I’m excited to announce a new MapRoulette challenge: Norway - Fix Spiky Buildings (MapRoulette)! We’ve identified buildings on OpenStreetMap (OSM) with oddly sharp angles. Most building corners are close to 90 degrees, but this challenge focuses on buildings where angles are less than 15 degrees. While some might be real, many are likely incorrect data.
The Task:
Improve OSM data quality by reviewing the identified buildings and see if their shapes need correction.
Important Considerations:
Some buildings, especially historical ones, may have unique angles.
If unsure, consult the community forum for guidance.
Why Fix Spiky Buildings?
Accurate building shapes on OSM benefit everyone! They improve:
Data quality for various applications
Navigation services
Urban planning
Emergency response
Have Questions?
Feel free to reply to this message for help, modifying instructions, or suggestions on resources.
Thanks for your contributions!
Hi there OSM community in Norway and in particular @anderfo and @hansfn (as you already showed interest in this MapRoulette challenge.
We in TomTom just wanted to follow up with you on this challenge as we noticed that only about 13% of the cases is solved and we have some concern about the challenge. Is there anything else needed from our side to make that would help to make this challenge more approachable/attractive? Do you have any comments or concerns? We are as always listening
The biggest problem was that many (most) of the issue I tried to fix, required zooming in (really close) on each node in the building to find the extra node that was causing the spike. It was just slow / cumbersome.
Hi all! How is this Challenge going, still having difficulties to find the problematic nodes? Improvements needed or is this an OSM internal process issue?
I have made some progress so the main problem for me is time
Anyway, the challenge has major UX problems:
The edit link is to “…node=XXX” instead of “…way=XXX” making it hard to find the building with the spike (if the building is surrounded by other buildings).
As already mentioned, it is often hard to find the problematic node(s). Some times iD warning about nodes being very close, but mostly you need to inspect / zoom in.
So, a task that should be very quick - typically deleting a extra node - just isn’t …
All the cases I checked were false positives, i.e. no action needed.
All the 4 million buildings which were imported in Norway some time ago have already been automatically rectified. Most of the remaining buildings are not supposed to be square.
I think it would be more productive if this challenge were able to identify big mistakes such as self-crossing building polygons, dragged building nodes or very large spikes/very small angles.