[Tagging] part_of:wikidata key

[Christoph Hormann]

On Wednesday 29 November 2017, Marc Gemis wrote:
>
> I assume you mean verifiable without accessing the Wikidata website.
> I often hear contradicting information about "verifiable". Some
> people say it's only "on the ground", others find it OK to ask
> locals. It would be nice to know what you understand under
> "verifiable".

Verifiability ultimately has relatively little to with the sources used.  
On the contrary being independent of specific sources is a key aspect 
of it.

For physically observable things this is pretty easy because it often 
boils down to if you can measure something - either directly as a 
quantity (like a geometry or the iconic height tag of a building) or 
using statistics (an area can qualify as natural=wood if it features a 
certain density of trees).  You can usually measure with better 
reliability and more accurately when you are locally on the ground so 
in cases of doubt the observation on the ground stands above remote 
assessment.

For things like a name it is more complex because it is not usually 
observable physically - unless there are signs.  But you can verifiably 
determine the name of a lake for example by spending sufficient time 
around it and asking every local you meet for the name of the lake.  If 
you get consistent results (like 3/4 of the people giving you the same 
name) you have a verifiable name.  No one practically does this test of 
course but as a local mapper you have an intuitive feeling for what the 
test will have as results.

Now the wikidata ID is different - not mainly because you have to look 
it up in a separate database (which you could argue is similar to the 
idea of asking a local for the name of something) but because what you 
look at there is not first hand local knowledge and because you cannot 
independently verify if this information is true or false.

If i assume for a moment that every piece of information in wikidata is 
diligently recorded according to the rules there it would be possible 
to reconstruct this information from the "serious and publicly 
available references" used but in many cases it could *only* be 
reconstructed based on these sources and not independently.  This is 
the main reason why wikidata information is not verifiable according to 
the OSM understanding of verifiability.  

In other words: If a sufficient number of "serious and publicly 
available references" assert something is true this qualifies it for 
being recorded in wikidata.  So a lot of things in wikidata are not 
verifiable in the OSM sense because if they are true or false depends 
on what sources exactly you consider for your assessment.  And because 
of that the mapper in OSM cannot verifiably determine if a wikidata 
object created based on such information qualifies for being specified 
in a wikidata tag.

-- 
Christoph Hormann
http://www.imagico.de/