I propose to systematically remove all subarea members (e.g., states from the US boundary relation, counties from state relations, municipalities from county relations) from boundary relations in the United States. No OSM features will be deleted as part of this edit.
Having these nested “subareas” provides no functional benefit and creates unnecessary complexity. Removing subareas would simplify our database, reduce maintenance efforts, and ensure consistent data modeling for boundary relations. Any hierarchical logic can be handled by data consumers spatially or via wikidata as is done in other data consumers.
Specifically, the problems with the current usage of subarea are:
- Redundancy: Boundary nesting is already represented spatially;
subareamembers offer no additional data utility. Also, there is no way to tell if thesubarealist is complete. - Maintenance Overhead: Keeping subarea lists current is time-consuming and prone to error, increasing the burden on mappers and setting an expectation for mappers to add subareas.
- Disuse: The
subarearelation role is only partially present in the United States, so they can’t be used by data consumers anyways. A data consumer expecting to use them may not realize until it’s too late that they are not universally proliferated. - Data Complexity: Sometimes cities overlap multiple counties and so forth. Subareas can’t represent the real complexity of the world.
- Recursion: Sub-areas may cause data consumers to unnecessarily download large amounts of data by recursing hierarchically through subarea relation links.
Proposed Approach for Removal:
- Use Overpass and QLever queries to identify boundary relations that currently include subareas.
- Systematically remove subarea members from these relations, ensuring each administrative boundary stands alone as a top-level entity.
I welcome feedback and discussion on this proposal.
The discussion that prompted this proposal to remove subarea relation roles is discussed here.