Well, I have to admit, I've only seen one problem with the armchair mapping with the smoothing in my local area of Pittsburgh, PA so far.
What happened was a long time ago, I cleanup up US-19 and separated segments that were divided and segments that had a 'Center Turn Lane' and tagged as such. Well, it seems that MapRoulette happened to flag one of the segments where it transitioned from divided highway to un-divided w/ a Center Turn Lane and then back to divided for another major intersection. The imagery in Bing clearly showed that there was a Center Turn Lane there and I had he un-divided segment tagged as such. Instead of it being marked as a 'false positive' on the MapRoulette site, the user twinned this segment, even when the Bing imagery didn't justified it (Changeset 22050738 [1]). (And yes, the imagery right now in Bing is still the same as when the change was made.) When I discovered this, I promptly reverted that changeset in Changeset 22262496 [2]. So, as you can see, there are still a few flaws in the MapRoulette smoothing challenge. And unfortunately, not everybody seems to acknowledge that there might be a false positive here and there and mark it as such. Maybe a stronger warning could be given, but I don't know. All I do know is that any time there is a 'Center Turn Lane' on a highway here in the USA, it stays as a single way and only becomes divided is when there is a physical item dividing the highway like a concrete divider. Maybe this particular flag could have been avoided where it wouldn't flag one-way roads merging into a single way if the angle wasn't like a ~90 degree turn or larger as everything is the same right now as when it was unfortunately falsely 'smoothed'. Any thought how this might be avoided in the future Martijn? -James [1] - https://www.openstreetmap.org/changeset/22050738 [2] - https://www.openstreetmap.org/changeset/22262496
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
