[Talk-us] Boundaries and verifiability (was Re: Retagging hamlets in the US)

Frederik Ramm frederik at remote.org
Mon Mar 23 07:27:11 UTC 2015


Greg,

> 3. It is my belief and experience that the "ground observable rule" is
> something that only applies to Europe or older metropolitan areas.

I think there's a misunderstanding here.

Of course even in European metropolitan areas there will *not* be a sign
bearing the name of every stream that you drive across! That doesn't
keep Europeans from mapping the stream (the fact that there *is* one is
at least observable), or naming it according to common knowledge or
whatever the locals will tell you the name is.

We usually draw the line when it is about features that cannot be seen
on the ground; these should be in OSM only in exceptional cases (for
example we do map administrative boundaries and post code areas even if
they're invisible; the discussion about how much of a railway must still
be there to map it as "abandoned" is going on elsewhere; the mapping of
airways is strongly discouraged; some people map long-distance radio
links but that is not likely to catch on).

Your remark that OSM is different from the "old GIS world" with ESRI and
$20k GPS receivers is correct, however it is not a suitable basis for
reasoning (following the same logical path as you did, I could say "they
use computers; we are different, so we should not use computers").

The "ground observable rule" kicks in most strongly when there's a
dispute. If one mapper happily maps an invisible boundary and another
mapper pops up and maps it differently, and they later apply to someone
to mediate in their conflict, that third person will ask whether there
is any proof for each mapper's version, and if there isn't any because
both just map from hearsay, then the feature will have to be tagged as
"disputed" or removed altogether.

> 9. Taking Serge's example of neighborhood boundaries to the logical
> conclusion, nothing should be put in OSM because an edit war __could__
> ensue.

Again, you've misunderstood Serge; because as long as we stick to
observable things, the edit war can be resolved by fact-checking.

This is what Serge hinted at when he talked about Alice and Bob.
Crucially he also mentioned that there's a high risk that if we allow
un-substantiated mapping of neighbourhoods, this might be at the expense
of the underprivileged who seldom participate in OSM. For some, it might
make a very big difference whether their address resolves to
neighbourhood A or neighbourhood B if they live just on the border. As
long as we're talking facts there's not much that can go wrong - an
able-bodied, college-educated caucasian male can trace a stream through
the slums from Bing without being in much danger of unwittingly applying
prejudice. The same is not true for the same able-bodied,
college-educated caucasian male drawing the boundary of the
neighbourhood they are unlikely to ever set foot in.

There's actually quite a few things apart from neighbourhoods that are
not defined. For example here in Germany, if a village can advertise
themselves as being "in the Black Forest", that's a plus, tourism-wise.
But the Black Forest is not a forest where you simply check the
treeline; it's a large region with not-really-well-defined boundaries.
There's places where 99% of interviewees would says "clearly that's in
the Black Forest", and places where 99% would say clearly not, but a
grey band in between. The kind of area that is labelled with a curved,
wide-spaced font on old-school maps. OSM doesn't have a good mechanism
to record these; OSM only accepts precise geometries, not fuzzy ones.

> 7. The "ground observable rule" is a barrier to new mappers. I helped a
> new mapper at a Editathon add taco stands.  She did everything wrong. I
> did say no you cannot add that node. We have not gone and surveyed that
> node exists.  I let her add the node with abbreviated street names and
> all.  She was so exited to add here research data to OSM.

There's absolutely no problem with adding Taco stands from memory as
they are observABLE (even if not observED) and if someone else starts a
fuss about the Taco stands, we can just go there and check.

People add data from memory all the time, and if it's wrong, it get
fixed. But that's not the point when discussing neighbourhood boundaries.

> I failed to
> map for months because it sounded like I had to have a GPS five years
> ago before I could map.

I think you're consistently misunderstanding the difference between
observable and observed.

Bye
Frederik

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



More information about the Talk-us mailing list