[JT] Jochen Topf's Blog
Mon 2017-08-28 20:35

Polygon Fixing Effort Concluded

In February of this year I started a community effort to fix broken multipolygons in OSM and switch the remaining old-style multipolygons (with the tags on the outer ways instead of on the relations) to a more modern tagging. Seven months later I am declaring this effort to be finished.

This was a huge community effort with over 400 OSM users contributing “to the cause”. In 15,000 changesets the community switched 250,000 multipolygons from the old to the modern tagging. No old-style multipolygons are now left (except a few that are created anew here and there, most likely by mappers who don’t know yet that the rules changed.) As a result, the main OSM map was switched over and doesn’t handle old-style multipolygons any more. Most of the major editors and other programs handling OSM data switched or are in the process of switching to eliminate old-style multipolygon handling (see the Github issue tracking this work if your are interested).

In addition the community fixed more than 100,000 bad (multi)polygons. This includes repairing various problems from self-intersections in polygon boundaries, over open rings, to lots of (inner or outer) ways that had the same tags as the multipolygon relation. To enable this work I created 76 Maproulette challenges on which the community worked in over 40,000 OSM changesets. Over the last months about half of all the changesets attributed to Maproulette were part of this effort.

Another invaluable tool for this effort was the OSM Inspector. Many thanks to Frederik Ramm and Michael Reichert of the Geofabrik who set up the OSM Inspector to show the data about (multi)polygon problems and changed it multiple times when I had new data for them. I am still creating daily reports of this data on my personal server and feed it to them.

I have also to thank Mapbox who supported the effort by allowing me to work on this as part of my consulting job for them. Without it I couldn’t have pushed this as far.

And, of course, I want to thank the awesome OSM community who did the actual work. Some OSM users went above and beyond and fixed thousands or even tens of thousands of problems single-handedly. And after all this they voted for me in this years OSM awards. Thanks for the “Greatness in Mapping Award”! It belongs really to the whole community and their great effort.

I hesitated a long time to actually start this project, because I wasn’t sure whether it was worth the effort and whether it could actually be done. Could we actually fix hundreds of thousands of problems manually? The result shows that the answer is a clear “Yes, we can!”. There are a lot of mappers who are willing to go the extra mile, to not only work on improving their neighborhood, but do the sometimes tedious work needed to improve the data for the whole world.

This gives me a lot of hope for the future of the OSM project. Sure, there is a lot of internal squabbling, and often it isn’t clear where the project is going. But any kind of future for OSM depends on us willing to learn from what we have been doing and fixing our work from yesterday. Our understanding of the world that we are mapping is evolving as are our data model and tagging standards. We need to keep cleaning up and reorganizing our data to keep it useful.

We are never finished, there are always more things to fix in OSM. The number of broken (multi)polygons is, again, slowly climbing and there are many more problems with the data unrelated to (multi)polygons. I have learned a lot about OSM data and the tools we need to fix it as well as about the great OSM community that rallied in this effort. So while I am declaring this effort to be done, I am already thinking about future work ahead.

Tags: openstreetmap · polygon