[OSM-talk] to all potlatch and JOSM users - automatic simplification of geometry

[Dave Stubbs]

On Sun, Aug 9, 2009 at 11:12 AM, Stefan de Konink<stefan at konink.de> wrote:
> On Sun, 9 Aug 2009, Richard Fairhurst wrote:
>
>> Stefan de Konink wrote:
>>
>> > And then you realise that the alignment of Yahoo Imagery is wrong
>> > on most
>> > places, and you have killed good vector material. Great job :)
>>
>> "Good vector material"? Tell me, have you ever _seen_ TIGER?
>
> Yes, and I have also _seen_ Yahoo been wrong on more places than it was
> right. So I am still skeptical people tracing Yahoo and claiming they are
> better than [insert source here] mainly because except that source they
> have 0 ground reference.
>
>> Seriously, you're kidding me. Yahoo may be out-of-alignment here and
>> there but by and large if the street looks straight in Yahoo, it is
>> straight. And US city streets generally are. TIGER streets are
>> bonkers mad wavy.
>
> Most likely this is true because the TIGER dataset was actually traced
> from/for a much lower precision map, hence wavy because you are plotting
> it on a far higher resolution then it was created for. With respect to
> TIGER you might say: TIGER is *so* bad, Yahoo will always be better. My
> point is, unless Yahoo is actualy validated where you are tracing it
> gives poor output too, no matter how bad you think TIGER is.
>

Um, no. It'll be potentially misaligned, potentially outdated, and
potentially misinterpreted.

So potentially poor, but it's entirely possible (and actually quite
likely) that it's fine and you get pretty good data out.
And it's definitely better than most tiger data I've ever seen.

So choosing between potentially poor, and very horrible isn't really
so hard is it.
Hopefully someone on the ground will eventually verify it -- they can
fix it if it's still broken.

Dave