[Talk-us] TIGER considered harmful

[Paul Johnson]

Dave Hansen wrote:

> On Sun, 2009-11-15 at 10:59 -0800, Paul Johnson wrote:
>> Dave Hansen wrote:
>> > If we can come up with a scheme for getting the addressing imported in a
>> > sane fashion and the consensus is that people want it done that way,
>> > it'll get imported.  There are still quite a few squeaky wheels that
>> > like to grumble about TIGER, but I haven't heard a single person say
>> > that it did more harm than good.
>> 
>> I firmly believe this.  TIGER contains so many errors in Oregon,
>> Washington and Idaho that it would likely be easier to start fresh than
>> fix.
>
> Just curious, but why weren't you mapping when Oregon and Washington
> were blank?    

No, I hadn't heard of the project yet.

> I'm also curious how you've made so many changes.

A lot of them are simple expansions of street names that TIGER wrongly
abbreviated that I regexp replaced within the OSM and uploaded the
correctly expanded name.  Which streets did TIGER name wrong?  All of
them.

>  Have you
> surveyed it, or are you using TIGER plus Yahoo! imagery?

In many cases, I surveyed it or frequent the location often enough to
have the items I mapped memorized.  I'm not using the Yahoo imagery,
it's rather low quality and very much out of alignment, but I am using
the Oregon Geospatial Enterprise Office and Metro Region's aerial
imagery.

> What would you
> have done without TIGER in these cases?

Used state data instead, if I were to do a mass import.  Oregon GEO
knows what they're doing, the US Census (along with the rest of the
federal government) barely acknowledges we exist.  Which would you
rather trust?

1) Known good data from a regional government agency that's receiving
it's information directly from the Department of Transportation...

or

2) A bunch of minimum wage lackeys who don't want to be out in the rain
and aren't motivated to do it right?

>> 1) TIGER data is so out of date for urban parts of Cascadia as to be
>> rendered entirely useless.
>
> Hmm.  I live in urban Cascadia.  My subdivision was off by ~100m and had
> a major street routed wrong.  I still don't consider it quite entirely
> useless.  Personally, I find it a lot easier to map with GPS traces, my
> memory and hints from TIGER than to go out and take detailed notes.  It
> looks to me like you do the same thing, so I'm really surprised that you
> don't like TIGER.

Visit Portland or Salem and there's vast tracts that are missing or off
by ~250+m

> This one might be my fault.  There are a bunch of TIGER->OSM feature
> code mappings, and it's quite possible that I just plain got them wrong
> in some cases, or that we disagree on how the mappings should have been
> done.

I'm pretty sure dual-carriageway expressway doesn't qualify as a single
way.

> If this is still widespread, I'd be happy to look into it to see what
> can be done to fix it up.

More often than not, TIGER is wrong if the tag is highway=residential. 
Most of Oregon's roads are forest service, BLM or private forest
tracks...

> BTW, did you go and survey these thousands of miles of roads to ensure
> that they really are what you think they are?

On the roads that I'm complaining about and have fixed, yes.

> Are you also saying that you'd rather have a blank space in the map than
> a wrongly-tagged expressway?

Blank space.  I'd rather have it done some way appreciably close to what
the reality the first time over a larger span of time than have a few
million objects to fix later.  The former isn't so overwhelming to
maintain, even if it is much more difficult to appear complete.

>> The TIGER import should never have been done.  I wonder how easy it
>> would be to undo this until an actually suitable data source can be
>> found, since the Fed is doing it on wet bar napkins with "cartographers"
>> who wear hockey helmets and ride the short bus to work.  Might as well
>> photograph a turd and call it aerial photography of central Idaho for
>> the accuracy of TIGER...heck, that photo might actually agree with the
>> TIGER data better!
>
> So, what I did initially was go and contact all of the mappers in the US
> that I could find.  I asked them all individually what should be done in
> their local areas and I believe I was able to follow their wishes
> without failure.  If you want to do the same with TIGER removal, I'd
> welcome it, especially if you have something superior to contribute.  On
> a small or large scale, we should *NEVER* keep any data in the map just
> because it is what was already there.
>
> TIGER doesn't provide the best possible map, but it did (and does)
> provide the best map that we have access to.  If anyone has better
> suggestions, I'm open to them.  

Contact local states first!  Assuming they actually buy into the
overarching political ideology this country was founded on (instead of
just pretending they do), this information is available to you free of
charge and likely available for download.  TIGER should be an absolute
last resort when no proper resource exists.