMap for it), but this is not a general solution for problems of this kind.
Tharâs a little unfair. Those nodes were created 16 years ago, and there is no evidence of bias in that change. OSM is forced to take a view on how many continents there are - it cannot have 5 and a half nodes. If that does not fit anyoneâs world view they can change the nodes in whatever they use to consume OSM data.
If it helps to document why OSM has 6 continents, then the OSM wiki may be the place to do that
it could have several versions of continents, according to different world views, similar to how disputed borders could be mapped.
As I wrote above, IMHO the continent topic is not worth it, but for other situations it could be useful to have a mechanism to deal with such situations.
Just for a second, take a step back, and think about the infrastructure changes that would be need to do thatâŠ
Itâs not an entirely silly idea - Natural Earth has the concept of âpoint of viewâ datasets (which seem to be âcultural themesâ now). The OSMF disputed territories policy declares it as an aim that people consuming OSM data can postprocess it to fit their needs.
It would be entirely possible (and relatively simple, as these things go) for someone to create a version of Nominatim (or other OSM tools) that have different base data for continents, just as people can create maps that show culturally different boundaries. What is not feasible is to expact all OSM tools (including the osm.org website) to magically understand all potential cultural differences at the same time.
There were suggestions elsewhere to âraise an issue with Nominatimâ - that is at best unhelpful. If someone wants to change the results they see in one instance of that they are entirely free to run their own instance. Complaining about the implementation at osm.org is a bit like saying âthank you for the free pony, but can I have it in a different colour please?â
Edited to add this bit:
As a DWG member I see the emails from people complaining about the geographical representation of things in OSM. I also see (via the list of Natural Earth issues) the complaints about their point-of-view data. Iâm not convinced that they get fewer complaints than us; in fact I suspect that they actually get more âinternet advocatesâ complaining that X âis not a valid point of viewâ than we do, perhaps because their issues list is public.
I think the point is that Wikidata can better reflect the complexity of this topic than OSM, because it doesnât have to rigidly insist on a single answer for something that varies along linguistic or cultural lines with no right answer, even âon the groundâ:
Whereas Wikidataâs data model has the capability to express that something is a continent according to one country or language or author and not according to another.
Nevertheless, I donât see the practical problem in mapping all the notions of continent across major languages and fields of study. Itâs very much like other natural features like place=ocean or natural=mountain_range that can have overlapping definitions, or place=region for that matter.
So what if we acknowledge a Southern Ocean that is more widely accepted in hydrography than in other fields? Technically, we also take a view on how many place=country nodes there are in the world, but we donât expect someone to get a perfect answer to âHow many countries are there?â by counting the occurrences of this tag.
It would only be able to âreflect the complexityâ if it stayed as an abstract concept and not interacted with the real world. As soon as you try and map an abstract concept like that to a real-world object in OSM, that breaks - wherever there is geographcal vaguesness, such as the examples I gave above where wikidata entries were created willy-nilly from wikipedia articles, that will always be the case.
Because it is rooted in the real world, OSM doesnât have the option of sitting in an ivory tower contemplating its navel; it has to âpick a sideâ and say that there are either 5 (or 6) continents, and explain why - with the caveat that we try and make it easy for people to postprocess the data for their own use so that their representation has 6 (or 5).
Meanwhile, we have Oceania as a place=continent. This may be suitable from a cultural and pedagogical standpoint, but from a geological standpoint, itâs merely a region that includes the continent of Australia (unmapped), which is not to be confused with the mainland alone.