Usuari:Pere prlpz/The center of the Earth according to Wikipedia
The centre of the Earth is in Morocco. It's about 2400 below a point in Khouribga Province, but that is just the centre of the Earth according to English language Wikipedia. According to German language Wikipedia, it's 1200 km under a point in western Sardinia, according to Japanese language Wikipedia it lies on the Shaanxi-Henan border (China), and according to Russian language Wikipedia it's in Bulgaria. How do we find the centre of the Earth? Just like we find the centre (or centroid) of anything else: we find where is every point on it, we compute its coordinates and we average them for all points. And what do we use to find where are the points on Earth? Wikipedia, but since articles with coordinates are unevenly distributed across the Earth's surface and the articles with coordinates are different for every edition of Wikipedia, by computing the centre of the geolocated articles from different Wikipedias we get different points according to their diverse content.
Results modifica
Here is a list with the centroids of geolocated articles from several Wikipedias. There is a location map below, but more detailed maps can be gotten by clicking in the "OSM", "Google Maps" or "Bing" links at right.
.svg)
Depth of centre is related to how spread are points from each Wikipedia. An edition of Wikipedia with articles evenly spread over the globe would yield a very deep centre that would fit the geometric centre of the Earth. A similar centre could be produced by a Wikipedia with articles clustered in a pair of areas placed in each other's antipodes.
Although deeper research would be needed to fully explain every centre, in general the centre from each Wikipedia can be explained from its content in three areas where most Wikipedias have a lot of geolocated articles:
- The area where the language of each Wikipedia is spoken.
- Europa, specifically Western Europa.
- North America and specifically the United States.
The remaining continental areas influence seems to be smaller.
This way, most Wikipedias produce centres arround the Mediterranean Sea (most of them Western Mediterranean), with some notable exceptions:
- Wikipedias in non-European languages: The big distance from the central (Mediterranean) zones and their linguistic areas produces a centre in some point between both and at great depth. Notably, South Eastern Asia languages and Afrikaans have centres moved to their areas. Spanish and Portuguese Wikipedia can be included in this group, since their American linguistic areas carries the centrer to the Atlantic Ocean.
- Botopedias: Wikipedias with hundreds of thousands of bot-made geographical stubs produce centres far from their linguistic area but close to the place where stubs are. An extreme case is Serbo-croatian Wikipedia with its mid Atlantic centre -the westernmost of all centres- because one third of geolocated articles in shwiki are Mexican places. Volapük Wikipedia, the oldest botopedia, shows a similar Atlantic centre.
Only a few Wikipedias yield a center in its own area: only Arabic, Occitan and Aragonese Wikipedias, probably because of a combination of local content and a location that makes the effect of other geolocated articles cancell. Not surprisingly, Arabic, Occitant and Aragonese languages are spoken in the Western Mediterranean areas where the centers of most Wikipedias lie.
The center of all Wikidata items with coordinates is in the Mediterranean Sea in front of the Algerian coast, among the centers of most Wikipedias but less deep than most of them. That's not far away from where it would be expected, but it would be interesting to see where are located the items that are in Wikidata and not in any Wikipedia.
Finding the centers modifica
The program used to get the data, do the math and format the table and the location map is Usuari:PereBot/robot centre viquipèdies.
Data modifica
Data for each Wikipedia are coordinates of all items in Wikidata with links to that Wikipedia. Items are filtered out using astronomical body property to only include items without this property and items with it filled with value Earth - when data were retrieved very few Earthly items had this property.
Items with coordinates in Wikidata aren't exactly the same items that show coordinates in their articles in Wikipedias, and when there are coordinates in Wikipedias and Wikidata they aren't always the same. Although, given that some bots upload coordinades from Wikipedias to Wikidata and that usually coordenates in Wikidata don't bear more errors than in Wikipedia, differences are expected to be small and aggregate measures -like centers or counts- of both sets of coordinates should be very similar.
Other works on geolocation in Wikipedias have used all coordinates appearing in articles, retrieved from template {{coord}} and the links it produces. The most notable difference with data used here is that we are using just one point for each article (and item) while other works used more than one point when the article had more than one geolocated point. Basically, this supposes that we are ignoring the points from lists with coordinates (example).
The math modifica
Wikidata stores geographical coordinates (latitude and longitude) for each item. They are used to compute Cartesian coordinates (x,y,z) in a reference frame with origin in the center of the Earth, which is taken as a sphere with radius equal to one. Cartesian coordinates for all points are averaged to find the center and its coordinates are again converted to geographical ones (latitude, longitude and radius).
This way of finding (and defining) the centroid of the geolocated articles in a given Wikipedia has three consequences:
- The definition of the center of a set of points depends on the definition of distance between two points. Here the Euclidian distance is used - that is, the distance along an straight line across the Earth. Had been used a distance measured along a great circle -the distance on the surface of the Earth- slightly different centers had been obtained.
- The Earth is approximated by a sphere. Since the biggest difference between mean radius and real local radius is about 15 km, the error caused by this approximation can be expected to be even smaller - I would guess a lot smaller.
- Height above (or under) sea level has not been used because this piece of information is seldom stored in Wikidata. However, most of geolocated items are placed just a few hundred meters above sea level so the error caused by ignoring height is expected to be a lot smaller than one kilometer.
See also modifica
- List_of_Wikipedias, in Meta.