[OSM-dev] TIGER field report, Columbia, Maryland

Tom Permutt natashafatale.54532583 at bloglines.com
Sat Sep 1 04:44:37 BST 2007


Dave has kindly uploaded Howard County, Maryland. I've had Tiles at Home render
one tile that I had already partly mapped. You can search for zipcode 21044
if you'd like to see it. Here's what I see:

1.  The script is correct, or very nearly so. The only thing that's likely
to be a rendering error, I think, is cul-de-sacs. We have a lot of dead-end
streets with little circles at the end. TIGER actually paints nodes densely
around the circles. These come out nearly right in OSM, except for spurious
stubs out the far end.

2.  TIGER is mostly right, nearly enough right that OSM is much better with
it than without, in its present state. Mappers can now spend our time
correcting and enhancing what's there rather than tracing raw tracks. This
may turn out to be more fun or less, but it will certainly produce useful
maps faster.

You can see some streets right on top of mine; some that I didn't have (some
of which actually exist, and some not); and a few about which we disagree.

We'll want some discussion of guidelines for mapping with TIGER data. If
it's wrong, do we delete it? Is it significantly better to try to move it to
where the street really is? What about if it's right? We won't want two
versions of the same way, or will we? Which one do we prefer? TIGER may have
a lot of tags, but only some of them may be right. Is this better or worse
than a verified name and class and nothing more? What about road classes?
TIGER and I disagree about MD 108. Will the TIGER classification become the
default standard in the US, or should we try to apply the existing
guidelines (which need better adaptation to US English, but that's another
issue)?

Kudos to Dave and everyone else who has worked to make this import possible.

Tom Permutt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20070831/75213f9d/attachment.html>


More information about the dev mailing list