[Talk-us] Low-quality NHD imports

Kevin Kenny kevin.b.kenny+osm at gmail.com
Fri Oct 13 14:51:51 UTC 2017


On Fri, Oct 13, 2017 at 10:06 AM, Frederik Ramm <frederik at remote.org> wrote:
> I only posted that on the talk list and not here, so for those on
> talk-us who don't read talk and who are familiar with the "imports are
> always bad for the community" discussion, you might want to have a look
> at a scientific paper recently written by Abhishek Nagaraj (UC
> Berkeley-Haas) which finds that:
>
> "... a higher level of information seeding significantly lowered
> follow-on knowledge production and contributor activity on OpenStreetMap
> and was also associated with lower levels of long-term quality."
>
> The paper can be freely downloaded here
> https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3044581 and there
> has been a little discussion about it over on the talk list, here:
> https://lists.openstreetmap.org/pipermail/talk/2017-October/079116.html

Interesting. The interesting take-away is that there may be an ideal
level of 'information seeding' so that potential contributors are
initially attracted to the platform (if there's nothing there, there's
no reason to believe that your contribution will ever have value)
and not driven off by a sense that the job is done. The
'inverted U' effect is hypothesized several times in the
paper.

I'm a post-TIGER mapper, so the TIGER import has always
been part of my environment. I know that when I was a beginner,
one thing that quite put me off trying to improve it was
all the noise tags that came with the import. Since at the
time, I had no idea what the purpose of that information
was, much less that it serves no purpose whatsoever, I
was reluctant to edit TIGER ways for fear of breaking
something. In the environment we live in now, we certainly
need some way to encourage new mappers to go ahead and
break things. (I know, this argument is anecdotal, but
I can't even think of how I could set out to test it.)

I also think the balance shifts some in the places where
there simply are never going to be enough mappers.
I could, in theory, lawfully get out in the field and
survey the boundaries of the big parks, for instance.
But the boundaries don't move much, and there aren't
enough people like me with the necessary motivation
and skills. But people relate to parks, and react
badly to an argument that they shouldn't be
on the map unless someone has devoted that
sort of effort. (Someone has. The state surveyor. And
even the state survey may reblaze the bounds only
once or twice a century.) Remember that in my part of
the word, there are even indefinite county boundaries.
County lines that have never been surveyed and
monumented. (It messes up everyone's idea of topology
not to draw lines on the maps, so people do.)

I think the ideal level of imports is greater than zero and less
than TIGER. We'll go on arguing about where the maximum
value is along that curve, and that's a good thing. In the
meantime, we all try to improve the map, and to recruit
mappers - which, we see, are sometimes conflicting goals.



More information about the Talk-us mailing list