[Talk-us] Imports - an attempt to explain

[Kevin Kenny]

On 12/20/2012 10:50 PM, Russ Nelson wrote:
> Now, the flip side of your idea of collaboration is that I knew about
> OSM, knew that my county was largely empty, and knew that TIGER would
> be a great starting point. So I DID NOT do any editing until TIGER got
> uploaded. Your theory is great, but I'm a refutation of it. I'm also
> only one person, so you might be mostly right. I just don't think so.

You're not alone. And I've been largely silent about it. Because
I've believed the OSM teaching that studies have conclusively shown
that imports are an impediment to recruitment. But, having seen this
thread, I can't believe it.

I came to OSM to use it as a tool: I occasionally go hiking, and I've
been trying to produce better maps of the places I go because I like
to have good ones. (Moreover, I like to have good open source ones so
that I can hand them out!) I found OSM to be an invaluable tool for
laying down the framework of the map. It's all at the quality level of
"better than nothing" - but without it, for several layers of the map
that I work with, "nothing" is what I have without it.

In the places where I understand the data, I have various other data
sources, and use these also as map layers. For instance, in doing my
mapping, I remove water features altogether from OSM and replace them
with corresponding features from NHD.  I bring in amenities from public
data sets from New York State agencies. I remove the results of the
import of "New York State DEC Lands" that you did some years ago,
and instead render the data from a current import. I add data from
separate files of public and Nature Conservancy lands. (On my personal
map, I also have a fairly complex and brittle script to deal with
integrating OSM, NYS DEC Roads and Trails, and my own GPS tracks. The
result is often a set of braided trails because I didn't get the script
right.) I put in wetlands from US Fish and Wildlife Service National
Wetlands Inventory.

If I didn't have the basic framework of public data sets for the
layers, I couldn't have begun.

So far, what I'm describing is being a taker - a leech - rather than
a contributor. And I don't like doing that. I want to give back. But
most of the actual work I could contribute back is in either small
corrections - places where I've gone, noted error in the public data
sets, and fixed them for next time, or else in the stuff that
"patches around the edges" - either conflates features from multiple
data sources or blesses a feature from one source over another.

In my opinion, the second of these is more broadly valuable. And it
had originally seemed to me that an obvious way to share it would be
to contributed the conflated result back into OSM (presuming that
appropriate licenses could be negotiated), and to offer a side channel
for the rules that helped conflation and the data describing what
was *not* included. This last data set is essential if the conflation
process is ever to be repeated: "feature with ID=xxx in data set yyy
still isn't there in the field."

So far, I haven't dared - instead, I've kept my mapping to myself.
I've confined my contributions to a handful of spots where I could
personally verify every single data point, literally with boots on the
ground. Because that seems to be the only thing that's
non-controversial. So I've outlined a handful of buildings that I
wanted on a map of a local park, sketched in some walking paths, and
made a few tweaks like that.

You, Russ, personally, and Richard Welty, have been extremely
encouraging off-list, but that hasn't been quite enough for me to
overcome the sense of dread that I have at proposing specific imports.

Even with the relatively non-controversial NHD, I get the sense that
a significant minority, if not a majority, of OSMers would consider
an import of a sub-basin unacceptable unless I were to personally
go out and verify every place where a highway crosses a stream and
retag it as a bridge (or whatever kind of crossing it is), and even
then I'd run afoul of controversy over what is a river and what is
a stream. And it's just not worth it to me, because this is a
particular data set that I can deal with readily - throw out all the
hydrography from OSM, and render my map with NHD. End of problem.

And the most daunting thing? I have no clue who actually speaks with
authority and who is merely a windbag. (OK, I recognized Richard's
name as being until recently a board member...) Until hearing the recent
discussion of an import committee, I got the distinct impression
that there was no governing body for imports, that ALL existing imports
had been done in "cowboy" fashion, and that the vast majority of OSMers
thought we'd be better off had none of them been done.

But, had none of them been done, we in the US would have no map!
At least if other mappers are like me, it's going to be really hard
to energize us to get out and duplicate data that already available,
paid for by our taxes. Even if there are errors and inconsistencies,
I'd rather spend my time on stitching up the errors than on trying to
create the tapestry out of whole cloth.

-- 
73 de ke9tv/2, Kevin