[OSM-talk] Just facts?

Frederik Ramm frederik at remote.org
Thu Jun 19 10:06:21 UTC 2014


Hi,

   this is an offspring from the discussion about whether or not we are
well advised to follow Wikimedia's example of requiring the disclosure
of paid contributions.

The discussion has been led here on talk and on osmf-talk. A statement
by Emilie Laffray on osmf-talk best summarizes the idea:

"Ultimately, map data is pretty much fact and whether it exists or not
is a binary statement. Now, could someone slip "advertisement" like
places of all shops of a specific brand? Yes! Do we care? No, as long as
the data is factual. I don't care if someone is being paid to put data
in OpenStreetMap as long as this data is correct and valid..."

Let us go back in time to when Wikipedia was started. I wasn't there
then but I am pretty sure that there would have been many at the time
who naively said something like: "An encyclopedia is a collection of
facts and knowledge. We can argue how the facts are organised and
presented but the content is verifiable and clear."

I would probably have agreed. Where and when a composer was born and
what their most famous work is - clearly a fact, no?

Fast-forward to the present. Wikipedia has learned the hard way that
there is much less fact and much more open to interpretation than
initially believed. Meanwhile in OSM, people say: "We don't have the
same kind of problems as Wikipedia because ours is effectively a
database of verifiable facts."...

This is undoubtedly true for a number of things in our database - but
the number of exceptions is much higher than you would naively expect.

Firstly, there are cultural issues mostly to do with names. We have edit
wars about name tags (most recently, disputed islands in South East Asia
and territories between Ukraine and Russia) - what is the "name on the
ground" in an area with active movement of troops?

Another very recent example is a long discussion-cum-edit-war in Germany
about whether or not something really *has* a German name X
("name:de=X") or if that is a thing of past occupation
("old_name:de=X"). This is not a fact that you can simply check, it is
something that requires research and brings us quite far into the
"[citation needed]" terrain already.

But there are other and much less obscure issues, starting with the
highway classification. There are no hard and fast rules about what is,
for example, a primary or a secondary road; this is mainly a distinction
that we leave to the local mapping community, and yes, there are edit
wars about that too, and there is potential for edits with an ulterior
motive (someone who lives on a street might downgrade it from primary to
secondary to have less vehicles routed there, or to make his property
look more attractive to a potential buyer).

We have many other situations in which we trust the mapper to do the
right thing without being 100% verifiable. Track types are an example,
or indeed the infamous "smoothness" tag. When we map which areas are
"residential" and which "commercial", there's quite a bit of leeway
there as well - you can gloss over a supermarket in a residential area,
or you can cut a hole in the area and mark it differently; you can map
the whole supermarket parking lot as a parking lot or you can map the
trees and the little bushes between the rows of cars and make it look
almost like a park.

Is something just a stream or already a river? Is something just a town
or already a city? Does the office of a private music teacher count as a
music school, thereby increasing the quarter's school density?

All these are judgment calls where we trust our mappers to make the
right decision. All these are a far cry from the "binary statement" and
the easy fact checking that people often ascribe to OSM.

I believe that it is still true that most of what we collect *is* facts,
and as long as we stick to the facts we're in the green. It may be
disputable whether something is tracktype=grade1 or tracktype=grade2 but
the fact that there is some kind of track in that location is not up for
discussion.

But this should not make ourselves blind to the fact that there's also
quite a lot of stuff in our database that is not as easy to fact-check.
I believe there is ample room to "interpret" reality in a way that is
not outright wrong, but has a "spin" on it - in OSM as in Wikipedia.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"



More information about the talk mailing list