[Talk-us] Neighborhoods / Zillow

stevea steveaOSM at softworkers.com
Wed Jun 12 17:55:56 UTC 2013


Martijn writes:
>I agree with the advantage of polygons when performing queries of 
>the type 'show me all bakeries in this neighborhood'. This will 
>however only work if that neighborhood is clearly defined in terms 
>of boundaries. If we agree that this is not the case, we are just 
>going to be creating confusion and perhaps even edit wars when we 
>settle on polygons for neighborhoods. A node location for a 
>neighborhood is something locals should be able to relatively easily 
>agree on. I think we can see much faster progress proceeding along 
>that avenue.

I'll say it again:  both polygons and nodes are useful as 
neighborhoods in the map.  We shouldn't outright dissuade either one, 
as each type of data has value and is valid.  However, we should be 
careful at encouraging non-locals from entering neighborhood data (of 
either type) as (IMHO) it truly is best for a local (person) to enter 
these.  At a minimum, a non-local entering neighborhood data should 
vet the data with a local, or do some research to verify its 
accuracy, as difficult as either or both may be.

>I think that we should show great restraint with importing any more 
>boundary polygons. They make mapping more difficult and confusing, 
>for example because they often overlap with roads. They do not 
>represent surveyable / verifiable data in many cases, which makes 
>for dead data, which we have enough of in the US.

Well, SOME polygons correspond, for example, to a sign on a road once 
the boundary is crossed saying "Welcome to (Neighborhood)."  That is 
surveyable, but I agree, it is not widespread.  Also, we have not 
well discussed those places where a street or small area 
realistically shares membership in two adjoining neighborhoods.  Is 
Jane Street (NYC) "in" Chelsea or Greenwich Village?  Well, kind of 
both.  This is where nodes work better.  And again, neighborhood 
nodes belong not in some mathematically-determined "center" but 
rather at a "cultural crossroads" that represents the "heart of the 
center" of that neighborhood.

>Back to my original question, rephrased slightly - would there be a 
>legal impediment to use Zillow or Geonames data to derive 
>neighborhood point data to increase coverage in OSM?

Very incumbent upon any import is an "honest brokerage" to verify the 
data are fresh and accurate.  This is true of not just neighborhood 
imports, but any import.  Checking the legality/license-ability is 
one (important) thing.  Checking its freshness and accuracy is 
another, and just as important.

>Why I care - because neighborhood data represents just what makes 
>OSM unique - local knowledge. Why use external sources then you say? 
>Well, the point would be to make it easy for locals to add 
>neighborhood data to OSM, by offering a data starting point.

A worthy goal, to be sure.  But imagine a new user coming to an early 
map with both noisy TIGER data and noisy neighborhood data:  possibly 
misnamed and mislocated centroid neighborhood nodes, and little else. 
Does that make for a good place for that user to jump-start mapping? 
I think not.  Let's be careful at importing non-local neighborhood 
data.  I'm OK with it being nodes or polygons, I'm OK with importing 
it, but it really should be accurate and verified data.  ESPECIALLY 
with neighborhoods, getting a local person who knows the geography is 
an exceedingly helpful (maybe even required?) component of this sort 
of data entry.

In short:  if an automated (import or import-like) process, like 
Map-A-Thon, were to bring into the map either neighborhood polygons 
or neighborhood nodes, I'm OK with that, so long as somebody local 
gets to verify the data and say "yup, that's where I'd put that, 
because I know that's where that neighborhood is."  Otherwise, it 
could very well be the TIGER import all over again.  This makes such 
a process a bit more difficult, but it has the upside potential of 
better developing local OSM community, by reaching out to those who 
know an area well.  Certainly, we can do exactly this, but let's do 
it right.  (Martijn, thank you for encouraging us to "reach high" 
like this:  it's a worthy goal, it's doable, and it challenges us in 
a rewarding way).

SteveA
California



More information about the Talk-us mailing list